jjc@jclark.UUCP (James Clark) (04/08/91)
I've got a Sun Sparcstation 2 on order, and I'm thinking about cancelling it and getting an HP 9000/730 instead. I'm more than a little bit nervous about switching over. Although I'm generally fed up with Sun, I've used their machines for several years, and I know that they provide a reasonably pleasant working environment. I have generally warm feelings towards HP, and I've been very impressed by the responsiveness of individual HP employees in this group, but I really don't know much about HP workstations. So I hope the net will excuse some rather general novice questions. (I know I could probably get answers to some of these from my salesman, but he didn't seem very clued up.) If I get some replies, I'll summarize to this group, since I suspect there are other who have similar concerns to me. If you reply but don't want your remarks to appear in the summary, or want to remain anonymous, please say so and I'll be glad to comply. I've got a lot of questions, but I would like to hear from you even if you want to answer only one of them. First a little bit of background. I work mostly on free software (my project for the last few years has been GNU groff); I write mainly in C++. I also run a uucp node. The things that are important to me are integer preformance, the quality of the available software development tools, the quality of the window system and GUI, and the overall robustness, standards conformance and functionality of the OS. Integer performance is important, but it is by no means everything; floating point performance is unimportant to me. My main concern is about HPUX. My second-hand impression of it is that it's based on some old version of System V, and that it's a bit old-fashioned and non-standard, and generally not quite as good as SunOS. Is this a fair assessment? Does HPUX include DWB? Does HPUX support long filenames, symlinks, job control, sockets, NFS? Is it POSIX compliant? With the 2-user version of HPUX, does a uucp login count as a user? Does a rlogin count as a user? Does uucp work reliably at 19200 bps? Is it a solid version of uucp? Does HPUX provide lpr or does it only provide lp? Does it support PostScript printers? Somebody said that OSF/1 was going to be available by the end of this year. Does this mean that a production version suitable for end users will be available by then? Does this mean HP will be shipping OSF/1 as their standard OS by then? Will OSF/1 run HPUX binaries? Is BSD 4.4 likely to be available? If so, will the machine dependent parts be freely redistributable? If not, does HP document their hardware enough that it would be possible for somebody to port it? How good (fast, robust) is HP's implementation of X? Does it include all of X11R4? Is it reasonably close to X11R4? Does the MIT X11R4 distribution build on a 9000/730? Is the R5 distribution likely to? Do HP document their graphics hardware enough to port a window system? How good is HP hardware and software support? Is it reasonably priced? Can one access support by email? How does HP's 19'' monitor compare to the Trinitron that Sun ships? How about the keyboard and mouse? How noisy/hot would a system with two internal 400Mb disks be? Would it be reasonably pleasant in an office environment? Does it connect directly to thinwire ethernet? Is HP hardware generally reliable? I'm concerned about exchanging data with people who don't have HPs. Can I get a third party SCSI tape that handles Sun-compatible QIC-24 or QIC-150 tapes? My salesman said that 400Mb was the biggest external disk that HP offered on the 730. This seems a bit on the small side. Is HP likely to offer bigger disks, or does it want me to buy a 750 if I need lots of disk space? Can I use third-party SCSI disks? In addition to DAT, I was told there was a writable CD-ROM available (C1701A or something like that). Is this intended as an alternative to DAT as a mass-backup device? I suspect not, since it seems to be slower, have lower capacity, and be more expensive than DAT. Can it mounted like a normal disk, or does one treat it like a tape? Would CD-ROMs written with this drive be readable with Sun's CD-ROM drive? Are gcc and g++ available? If not, do HP provide enough information about their architecture that somebody could port them? If so, is the code quality competitive with HP? How about gdb? What debugging format does HP use? Is it COFF, stabs, or something peculiar to HP? If the last, is it documented? Does HP offer dbx? How reliable is the standard C compiler? Are the higher optimization levels sufficiently reliable to be usable? Is ANSI C included with HPUX or available separately? How good is HP's C++ (Softbench) offering? Is it based on cfront 2.0 or 2.1? How good is the debugger? Is it as good as Saber C++? Is Saber C++ likely to become available? In general, how easy is it to port free software to HPUX? Most things seem to port to SunOS with little effort. To what extent is this true of HPUX? I would also be interested in any other random comments that you might have about the relative merits of Sun and HP workstations. If you've got this far, thanks for bearing with me. This is an important decision to me: any help you can provide really will be appreciated. James Clark jjc@jclark.uucp jjc%jclark@mcsun.eu.net jjc@ai.mit.edu
tml@tik.vtt.fi (Tor Lillqvist) (04/09/91)
In article <JJC.91Apr8133840@jclark.UUCP> jjc@jclark.UUCP (James Clark) writes:
First a little bit of background. I work mostly on free software (my
project for the last few years has been GNU groff); I write mainly in
C++.
Many thanks for groff! It compiles and works fine with HP C++ (both s300/400
and s800).
My main concern is about HPUX. My second-hand impression of it is
that it's based on some old version of System V, and that it's a bit
old-fashioned and non-standard, and generally not quite as good as
SunOS. Is this a fair assessment?
Well, not really. The HP-UX kernel is as far as I know originally based on
BSD4.2. The system call semantics have been changed to be SVID (and now,
POSIX) conformant, though. User level programs are System V versions. As for
standards, well... If you define "standard" as "SunOS" or "(pure) BSD," I
guess HP-UX isn't standard. It does confirm to the *real* standards POSIX,
X/Open etc, though.
But you are right that SunOS does have some nice features that HP-UX hasn't
got (yet). Like dynamic linking, enhanced ptrace(2), interesting new
filesystem types, automounter... It's not fair to call these featrures
"standards," though. On the other hand, the HP-UX approach to diskless
workstations might be better than SunOS's.
Does HPUX support long filenames, symlinks, job control, sockets, NFS?
Is it POSIX compliant?
Yes, yes. Sure.
Does HPUX provide lpr or does it only provide lp? Does it support
PostScript printers?
It has lp, but with support for remote printing (both as client and server).
You must put together your own PS interface model. (Trivial, just a small
shell script, that could check if the stuff is already PS, and if not, pipe
through the ascii-to-PS filter of your choice.)
How good (fast, robust) is HP's implementation of X? Does it include
all of X11R4? Is it reasonably close to X11R4? Does the MIT X11R4
distribution build on a 9000/730? Is the R5 distribution likely to?
Do HP document their graphics hardware enough to port a window system?
I guess HP-UX 8.0 will have X11R4. HP-UX 7.0 has X11R3, but you can get a
X11R4 server and pre-compiled clients and libraries by ftp. Or compile X11R4
yourself.
How does HP's 19'' monitor compare to the Trinitron that Sun ships?
HP ships Trinitron monitors, too.
Are gcc and g++ available? If not, do HP provide enough information
about their architecture that somebody could port them? If so, is the
code quality competitive with HP?
There is a version of gcc for the Series 800 available from Utah. The code
quality is not yet as good as that the HP compiler produces.
What debugging format does HP use? Is it COFF, stabs, or something
peculiar to HP? If the last, is it documented? Does HP offer dbx?
It's something peculiar to HP, and not documented :-(. HP doesn't offer dbx,
but xdb, which is different, but at least as good. Then there is SoftBench, a
whizzy Motif-based interface you probably will get much glossy material on...
However, if you are short of resources (memory), remember that you can do
almost anything SoftBench does "by hand."
In general, how easy is it to port free software to HPUX? Most things
seem to port to SunOS with little effort. To what extent is this true
of HPUX?
Depends on how the writer of the free software interprets the word "standard."
Most free software port fairly straightforward. X software usually compiles
without changes. Some software depend heavily on the more obscure
peculiarities of BSD or SunOS, and are thus more of a challenge. It's
interesting to note that Andrew, however, seems to be easier to compile and
run on HP-UX (on the Series 300, alas) than on SunOS.
--
Tor Lillqvist,
working, but not speaking, for the Technical Research Centre of Finland
beshers@cs.columbia.edu (Clifford Beshers) (04/09/91)
I'm sure everyone is going to hit on this one, but... Yes, it does support long filenames. You have to configure the filesystems and the kernel correctly. -- ----------------------------------------------- Clifford Beshers 450 Computer Science Department Columbia University New York, NY 10027 beshers@cs.columbia.edu
luism@genesis.Berkeley.EDU (Luis Miguel) (04/09/91)
In article <TML.91Apr9103738@nyyti.tik.vtt.fi>, tml@tik.vtt.fi (Tor Lillqvist) writes: |> |> Does HPUX support long filenames, symlinks, job control, sockets, NFS? |> Is it POSIX compliant? |> |> Yes, yes. Sure. |> Baloney!!!. I run HPUX 7.0, and it DOES NOT support long filenames (A pain when untarring things with long filenames. In more than one occasion the clipping of the name meant that some files were written on top of others). -- Luis Miguel. Computer Science Division, UC Berkeley. arpanet: luis@postgres.Berkeley.EDU uucp: {ihnp4,decvax}!ucbvax!postgres!luis at&t: (415) 642-8234
ba7@venus.ornl.gov (M. Lee Bailey III) (04/10/91)
In article <41652@ucbvax.BERKELEY.EDU> luis@postgres.berkeley.edu writes: >In article <TML.91Apr9103738@nyyti.tik.vtt.fi>, tml@tik.vtt.fi (Tor Lillqvist) writes: >|> >|> Does HPUX support long filenames, symlinks, job control, sockets, NFS? >|> Is it POSIX compliant? >|> >|> Yes, yes. Sure. >|> >Baloney!!!. I run HPUX 7.0, and it DOES NOT support long filenames >(A pain when untarring things with long filenames. In more than one >occasion the clipping of the name meant that some files were written >on top of others). > > >-- >Luis Miguel. Computer Science Division, UC Berkeley. >arpanet: luis@postgres.Berkeley.EDU >uucp: {ihnp4,decvax}!ucbvax!postgres!luis >at&t: (415) 642-8234 Baloney!!!. I run HPUX 7.0, and it DOES support long filenames. We, however, read the man page on convertfs(1M) in order to enable this feature. _M_._ _L_e_e_ _B_a_i_l_e_y_ _I_I_I_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _M_a_r_t_i_n_ _M_a_r_i_e_t_t_a_ _E_n_e_r_g_y_ _S_y_s_t_e_m_s_,_ _I_n_c_. Internet: _b_a_7_@_o_r_n_l_._g_o_v BITNET: _B_I_T_N_E_T_%_"_b_a_7_@_o_r_n_l_s_t_c_" Slo-Mail: _P_._ _O_._ _B_o_x_ _2_0_0_9_,_ _O_a_k_ _R_i_d_g_e_,_ _T_e_n_n_e_s_s_e_e_ _ _3_7_8_3_1_-_8_2_0_3
luism@genesis.Berkeley.EDU (Luis Miguel) (04/10/91)
In article <41652@ucbvax.BERKELEY.EDU>, luism@genesis.Berkeley.EDU (Luis Miguel) writes: |> In article <TML.91Apr9103738@nyyti.tik.vtt.fi>, tml@tik.vtt.fi (Tor Lillqvist) writes: |> |> |> |> Does HPUX support long filenames, symlinks, job control, sockets, NFS? |> |> Is it POSIX compliant? |> |> |> |> Yes, yes. Sure. |> |> |> Baloney!!!. I run HPUX 7.0, and it DOES NOT support long filenames |> (A pain when untarring things with long filenames. In more than one |> occasion the clipping of the name meant that some files were written |> on top of others). |> SORRY, this is wrong. HPUX does support long filenames as an option. Next time I will not be so quick to speak........ -- Luis Miguel. Computer Science Division, UC Berkeley. arpanet: luis@postgres.Berkeley.EDU uucp: {ihnp4,decvax}!ucbvax!postgres!luis at&t: (415) 642-8234
hardy@golem.ps.uci.edu (Meinhard E. Mayer (Hardy)) (04/10/91)
And how it supports longfilenames! (And tcsh even will spellcheck them for you!). Greetings, Hardy -------****------- Meinhard E. Mayer (Hardy); Department of Physics, University of California Irvine CA 92717; (714) 856 5543; hardy@golem.ps.uci.edu or MMAYER@UCI.BITNET
puglia@cunixa.cc.columbia.edu (Paul Puglia) (04/10/91)
In article <41652@ucbvax.BERKELEY.EDU> luis@postgres.berkeley.edu writes: >In article <TML.91Apr9103738@nyyti.tik.vtt.fi>, tml@tik.vtt.fi (Tor Lillqvist) writes: >|> >|> Does HPUX support long filenames, symlinks, job control, sockets, NFS? >|> Is it POSIX compliant? >|> >|> Yes, yes. Sure. >|> >Baloney!!!. I run HPUX 7.0, and it DOES NOT support long filenames >(A pain when untarring things with long filenames. In more than one >occasion the clipping of the name meant that some files were written >on top of others). > > >-- >Luis Miguel. Computer Science Division, UC Berkeley. >arpanet: luis@postgres.Berkeley.EDU >uucp: {ihnp4,decvax}!ucbvax!postgres!luis >at&t: (415) 642-8234 Untrue. I believe that even 6.5 supported long file names. You have to convert your filesystem to use them though. Pages 6-63 thru 6-72 in the HP-UX Systems Administration Tasks manual describes the procedure for, and the consequences of making the change. Paul Puglia Dept. of Civil Engineering Columbia University puglia@cunixf.cc.columbia.edu
riley@a243.even.ge.com (U-E59331-Dana Riley) (04/10/91)
luism@genesis.Berkeley.EDU (Luis Miguel) writes: >In article <TML.91Apr9103738@nyyti.tik.vtt.fi>, tml@tik.vtt.fi (Tor Lillqvist) writes: >|> >|> Does HPUX support long filenames, symlinks, job control, sockets, NFS? >|> Is it POSIX compliant? >|> >|> Yes, yes. Sure. >|> >Baloney!!!. I run HPUX 7.0, and it DOES NOT support long filenames >(A pain when untarring things with long filenames. In more than one >occasion the clipping of the name meant that some files were written >on top of others). I also run HPUX 7.0 and it does support long filenames. Specifically, short filenames is the default, but the disk can be configured to support long filenames. At version 8.0 of HP-UX, it comes configured for long filenames. -- ____________________________________________________________ | | | Dana M. Riley, Senior Analyst | | General Electric Aircraft Engines |
bigmac@erg.sri.com (Bryan McDonald) (04/10/91)
In article <41652@ucbvax.BERKELEY.EDU> luis@postgres.berkeley.edu writes: >In article <TML.91Apr9103738@nyyti.tik.vtt.fi>, tml@tik.vtt.fi (Tor Lillqvist) writes: >|> >|> Does HPUX support long filenames, symlinks, job control, sockets, NFS? >|> Is it POSIX compliant? >|> >|> Yes, yes. Sure. >|> >Baloney!!!. I run HPUX 7.0, and it DOES NOT support long filenames >(A pain when untarring things with long filenames. In more than one >occasion the clipping of the name meant that some files were written >on top of others). Actually, HP-UX 7.0 goes both ways. ;-) The default is to stay with short file names, but you can convert the filesystem to long file names at will. Of course, you then have to worry about what other HP sites that you interact with are doing, but that is another administrative headache... -------------------------------------------------------------------- Bryan McDonald | Computer, Hardware, And Operations Support bigmac@erg.sri.com | CHAOS | ITAD - SRI International "I refuse to have a battle of wits with an unarmed person."
mjs@hpfcso.FC.HP.COM (Marc Sabatella) (04/10/91)
I work for HP, but am not an official spokesperson. Take my comments accordingly: >My main concern is about HPUX. My second-hand impression of it is >that it's based on some old version of System V, and that it's a bit >old-fashioned and non-standard, and generally not quite as good as >SunOS. Is this a fair assessment? Obviously this is subjective. The sources to HP-UX come from SVR2 as well as BSD 4.2, and have had a *lot* of work done by HP. The claims of "nonstandard" mainly come from people who equate "standard" with BSD and/or SunOS. The 700's will come with some flavor of HP-UX 8.0, which will include shared libraries, X11R4, Motif 1.1, and most other "modern" features. Missing are some BSD functions from the C library, and some things that require the separate AT&T toolchest license (I think nmake falls into this class). >Does HPUX support long filenames, symlinks, job control, sockets, NFS? >Is it POSIX compliant? Yes on all counts, depending on how you define "POSIX compliant" - there are many POSIX specs, and various different suites to measure compliance. I don't know our full story, but I know that 8.0 contains a lot of code in the commands, libraries, and kernel for drafts 1 and 2. >Does HPUX provide lpr or does it only provide lp? lp - but the spooler accepts requests from Berkeley "lpr". Is this actually an issue? They are plug compatible in many cases. >Somebody said that OSF/1 was going to be available by the end of this >year. Does this mean that a production version suitable for end users >will be available by then? Does this mean HP will be shipping OSF/1 >as their standard OS by then? Will OSF/1 run HPUX binaries? The OSF picture is not clear. OSF itself has said they don't intend to OSF/1 to be for end users - for developers only. At least that's what I heard. Don't believe everything you hear, particularly about OSF. Don't expect HP-UX to go away any time soon, though. >Is BSD 4.4 likely to be available? If so, will the machine dependent >parts be freely redistributable? Much of the BSD 4.4 development is being done on HP machines, so the answer to the first question is undoubtedly yes. You'll have to contact BSD as to the "freely distributatble" stuff - I don't think HP has any involvement with the port. >How good (fast, robust) is HP's implementation of X? Does it include >all of X11R4? Is it reasonably close to X11R4? Does the MIT X11R4 >distribution build on a 9000/730? Starting with 8.0, it is standard X11R4, except I don't think we include some of the contributed stuff that we don't get a chance to test. People have indeed successfully compiled the MIT stuff directly. As for speed, the posted benchmarks should speak for themselves - the 730's X performance blows everything else away. >Does it connect directly to thinwire ethernet? Is >HP hardware generally reliable? Yes and yes - the reason many people couldn't get 400's for a while is that the US Army was snatching them up to use in tanks in the gulf. [ re: writable CD-ROM ] Perhaps they were referring to the optical disk and the autochanger. These are not CD-ROM per se. Your assessment is otherwise accurate, if we are talking about the same thing - it is awkward for use as a backup device (but it is not low capacity - the autochanger basically gives you removable media and automatically switches between I believe 32 surfaces. It can be mounted. We use it for archiving sources and build environments. >Are gcc and g++ available? Yes, but are nowhere near as good as HP's in terms of generated code quality. I've heard factors of 2 for gcc vs HP's cc on the 700. >How about gdb? What debugging >format does HP use? Is it COFF, stabs, or something peculiar to HP? >If the last, is it documented? Does HP offer dbx? HP uses a proprietary debug format, and a debugger called "xdb". "xdb" is somewhat more powerful than "dbx", but some people find it harder to get used to (it is not as freindly). "xdb"'s biggest advantage is its unrivaled C++ support - if you use HP's C++. >How reliable is the standard C compiler? Are the higher optimization >levels sufficiently reliable to be usable? Is ANSI C included with >HPUX or available separately? Available separately, I think. There will always be stories of optimizers breaking code, and it is hard to get objective numbers. The higher optimization levels include some interprocedural stuff, so in addition to any problems with the optimizer, it uncovers new classes of bugs in programs being optimized (much in the same way register allocators found a lot of people's uninitialized variables, the hard way). >In general, how easy is it to port free software to HPUX? Most things >seem to port to SunOS with little effort. To what extent is this true >of HPUX? If they don't have BSD dependencies, they port fine. That is my gut impression. -------------- Marc Sabatella (marc@hpmonk.fc.hp.com) Disclaimers: 2 + 2 = 3, for suitably small values of 2 Bill and Dave may not always agree with me
mah@dec1.wu-wien.ac.at (Michael Haberler) (04/10/91)
In article <41652@ucbvax.BERKELEY.EDU>, luism@genesis.Berkeley.EDU (Luis Miguel) writes: |> In article <TML.91Apr9103738@nyyti.tik.vtt.fi>, tml@tik.vtt.fi (Tor Lillqvist) writes: |> |> |> |> Does HPUX support long filenames, symlinks, job control, sockets, NFS? |> |> Is it POSIX compliant? |> |> |> |> Yes, yes. Sure. |> |> |> Baloney!!!. I run HPUX 7.0, and it DOES NOT support long filenames No baloney. Consult your FM on newfs and cvtfs. If you install hp-ux it's hard to overlook the point where you're asked for long or short filenames. - michael
dlj@hpfcdc.HP.COM (David Jobusch) (04/10/91)
/ hpfcdc:comp.sys.hp / luism@genesis.Berkeley.EDU (Luis Miguel) / 10:45 am Apr 9, 1991 / >In article <TML.91Apr9103738@nyyti.tik.vtt.fi>, tml@tik.vtt.fi (Tor Lillqvist) writes: >|> >|> Does HPUX support long filenames, symlinks, job control, sockets, NFS? >|> Is it POSIX compliant? >|> >|> Yes, yes. Sure. >|> >Baloney!!!. I run HPUX 7.0, and it DOES NOT support long filenames >(A pain when untarring things with long filenames. In more than one >occasion the clipping of the name meant that some files were written >on top of others). > > >-- >Luis Miguel. Computer Science Division, UC Berkeley. >arpanet: luis@postgres.Berkeley.EDU >uucp: {ihnp4,decvax}!ucbvax!postgres!luis >at&t: (415) 642-8234 >---------- Er, incorrect. HP-UX 7.0 (and of course later) does indeed support long filenames. Check the man pages for "fs(4)", "mkfs(1M)", "newfs(1M)" and "convertfs(1M)". ___ Dave Jobusch / / HP Fort Collins, CO HEWLETT/hp/PACKARD OSSD/UDL Kernel Project /__/ dlj@hpfcls.hp.com
hooft@prl.philips.nl (Peter van Hooft) (04/10/91)
In <41652@ucbvax.BERKELEY.EDU> luism@genesis.Berkeley.EDU (Luis Miguel) writes: >In article <TML.91Apr9103738@nyyti.tik.vtt.fi>, tml@tik.vtt.fi (Tor Lillqvist) writes: >|> >|> Does HPUX support long filenames, symlinks, job control, sockets, NFS? >|> Is it POSIX compliant? >|> >|> Yes, yes. Sure. >|> >Baloney!!!. I run HPUX 7.0, and it DOES NOT support long filenames man 1m convertfs Peter van Hooft Philips Research Labs, Eindhoven, the Netherlands Email: hooft@prl.philips.nl SERI: HOOFT:NLWAYA01 Voice: +31 40744327 X400: /PN=PJG.VanHooft/O=research/PRMD=philips400/ADMD=400net/C=nl/
rodean@hpfcdc.HP.COM (Bruce Rodean) (04/10/91)
In article <41652@ucbvax.BERKELEY.EDU> luism@genesis.Berkeley.EDU (Luis Miguel) writes: >In article <TML.91Apr9103738@nyyti.tik.vtt.fi>, tml@tik.vtt.fi (Tor Lillqvist) writes: >|> >|> Does HPUX support long filenames, symlinks, job control, sockets, NFS? >|> Is it POSIX compliant? >|> >|> Yes, yes. Sure. >|> >Baloney!!!. I run HPUX 7.0, and it DOES NOT support long filenames >(A pain when untarring things with long filenames. In more than one >occasion the clipping of the name meant that some files were written >on top of others). Long file names have been supported on the Series 300 since the 6.2 release which was delivered in mid-1988. Bruce Rodean rodean@hpfclg.fc.hp.com Not an official statement of Hewlett-Packard Co. Opinions expressed are solely mine.
hardy@golem.ps.uci.edu (Meinhard E. Mayer (Hardy)) (04/11/91)
||>In general, how easy is it to port free software to HPUX? Most ||things ||>seem to port to SunOS with little effort. To what extent is this ||true ||>of HPUX? || ||If they don't have BSD dependencies, they port fine. That is my gut ||impression. In fact -- without too much computing experience -- I have managed to compile most PD stuff (with the obvious changes: #include string.h -> #include strings.h and similar annoying trivialities; last example string.h does not define _index0). As soon as enough Cobras populate academic departments and Sun adopts SysV people will stop writing completely Heliocentric software and stop discriminating against minorities. Many thoughtful people do: often make HPUX or make SysV will do the job. Greetings, Hardy -------****------- Meinhard E. Mayer (Hardy); Department of Physics, University of California Irvine CA 92717; (714) 856 5543; hardy@golem.ps.uci.edu or MMAYER@UCI.BITNET
kriso@lombard.dartmouth.edu (Kris Olander) (04/11/91)
Excerpts Re: Switching from Sun to H.. Tor Lillqvist@tik.vtt.fi (3856) ... ... It's interesting to note that Andrew, however, seems to be easier to compile and run on HP-UX (on the Series 300, alas) than on SunOS. -- Tor Lillqvist, working, but not speaking, for the Technical Research Centre of Finland Yeah, but what about a working Typescript program. I'm still waiting for patchlevel 10 so that our hp9000 340s will be able to handle job control, etc., etc... And it would be nice if Console could intercept /dev/console so that I don't get garbage printed out on my X root window. It may compile easier, but the important question is... does it run? -- Kris Olander --------------kris.olander@dartmouth.edu----------------
milburn@me10.lbl.gov (John Milburn) (04/11/91)
kriso@lombard.dartmouth.edu (Kris Olander) writes: >Yeah, but what about a working Typescript program. I'm still waiting >for patchlevel 10 so that our hp9000 340s will be able to handle job >control, etc., etc... And it would be nice if Console could intercept >/dev/console so that I don't get garbage printed out on my X root window. I don't know what Typescript is, but as for console output, the HP X server enables capture of console output. Be sure xterm is suid root, and invoked with the -C option. Works fine for me. >It may compile easier, but the important question is... does it run? I've found very little useful PD stuff which does not run properly, with a minimum of futzing. -jem -- John Milburn milburn@me10.lbl.gov (415) 486-6969 "X windows. It could be worse, but it'll take time." - from a SIGGRAPH flyer
jrc@hpdmd48.boi.hp.com (Jim Conrad) (04/11/91)
Hmm. I worked on HP's LAN Manager/X product. Support for HP-UX Ver 7's long file names was one of the really nasty problems we addressed. They are there, but no one forces you to use them. JimC
decot@hpisod2.cup.hp.com (Dave Decot) (04/11/91)
Here are answers to some of the questions. I don't know reliable answers for the others. However, see the disclaimer at the bottom. > Does HPUX include DWB? No, but Documenter's Work Bench for HP-UX can be obtained from companies such as ELAN. Note, however that nroff, eqn, and tbl are all provided in stock HP-UX. > Does HPUX support long filenames, symlinks, job control, sockets, NFS? Yes, since HP-UX Releases 6.5 [s300] and 2.0 [s800], respectively. > Is it POSIX compliant? Yes. HP-UX has been compliant with POSIX.1 and the XPG3 Base branding criteria since Release 7.0. > Does uucp work reliably at 19200 bps? Is it a solid version of uucp? It is HoneyDanBer UUCP, if that means anything to you. > Does HPUX provide lpr or does it only provide lp? There is no command called "lpr", but HP's lp also provides most of the other services associated with lpr, including remote spooling and cancellation. > Somebody said that OSF/1 was going to be available by the end of this > year. Does this mean that a production version suitable for end users > will be available by then? Does this mean HP will be shipping OSF/1 > as their standard OS by then? Will OSF/1 run HPUX binaries? Nobody in HP can comment on future products. > Is BSD 4.4 likely to be available? If so, will the machine dependent > parts be freely redistributable? If not, does HP document their > hardware enough that it would be possible for somebody to port it? Someone (the University of Utah, I think) has ported BSD 4.3 for HP hardware, so it should be possible. I don't know its distribution status, and obviously can't guarantee anything about future BSD releases. > How good (fast, robust) is HP's implementation of X? Does it include > all of X11R4? Is it reasonably close to X11R4? Does the MIT X11R4 > distribution build on a 9000/730? Is the R5 distribution likely to? > Do HP document their graphics hardware enough to port a window system? Standard benchmarks of "good", "fast", and "robust" have not been developed, to my knowledge. It's certainly seems to be very fast on the Series 700. I think the answer is yes to all the rest of these. > How does HP's 19'' monitor compare to the Trinitron that Sun ships? > How about the keyboard and mouse? How noisy/hot would a system with > two internal 400Mb disks be? Would it be reasonably pleasant in an > office environment? Does it connect directly to thinwire ethernet? Is > HP hardware generally reliable? HP's monitors are all manufactured by Sony, I believe. Reducing environmental noise and dissipated heat are important design objectives for all HP equipment, but since I run diskless on my desk I can't provide first-hand knowledge of this particular set-up. In general, HP hardware is the most reliable available from any major vendor. > How reliable is the standard C compiler? Are the higher optimization > levels sufficiently reliable to be usable? Reliability is relative and somewhat subjective, but I haven't had any problems with either the Classic or ANSI C compilers or their shared optimizer in the recent releases. > In general, how easy is it to port free software to HPUX? Most things > seem to port to SunOS with little effort. To what extent is this true > of HPUX? It depends on the general portability of the particular free software, obviously. I'm sure that free software has been written that only runs on PC-DOS. However, most Un*x-oriented things I've pulled off the net port easily on HP-UX. *********************************************************************** DISCLAIMER: All of the information above is provided for informative purposes only, as my own opinion, and may be false. I am not authorized by HP to make binding statements. An HP sales representative will be able to give you such answers. *********************************************************************** Dave Decot
ian@rathe.cs.umn.edu (Ian Hogg) (04/11/91)
In article <JJC.91Apr8133840@jclark.UUCP> jjc@jclark.UUCP (James Clark) writes: > >If I get some replies, I'll summarize to this group, since I suspect >there are other who have similar concerns to me. If you reply but >don't want your remarks to appear in the summary, or want to remain >anonymous, please say so and I'll be glad to comply. I've got a lot >of questions, but I would like to hear from you even if you want to >answer only one of them. > > >My main concern is about HPUX. My second-hand impression of it is >that it's based on some old version of System V, and that it's a bit >old-fashioned and non-standard, and generally not quite as good as >SunOS. Is this a fair assessment? I find HP-UX to quite solid and has most of the features I'd like except for mmap & threads. These are in OSF/1. > >Does HPUX include DWB? > Yes but it doesn't have troff. I've used xroff from Image Network and several people in the newgroup recommend eroff from Elan. >Does HPUX support long filenames, symlinks, job control, sockets, NFS? >Is it POSIX compliant? > Yes on all counts. (I'm not 100% sure on job control but I think I've diddled with it in ksh. >With the 2-user version of HPUX, does a uucp login count as a user? >Does a rlogin count as a user? > I've had some problems with this in the past. I recommend that you log in to your system via VUE or xdm. If you use VUE HP has a patch for login so that VUE doesn't suck up both users. All logins via the network count as one user. So if you log in via VUE and severl other people are rlogin'ed or telnet'ed in you will only count as one user. This leaves one login available for uucp. >Does uucp work reliably at 19200 bps? Is it a solid version of uucp? > I ran UUCP at 19200 off a MUX card. Some people have pointed out the limitations of the built in serial card. >Does HPUX provide lpr or does it only provide lp? Does it support >PostScript printers? > Lp but it does support Berkely remote spooling. I had to roll my own stuff to spool to a postscript printer. If you still have a Sun laying around, attach your Postscript printer to it. >Somebody said that OSF/1 was going to be available by the end of this >year. Does this mean that a production version suitable for end users >will be available by then? Does this mean HP will be shipping OSF/1 >as their standard OS by then? Will OSF/1 run HPUX binaries? > I'll let someone from HP handle this one. >Is BSD 4.4 likely to be available? If so, will the machine dependent >parts be freely redistributable? If not, does HP document their >hardware enough that it would be possible for somebody to port it? > I think you can buy BSD from Mt. Xinu. >How good (fast, robust) is HP's implementation of X? Does it include >all of X11R4? Is it reasonably close to X11R4? Does the MIT X11R4 >distribution build on a 9000/730? Is the R5 distribution likely to? >Do HP document their graphics hardware enough to port a window system? > HP's X is fasr and robust. Do not use the X server off the MIT tape. It is significantly slower. Optimized servers have been place on some HP machine somewhere. I don't know exactly what server you'll get with your system. I was running HP's server (R3 based but has R4 extensions like SHAPE) with the MIT R4 Xlib, Xt, Xaw, and Motif 1.1 (we licensed it from OSF). I don't know when HP will ship R4 libs and Motif 1.1. >How good is HP hardware and software support? Is it reasonably >priced? Can one access support by email? > >How does HP's 19'' monitor compare to the Trinitron that Sun ships? >How about the keyboard and mouse? How noisy/hot would a system with >two internal 400Mb disks be? Would it be reasonably pleasant in an >office environment? Does it connect directly to thinwire ethernet? Is >HP hardware generally reliable? > We've had very good experiences with HP. Connects to either thin or thick ethernet (at least 375 and 400's do). >I'm concerned about exchanging data with people who don't have HPs. >Can I get a third party SCSI tape that handles Sun-compatible QIC-24 >or QIC-150 tapes? I don't know, I always read/wrot QIC tapes via rsh and a SUN. I have used Exabyte's to transfer data. > >My salesman said that 400Mb was the biggest external disk that HP >offered on the 730. This seems a bit on the small side. Is HP likely >to offer bigger disks, or does it want me to buy a 750 if I need lots >of disk space? Can I use third-party SCSI disks? > We don't buy any disks from HP. We buy our own usually Imprimis Sabres, Wrens, but haven't tried Elites. I posted a program to this newsgroup that generates "/etc/disktab" entries for Imprimis disk drives (haven't triesd it with other drives) >In addition to DAT, I was told there was a writable CD-ROM available >(C1701A or something like that). Is this intended as an alternative >to DAT as a mass-backup device? I suspect not, since it seems to be >slower, have lower capacity, and be more expensive than DAT. Can it >mounted like a normal disk, or does one treat it like a tape? Would >CD-ROMs written with this drive be readable with Sun's CD-ROM drive? > I used Exabytes for backup. You can either buy as an OEM from Exabyte (I think we payed about $2500) or buy from various suppliers. I am quite happy with Perfect Byte because they provide a device driver and monitor program. >Are gcc and g++ available? If not, do HP provide enough information >about their architecture that somebody could port them? If so, is the >code quality competitive with HP? How about gdb? What debugging >format does HP use? Is it COFF, stabs, or something peculiar to HP? >If the last, is it documented? Does HP offer dbx? > The 900/8xx versions of these should work. >How reliable is the standard C compiler? Are the higher optimization >levels sufficiently reliable to be usable? Is ANSI C included with >HPUX or available separately? > At last check it was available separately. >How good is HP's C++ (Softbench) offering? Is it based on cfront 2.0 >or 2.1? How good is the debugger? Is it as good as Saber C++? Is >Saber C++ likely to become available? > I was very happy using C++ (Softbench). I was a beta site last fall for C++ Softbench. The beta software was based on 2.1 (I think). HP and Saber have announced Saber C++ for Softbench. Last summer I had to develop a demo to be presented at the IEEE Power Society Summer Meeting. I had 1 month to do it in and it was based on HP OpenView. I was the first person to actually use OpenView. So, in one month I learned Softbench, OpenView and developed the demo. I doubt I could have done it in two months without Softbench. >In general, how easy is it to port free software to HPUX? Most things >seem to port to SunOS with little effort. To what extent is this true >of HPUX? > I've ported quite a few things and the degree of difficulty is varying. I've built up a library of BSD routines for HP-UX. I don't think porting is all that difficult. >I would also be interested in any other random comments that you might >have about the relative merits of Sun and HP workstations. > I would rather have a 9000/375 than the SparcStation I I have know. The new 700's are the obvious choice. >James Clark >jjc@jclark.uucp >jjc%jclark@mcsun.eu.net >jjc@ai.mit.edu -- Ian Hogg email: ian@rathe.cs.umn.edu ...!umn-cs!rathe!ian Rathe, Inc ianhogg@cs.umn.edu 366 Jackson Street phone: (612) 225-1401
fkittred@bbn.com (Fletcher Kittredge) (04/12/91)
In article <TML.91Apr9103738@nyyti.tik.vtt.fi> tml@tik.vtt.fi (Tor Lillqvist) writes: >In article <JJC.91Apr8133840@jclark.UUCP> jjc@jclark.UUCP (James Clark) writes: > >But you are right that SunOS does have some nice features that HP-UX hasn't >got (yet). Like dynamic linking, enhanced ptrace(2), interesting new >filesystem types, automounter... It's not fair to call these featrures >"standards," though. On the other hand, the HP-UX approach to diskless >workstations might be better than SunOS's. This is out of date. HP-UX 8.0.1 does include dynamic linking. However, automounter not being there is a major drawback, though you can use AMD. Fletcher Kittredge Platforms and Tools Group, BBN Software Products 10 Fawcett Street, Cambridge, MA. 02138 617-873-3465 / fkittred@bbn.com / fkittred@das.harvard.edu
gates@hpfcdc.HP.COM (Bill Gates) (04/12/91)
>>Does HPUX support long filenames, symlinks, job control, sockets, NFS? >>Is it POSIX compliant? > >Yes on all counts, depending on how you define "POSIX compliant" - there are >many POSIX specs, and various different suites to measure compliance. I >don't know our full story, but I know that 8.0 contains a lot of code in the >commands, libraries, and kernel for drafts 1 and 2. HP-UX 8.0 will be base compliant with XPG3, POSIX 1003.1, FIPS 151-1, and SVID Issue 2, volumes 1, 2, and 3. Additional work has been done towards conformance with POSIX 1003.2 (actually targeted at conformance with a Federal FIPS published in the Federal Register which never actually was adopted). XPG4 conformance work has also been done in 8.0, but only in those areas which agree with 1003.2. 8.0 won't be fully compliant with 1003.2 or XPG4 due to the lack of final standards. Bill
fkittred@bbn.com (Fletcher Kittredge) (04/12/91)
In article <41652@ucbvax.BERKELEY.EDU> luis@postgres.berkeley.edu writes: >In article <TML.91Apr9103738@nyyti.tik.vtt.fi>, tml@tik.vtt.fi (Tor Lillqvist) writes: >|> >|> Does HPUX support long filenames, symlinks, job control, sockets, NFS? >|> Is it POSIX compliant? >|> >|> Yes, yes. Sure. >|> >Baloney!!!. I run HPUX 7.0, and it DOES NOT support long filenames >(A pain when untarring things with long filenames. In more than one >occasion the clipping of the name meant that some files were written >on top of others). No, you are wrong. HP-UX has supported long file names since version 2.0. You don't have your system configured properly. Check the man page for SAM and try to be less vehement in your erroneous assertions. Fletcher Kittredge Platforms and Tools Group, BBN Software Products 10 Fawcett Street, Cambridge, MA. 02138 617-873-3465 / fkittred@bbn.com / fkittred@das.harvard.edu
ian@rathe.cs.umn.edu (Ian Hogg) (04/12/91)
In article <1991Apr10.180712.7815@dartvax.dartmouth.edu> kriso@lombard.dartmouth.edu (Kris Olander) writes: >Excerpts Re: Switching from Sun to H.. Tor Lillqvist@tik.vtt.fi (3856) > > ... > ... > It's interesting to note that Andrew, however, seems to be > easier to compile and > run on HP-UX (on the Series 300, alas) than on SunOS. > -- > Tor Lillqvist, > working, but not speaking, for the Technical Research Centre of Finland > > > >Yeah, but what about a working Typescript program. I'm still waiting >for patchlevel 10 so that our hp9000 340s will be able to handle job >control, etc., etc... And it would be nice if Console could intercept >/dev/console so that I don't get garbage printed out on my X root window. > I don't know what the problem with Console is, but I ported xconsole to HP-UX and have an X window intercept /dev/console. You can also do this with an hpterm. >It may compile easier, but the important question is... does it run? > > >-- >Kris Olander >--------------kris.olander@dartmouth.edu---------------- -- Ian Hogg email: ian@rathe.cs.umn.edu ...!umn-cs!rathe!ian Rathe, Inc ianhogg@cs.umn.edu 366 Jackson Street phone: (612) 225-1401
hardy@golem.ps.uci.edu (Meinhard E. Mayer (Hardy)) (04/13/91)
Aas an addendum to Ian's remark about disks: I have had no problem adding a Maxtor 8730-S 650 MB (that'ts what I ultimately got out of it) drive; I hav used Ian's disktab-generating program as a first approximation, then fine-tuned news by hand. Much faster and much quieter than the original HP 7939S which cost more 2.5 times more! Take note HP-pricemakers! Greetings, Hardy -------****------- Meinhard E. Mayer (Hardy); Department of Physics, University of California Irvine CA 92717; (714) 856 5543; hardy@golem.ps.uci.edu or MMAYER@UCI.BITNET
campbelr@hpcuhe.cup.hp.com (Bob Campbell) (04/13/91)
>> How does HP's 19'' monitor compare to the Trinitron that Sun ships? > HP ships Trinitron monitors, too. The 700 series uses monitors that run at 72Hz, the old Trinitrons will not work on the new workstations and I do not think Trinitrons are currently made to run at 72Hz. --------------------------------------------------------------------------- Bob Campbell Some times I wish that I could stop you from campbelr@hpda.cup.hp.com talking, when I hear the silly things you say. Hewlett Packard - Elvis Costello
lang@hpfcso.FC.HP.COM (John J. Lang) (04/13/91)
/ hpfcso:comp.sys.hp / jjc@jclark.UUCP (James Clark) / 6:38 am Apr 8, 1991 / > With the 2-user version of HPUX, does a uucp login count as a user? > Does a rlogin count as a user? I don't know about uucp logins. I don't think rlogin's count. > Does HPUX provide lpr or does it only provide lp? Does it support > PostScript printers? Some people have said no, but there is a /usr/bin/lpr file on the S700 I have access to. It does support PostScript Printers. > How does HP's 19'' monitor compare to the Trinitron that Sun ships? HP does have Sony Monitors, but they are not used by the low end Graphics devices supported by the S700 (CRX and GRX). These use a 72 Hz Hitachi monitor, which is not as good as the 60Hz Sony on my s400. There is a 72 Hz Sony, but I have not information on how compatible or expensive it is. As far as I know it is not being offered as an option by HP. > How about the keyboard and mouse? How noisy/hot would a system with > two internal 400Mb disks be? Would it be reasonably pleasant in an > office environment? I have had both sitting on/under my desk, and have not noticed a problem. I find the keyboard and mouse to be fine. > Does it connect directly to thinwire ethernet? Yes. > How reliable is the standard C compiler? Are the higher optimization > levels sufficiently reliable to be usable? Is ANSI C included with > HPUX or available separately? We compile about a 'billion' lines of graphics code with the standard C compiler, using full optimization. There have been problems in the past, most have been fixed, a few have been worked around. I think ANSI C is included, and uses the same optimizer.
bdsz@cbnewsl.att.com (bruce.d.szablak) (04/13/91)
I understand that HP-UX runs NFS 3.2 instead of NFS 4.0. Now, can anyone tell me what difference does that make?
anthony@nntp-server.caltech.edu (Lawrence Anthony) (04/13/91)
In article <7370361@hpfcso.FC.HP.COM> lang@hpfcso.FC.HP.COM (John J. Lang) writes: >/ hpfcso:comp.sys.hp / jjc@jclark.UUCP (James Clark) / 6:38 am Apr 8, 1991 / >> How reliable is the standard C compiler? Are the higher optimization >> levels sufficiently reliable to be usable? Is ANSI C included with >> HPUX or available separately? > >I think ANSI C is included, and uses the same optimizer. my local hp sales rep tells me that C/ANSI C is NOT bundled with the base HPUX operating system. he claims that HPUX includes a stripped-down version of the C compiler that contains neither a debugger nor an optimizer. (pretty slimy, if you ask me.) any hp folk care to confirm or refute this claim?
tml@tik.vtt.fi (Tor Lillqvist) (04/13/91)
In article <1991Apr11.141050.5207@rathe.cs.umn.edu> ian@rathe.cs.umn.edu (Ian Hogg) writes: >Does HPUX include DWB? Yes but it doesn't have troff. I've used xroff from Image Network and several people in the newgroup recommend eroff from Elan. Uh, doesn't DWB, besides n/troff, eql, tbl, also contain programs like pic and those writing style checkers whose names I can't remember? Anyway, none of this is in HP-UX either. Just neqn, tbl and nroff. I recommend GNU roff, which has eqn, tbl, pic and ditroff equivalents. (Wasn't it the GNU roff author who posted the original article...) BTW, for those who haven't got C++, I guess I could put up pre-compiled HP-UX (s300 & s800) binaries of the GNU roff programs for ftp. Anyone interested? -- Tor Lillqvist, working, but not speaking, for the Technical Research Centre of Finland
smith@wallaby.mrc.uidaho.edu (04/13/91)
In article <1991Apr13.005946.9689@nntp-server.caltech.edu> anthony@nntp-server.caltech.edu (Lawrence Anthony) writes: >In article <7370361@hpfcso.FC.HP.COM> lang@hpfcso.FC.HP.COM (John J. Lang) writes: >>/ hpfcso:comp.sys.hp / jjc@jclark.UUCP (James Clark) / 6:38 am Apr 8, 1991 / >>> How reliable is the standard C compiler? Are the higher optimization >>> levels sufficiently reliable to be usable? Is ANSI C included with >>> HPUX or available separately? >> >>I think ANSI C is included, and uses the same optimizer. > >my local hp sales rep tells me that C/ANSI C is NOT bundled with the base >HPUX operating system. he claims that HPUX includes a stripped-down version >of the C compiler that contains neither a debugger nor an optimizer. (pretty >slimy, if you ask me.) any hp folk care to confirm or refute this claim? This is wrong. First of all, C for the 800/700 series is ANSI complient. The C for the 300/400 series is not ANSI, is a very nice full featured C with many ANSI extensions. On the 300/400 version of HPUX 7.0 there is a ANSI C preprocessor (/lib/cpp.ansi). This leads me to believe the rumer that I have heard that HPUX 8.0 will have ANSI C for all hardware platforms. The compiler seems to be mostly bug free. We have compiled tens of thousands of lines of code and have had no noticable problems. The compiler does have an optimizer. Its good, but doesn't seem to be quite as good as gcc. HPUX also contains cdb as a standared debugger. This works in split screen mode and works fairly well. William Smith Microelectronics Research Center University of Idaho, Moscow, ID 83843 (208)885-6500 E-mail: wsmith@groucho.mrc.uidaho.edu
garvey@johnny5.uucp (Joe Garvey) (04/14/91)
In article <16710082@hpisod2.cup.hp.com>, decot@hpisod2.cup.hp.com (Dave Decot) writes: > > Does HPUX include DWB? > > No, but Documenter's Work Bench for HP-UX can be obtained from companies > such as ELAN. Note, however that nroff, eqn, and tbl are all provided in > stock HP-UX. But, HP actively discourages use of these tools by suspending internal support and upgrades for them years ago. Bugs go unfixed, and not everything you'd expect is there. I know, I bitched about it... and got the above answer. It is marginally useful... portable is another thing. > > Does HPUX support long filenames, symlinks, job control, sockets, NFS? > > Yes, since HP-UX Releases 6.5 [s300] and 2.0 [s800], respectively. Long file name support is there only if you've converted your file systems. If you're writing commercial software you'll have to work with the 14 character limit. Not all third parties have adapted to the 6.0 changes much less 6.5 and 7.0. There are issues regarding building your own libraries too. The 14 character limit crops up again. > > Does uucp work reliably at 19200 bps? Is it a solid version of uucp? > > It is HoneyDanBer UUCP, if that means anything to you. Yes, HP runs HDB, and is solid an 19200. If you're using a series 300 machine the internal port is a no-no (it is serviced by main CPU... and thus slows things down). You'll want a 4 port mux... of which only one port can be used with a modem. Argh. > > Does HPUX provide lpr or does it only provide lp? > > There is no command called "lpr", but HP's lp also provides most of the > other services associated with lpr, including remote spooling and > cancellation. The printing subsystem is based on SysV. It uses a printer script and there is no printcap (ala BSD printer subsystem). Having used both I prefer the System V... I found it easier to bend to my will. > > Somebody said that OSF/1 was going to be available by the end of this > > year. Does this mean that a production version suitable for end users > > will be available by then? Does this mean HP will be shipping OSF/1 > > as their standard OS by then? Will OSF/1 run HPUX binaries? > > Nobody in HP can comment on future products. BS. Some do, and marketing/sales advertises. HP has a clear path mapped to OSF/1. You'll see pieces of it as time goes on. I beleive last I heard (probably out of date) that 8.0 was to be OSF/1. HP is *very* good about compatiblity issues. But this is the realease where the rubber meets the road. All of Apollo gets folded in. I'd expect this to be difficult to do. Count on HP to do its best (better than you'll ever see from Sun). As with such an extrodinary effort expect delays and a significant number of small problems. HP's OS QA is very good. It'll be their applications that really hickup ME10, database, etc. > > Is BSD 4.4 likely to be available? If so, will the machine dependent > > parts be freely redistributable? If not, does HP document their > > hardware enough that it would be possible for somebody to port it? > > Someone (the University of Utah, I think) has ported BSD 4.3 for HP > hardware, so it should be possible. I don't know its distribution > status, and obviously can't guarantee anything about future BSD releases. HP has pretty good BSD compatibility. To do this they have a special BSD library that you link in. I expect HP will keep it reasonably up to date. > > How good (fast, robust) is HP's implementation of X? Does it include > > all of X11R4? Is it reasonably close to X11R4? Does the MIT X11R4 > > distribution build on a 9000/730? Is the R5 distribution likely to? > > Do HP document their graphics hardware enough to port a window system? > Standard benchmarks of "good", "fast", and "robust" have not been > developed, to my knowledge. It's certainly seems to be very fast on > the Series 700. I think the answer is yes to all the rest of these. I agree. HP does a solid job here. X is really to be considered part of the OS groups responsiblity. It'll be clean (er than Sun's for sure). Note: HP does add a few of their own eccentricites (compressed fonts for example). This makes X-terminals barf... just uncompress them and put them in their own directory. > > How does HP's 19'' monitor compare to the Trinitron that Sun ships? > > How about the keyboard and mouse? How noisy/hot would a system with > > two internal 400Mb disks be? Would it be reasonably pleasant in an > > office environment? Does it connect directly to thinwire ethernet? Is > > HP hardware generally reliable? > > HP's monitors are all manufactured by Sony, I believe. Reducing environmental > noise and dissipated heat are important design objectives for all HP > equipment, but since I run diskless on my desk I can't provide first-hand > knowledge of this particular set-up. In general, HP hardware is the > most reliable available from any major vendor. HP's monitors are superior in resolution and clarity to what Sun ships. I consider Sun's monitors to be PC class in resolution, clarity, and quality. HP's monitors are professional. Higher resolution (if you opt for it), always clearer, and very reliable (come on now, we're talking HP here, hardware quality is a specailty of this company). > > How reliable is the standard C compiler? Are the higher optimization > > levels sufficiently reliable to be usable? > > Reliability is relative and somewhat subjective, but I haven't had any > problems with either the Classic or ANSI C compilers or their shared > optimizer in the recent releases. I believe ANSI C is avaliable. No, the conventional wisdom (last I heard, though there haven't been enough releases to convince me this has changed) is to avoid the higher optimizations. > > In general, how easy is it to port free software to HPUX? Most things > > seem to port to SunOS with little effort. To what extent is this true > > of HPUX? HPUX is an excellent integration of BSD/SYSV. You should have very little effort porting (esp if you designed your software to do so). I've ported quite a few of my non-portably-written applications with little to no trouble. If a problem occurred is was almost always my fault (for doing something sloppy). You should expect few problems here. Like I said before, expect some of the libraries to change (headers too, probably). > It depends on the general portability of the particular free software, > obviously. I'm sure that free software has been written that only runs > on PC-DOS. However, most Un*x-oriented things I've pulled off the net > port easily on HP-UX. Almost all free software (and I've compiled a lot on HP) compiles without problems. GNU stuff has been tough in the past... but is reportedly getting quite a bit easier. Enough that you shouldn't worry about this. > *********************************************************************** > DISCLAIMER: All of the information above is provided for informative > purposes only, as my own opinion, and may be false. I am not authorized > by HP to make binding statements. An HP sales representative will > be able to give you such answers. In general no, an HP sales rep can't (I SAID IN GENERAL, REREAD THE BEGINNING OF THE LINE). My experience says you'll have to weasel past the sales rep to the SE. The SE should know, or point you to someone who knows (SE's specailize, so if you change the subject area, you may well need to talk to another SE). > *********************************************************************** > > Dave Decot Who'm I? A HP system admin/programmer in the HP environment for 4+ years. I really dug in the HP stuff. The comments above are based on my experinces. I'm also a Sun admin/programmer for 1yr-. Other things you should know. HP wrote their own networking years ago. It's a little wierd. They're fixing it. 8.0 is the magic release again. They may not actually be planing to call it 8.0... but it should give you the right idea. HP support is quite a bit better that Sun support. In general they are better trained, and more able to solve problems. I suspect this derives from organizational differences more than anything. You also get pretty good help from the open net from HP (better than what I've experienced from Sun). On the other hand, everyone in Sun uses e-mail... that makes things easier to do... like report bugs. HP's learning, but it takes time to accelerate such a large mass. Then again HP doesn't have everyone and their brother calling... making the customer base HP has to work with better... maybe the "snakes" will change that :-). You can also plan on a little annoyance adjusting to a different set of manuals... no better... no worse, just different. HP's ability to support diskless/dataless nodes is vastly superior to Sun's. This is quite a (un)pleasant surprise. Depending on the direction you're migrating. Hope that helps. -- Joe Garvey uucp: sumax!quick!johnny5!garvey J5 Research map entries are wrong for johnny5. They're Bothell, Wa. being fixed. Please use address above.
steve-t@hpfcso.FC.HP.COM (Steve Taylor) (04/15/91)
In /comp.sys.hp/ smith@wallaby.mrc.uidaho.edu / quotes & writes: | anthony@nntp-server.caltech.edu (Lawrence Anthony) writes: | > my local hp sales rep tells me that C/ANSI C is NOT bundled with the base | > HPUX operating system. he claims that HPUX includes a stripped-down version | > of the C compiler that contains neither a debugger nor an optimizer. | This is wrong. First of all, C for the 800/700 series is ANSI complient. | The C for the 300/400 series is not ANSI, is a very nice full featured C | with many ANSI extensions. On the 300/400 version of HPUX 7.0 there is a | ANSI C preprocessor (/lib/cpp.ansi). This leads me to believe the rumer that | I have heard that HPUX 8.0 will have ANSI C for all hardware platforms. Starting with 8.0, C/ANSI C on the 700/800 is unbundled. The "RunTime" (base) system contains a C compiler which is not ANSI, does not generate debug info, (I'm not sure about no optimization ...) which is used to rebuild kernels, handle shar files, and such things. If you're buying a bundle greater than the "RunTime" system, it may include the full compiler (some do). C/ANSI C for the 300/400 became available post 7.0 release. The current version of the 300/400 compilers (C, Fortran, Pascal) is 7.40. These are separate products also, not bundled with 7.0x HP-UX. | HPUX also contains cdb as a standard debugger. This works in split | screen mode and works fairly well. Thanks. Note that we tend to think of xdb as the standard debugger, since it appears on series 300/400, series 700, series 600/800, and series 900 (MPE). cdb is the same debugger with a different command syntax which only appears on the series 300/400. Regards, Steve taylor NOT A STATEMENT, OFFICIAL OR OTHERWISE, OF THE HEWLETT-PACKARD COMPANY.
nenaas@ulrik.uio.no (Nils-Eivind Naas) (04/15/91)
In article <HARDY.91Apr12105934@golem.ps.uci.edu> hardy@golem.ps.uci.edu (Meinhard E. Mayer (Hardy)) writes:
Aas an addendum to Ian's remark about disks:
I have had no problem adding a Maxtor 8730-S 650 MB (that'ts what I
ultimately got out of it) drive; I hav used Ian's disktab-generating
program as a first approximation, then fine-tuned news by hand.
Much faster and much quieter than the original HP 7939S which cost
more 2.5 times more!
Take note HP-pricemakers!
Greetings,
Hardy
-------****-------
Meinhard E. Mayer (Hardy); Department of Physics, University of California
Irvine CA 92717; (714) 856 5543; hardy@golem.ps.uci.edu or MMAYER@UCI.BITNET
Apparently things work differently in different countries; in Norway at least,
HP peripherals are also sold through distributors who charge quite low
prices. Disk prices are comparable to what is usually charged for Seagate
Imprimis disks. The only difference is that HP disks carry a 5-year guarantee!
Nils-Eivind Naas nen@isaf.no or
ISAF, Oslo,Norway nenaas@ulrik.uio.no
jim@cs.strath.ac.uk (Jim Reid) (04/15/91)
In article <1991Apr14.053212.24266@johnny5.uucp> garvey@johnny5.uucp (Joe Garvey) writes:
HPUX is an excellent integration of BSD/SYSV. You should have very little
effort porting (esp if you designed your software to do so).
I must disagree with this. HP-UX is an awful thing to port code to.
The kernel is BSD dressed up to look like System V. [The older
releases were infuriating because the kernel contained the code for
things like long file names and reliable signals, but wouldn't let you
run it for the sake of SVID compilance!] HP has its own object file
format which makes life hard if you're writing a debugger/linker etc.
(It was really painful to get gcc up on HP boxes.)
Practically nothing can be installed without tweaking makefiles or
config files. In my experience, it has never, ever been possible to
install anything on HP-UX simply by reading in the tape and just typing
make, as can be done on other platforms like SunOS. This is because of
the schizophrenic SysV/BSD OS. If you say the OS is System V, most
software assumes you don't have sockets (you've got streams instead)
and you've got COFF format. Neither assumption is true for HP-UX. If
you say you've got BSD, the compiles fail because they see SysV
include files and #defines.
Unfortunately, not many people write code to cover the case where an
OS has BSD sockets, System V include files, a non-standard object
format, a System V C library with some BSD bits flung in (and others
in a "compatibility" library) and numerous other weirdoes.
Jim
frank@grep.co.uk (Frank Wales) (04/15/91)
In article <1991Apr13.005946.9689@nntp-server.caltech.edu> anthony@nntp-server.caltech.edu (Lawrence Anthony) writes: >my local hp sales rep tells me that C/ANSI C is NOT bundled with the base >HPUX operating system. he claims that HPUX includes a stripped-down version >of the C compiler that contains neither a debugger nor an optimizer. (pretty >slimy, if you ask me.) any hp folk care to confirm or refute this claim? Your sales rep is broken. According to sales literature I have to hand (plus the HP9000/345 I'm writing this on), ANSI C is a separate product on the *300s* and is likely to stay that way -- HP-UX on these machines continues to ship with a pcc-derived C compiler (maybe this applies to the 400s too, I don't know). HP-UX for PA machines (700/800) has ANSI C bundled. All versions of HP-UX ship with decent debuggers and fairly reliable optimisers, although the code quality seems better on PA. -- Frank Wales, Grep Limited, [frank@grep.co.uk<->uunet!grep!frank] Kirkfields Business Centre, Kirk Lane, LEEDS, UK, LS19 7LX. (+44) 532 500303
scottm@spiderman.spider.co.uk (Scott Mackie) (04/16/91)
In article <JIM.91Apr15115715@baird.cs.strath.ac.uk>, jim@cs.strath.ac.uk (Jim Reid) writes: |> In article <1991Apr14.053212.24266@johnny5.uucp> garvey@johnny5.uucp (Joe Garvey) writes: |> |> HPUX is an excellent integration of BSD/SYSV. You should have very little |> effort porting (esp if you designed your software to do so). |> |> I must disagree with this. HP-UX is an awful thing to port code to. I've got to agree with this sentiment. The only code I've ever been able to port without endless tweaks is the X11R4 libraries and clients (and that's only because I've done it so many times for other people). The GNU stuff especially has been a pain (I even got assembler patches from HP Support for Emacs 18.55) although I'm told recent releases have better support for HPUX. |> Unfortunately, not many people write code to cover the case where an |> OS has BSD sockets, System V include files, a non-standard object |> format, a System V C library with some BSD bits flung in (and others |> in a "compatibility" library) and numerous other weirdoes. |> |> Jim Unless you're working on a MIPS box!! ;-) Scott......... -- Spider Systems Limited Net: scottm@spider.co.uk Spider Park, Stanwell Street, Edinburgh YellNet: +44 31 554 9424 #include <disclaimer.h> "Rockin' Good, Peanut!"
tml@tik.vtt.fi (Tor Lillqvist) (04/16/91)
In article <JIM.91Apr15115715@baird.cs.strath.ac.uk> jim@cs.strath.ac.uk (Jim Reid) writes:
non-standard object format,
I didn't know there was a standard for object formats... unless you
define standard as "Version 7 + various hacks" or "SunOS"?
HP-UX is hardly any worse than other System V / BSD hybrids. Remember
that all the world is not a VAX, nor a Sun3.
--
Tor Lillqvist,
working, but not speaking, for the Technical Research Centre of Finland
hwt@bwdlh490.BNR.CA (Henry Troup) (04/17/91)
In article <5570604@hpfcdc.HP.COM>, rodean@hpfcdc.HP.COM (Bruce Rodean) writes: |>Long file names have been supported on the Series 300 since the 6.2 |>release which was delivered in mid-1988. Check out convertfs(1M). It's a one-way conversion, and your administrator may not have done it (or want to). Henry Troup - HWT@BNR.CA (Canada) - BNR owns but does not share my opinions
perry@hpfcdc.HP.COM (Perry Scott) (04/18/91)
>> > Does uucp work reliably at 19200 bps? Is it a solid version of uucp? >> >> It is HoneyDanBer UUCP, if that means anything to you. > >Yes, HP runs HDB, and is solid an 19200. If you're using a series 300 machine >the internal port is a no-no (it is serviced by main CPU... and thus slows >things down). The basenote specifically mentions the 720. The 720 has two internal buffered serial ports. Between the power of the 720, and the 32(?) byte FIFO on the chip, there shouldn't be any overruns. >You'll want a 4 port mux... of which only one port can be >used with a modem. Argh. I was the midwife for the 98638A, an 8-port modem MUX for the 300/400. It takes up one wide (CPU-sized) slot. Perry Scott HP Ft Collins
garvey@johnny5.uucp (Joe Garvey) (04/18/91)
> I didn't know there was a standard for object formats... unless you > define standard as "Version 7 + various hacks" or "SunOS"? Yes, there is a "standard" format. It's called COFF (Common Object File Format). You can expect to see more of it. I know that some of HP's embedded processor development tools use it, and my SparcStation has a man page... Available only on Sun 386i systems running a SunOS 4.0.x release or earlier. Not a SunOS 4.1 release feature. So, it's gradually becoming *the* standard. BTW, I even have the Nutshell handbook on it... you may want to pick up a copy if you're interested in learning more about it. -- Joe Garvey uucp: sumax!quick!johnny5!garvey J5 Research map entries are wrong for johnny5. They're Bothell, Wa. being fixed. Please use address above.
tay@hpcvlx.cv.hp.com (Mike Taylor) (04/19/91)
>Long file name support is there only if you've converted your file systems. >If you're writing commercial software you'll have to work with the 14 character >limit. Not all third parties have adapted to the 6.0 changes much less 6.5 >and 7.0. There are issues regarding building your own libraries too. The 14 >character limit crops up again. It seems like it would be possible for your software to check to see if long filenames have been enabled during installation or startup by touching a 15+ character filename and checking the results. Then tell the user to convert since some people don't until they have to (I haven't heard of any good reasons not to convert). Also, explain the dependency in the sales literature's system requirements. The makefile for your own libraries could have a similar check. Peace, Mike Taylor Current Products Engineering & Online Interface Technology Operation ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ "I get stranger things than you free with my breakfast cereal!" - Zaphod Beeblebrox
lynch@theory.tn.cornell.edu (Tim Lynch) (04/20/91)
In article <1991Apr16.085958.17880@spider.co.uk> scottm@spider.co.uk writes: >The GNU stuff especially has been a pain (I even got assembler patches from >HP Support for Emacs 18.55) although I'm told recent releases have better >support for HPUX. I don't understand. We're running 18.55 on an 800 series machine and built it with no problem. No assembler patches needed. What version of HP-UX you referring to, Scott? >Scott......... >Spider Systems Limited Net: scottm@spider.co.uk >Spider Park, Stanwell Street, Edinburgh YellNet: +44 31 554 9424 Tim Lynch, Mann Library, Cornell University (607) 255 1381 Ithaca, New York 14853-4301 lynch@albert.mannlib.cornell.edu
scottm@spiderman.spider.co.uk (Scott Mackie) (04/22/91)
In article <1991Apr19.200733.422@batcomputer.tn.cornell.edu>, lynch@theory.tn.cornell.edu (Tim Lynch) writes: |> In article <1991Apr16.085958.17880@spider.co.uk> scottm@spider.co.uk writes: |> >The GNU stuff especially has been a pain (I even got assembler patches from |> >HP Support for Emacs 18.55) although I'm told recent releases have better |> >support for HPUX. |> |> I don't understand. We're running 18.55 on an 800 series machine and built |> it with no problem. No assembler patches needed. What version of HP-UX |> you referring to, Scott? |> Well, a quick look at uname says $ uname -a HP-UX spiderma 7.0 B 9000/375 spiderma Not that old.... The patches that I applied were to some 68XXX assembler in the src dir... I think it was alloca.s. Obviously that doesn't apply to the 800's.... I can dig out the notes I got from HP Support if anyone's interested... Scott........... -- Spider Systems Limited Net: scottm@spider.co.uk Spider Park, Stanwell Street, Edinburgh YellNet: +44 31 554 9424 #include <disclaimer.h> "Rockin' Good, Peanut!"
jbb@hpcvlx.cv.hp.com (Jim B. Byers) (04/23/91)
> HP's X is fasr and robust. Do not use the X server off the MIT tape. It > is significantly slower. Optimized servers have been place on some HP > machine somewhere. > I don't know exactly what server you'll get with your system. > I was running HP's server (R3 based but has R4 extensions like SHAPE) with > the MIT R4 Xlib, Xt, Xaw, and Motif 1.1 (we licensed it from OSF). I don't > know when HP will ship R4 libs and Motif 1.1. The HP-UX server that ships with 7.03 is an R4 server. There is a version of this server that runs fast on the 98550 graphics card available on hpcvaaz (15.255.72.15) in the XServers directory. This is availabe via anonymous ftp and is supported on all 7.x series 300 and 400s. An as yet unsupported Series 800 server is available in the unsupported directory. Note that these servers all have been based on R4. NOTE: When querried they will respond that they are vendor release 3. This is because we started shipping X11 with 6.2 and so are on our third major release. Only MIT is guaranteed to have a vendor release that is the same as the R# of the release (by definition). Therefore the vendor release number can have no relation to the actual MIT release it coresponds to. Don't be confused by it. R4 libraries compiled for 7.x are available in the MitX11R4 directory. R4, Motif 1.1 libraries and servers are standard with HP-UX 8.0. Also scalable typefacces in X11 are shipped as standard in 8.0. Motif 1.1 libraries can be purchased for 7.0 by ordering product B1177A. Hope this helps! Jim Byers Interface Technology Operation Hewlett Packard "Have you generated 1000 lines of Motif source code with a click of a button using HP Interface Architect today?"
mjs@hpfcso.FC.HP.COM (Marc Sabatella) (04/23/91)
Re: difficulties porting Gnu Emacs You may be referring to the patched crt0.s you need, not any patch for the assembler per se. See, these Gnu folks don't adhere to standards, and think that because they wrote an assembly version of the system-dependent startup code for BSD or whatever, that it should work on other systems? Baloney. You aren't supposed to be mucking with startup code. Same thing happened when Sun changed their startup conventions when they to shared libraries at 4.0. Software that was written to not depend on system dependent things like startup code or BSD-isms that aren't part of the Unix *standards* like SVID, Posix, etc will port with little or no difficult. >> I didn't know there was a standard for object formats... unless you >> define standard as "Version 7 + various hacks" or "SunOS"? > >Yes, there is a "standard" format. It's called COFF (Common Object File >Format). You can expect to see more of it. I know that some of HP's embedded >processor development tools use it, and my SparcStation has a man page... Oh, you mean the format that AT&T has now dumped in favor of "Elf"? No, there is no standard. At best, COFF was once a "de facto" standard, there are fewer and fewer systems that use it. >So, it's gradually becoming *the* standard. Quite the opposite - "Elf" may be the AT&T standard of the future. Except that OSF is promoting something they call "Rose", based on Mach-o... -------------- Marc Sabatella (marc@hpmonk.fc.hp.com) Disclaimers: 2 + 2 = 3, for suitably small values of 2 Bill and Dave may not always agree with me
darrylo@hpnmdla.hp.com (Darryl Okahata) (04/24/91)
In comp.sys.hp, scottm@spiderman.spider.co.uk (Scott Mackie) writes: > The GNU stuff especially has been a pain (I even got assembler patches from > HP Support for Emacs 18.55) although I'm told recent releases have better > support for HPUX. This year, people within HP contributed a number of programs to Interex (an HP user's group), for inclusion on their contrib tape. Among these was a copy of a version of GNU Emacs used within HP. It's basically vanilla GNU Emacs with a *lot* of added features; I've included a list of features at the end of this message. If anyone is bringing up GNU Emacs for the first time on an HP workstation, I recommend getting a copy of the Interex contrib tape with this version of Emacs, as it is a complete package with full laserjet-ready documentation. Note, however, that this is a completely unsupported program. -- Darryl Okahata UUCP: {hplabs!, hpcea!, hpfcla!} hpnmd!darrylo Internet: darrylo%hpnmd@relay.hp.com DISCLAIMER: this message is the author's personal opinion and does not constitute the support, opinion or policy of Hewlett-Packard or of the little green men that have been following him all day. =============================================================================== Many of these features have already been made available to the public, often by people outside HP; however, this may be the only publicly available version where all of these features have been combined into a single monolithic version of GNU Emacs (we imagine that other, similar, versions are probably available, but we are not aware of them). Features: 1. This is a "complete" version of GNU Emacs, with extensive modifications, additions, and documentation. Complete high-quality manuals, ready for printing on an HP Laserjet or compatible, are included. Many separate, but highly useful, packages have been integrated into this version, which means that novice users do not have to ask around or hunt through the Internet to find useful utilities. You no longer need access to a "GNU Emacs GURU" to have a powerful version of Emacs. 2. Extensive HP keyboard support, both on terminals and in X11. If you want, you can use hierarchical softkey menus (they exist even if you run Emacs in native X11 mode). 3. A sample .emacs file is provided, with copious comments describing how to configure Emacs to your liking. If an error occurs while reading the .emacs file, a useful error message is displayed, and not just the cryptic "error in init" message. 4. Enhanced X11 support: You can specify a class and resource name for Emacs. Regions being "cut" by the mouse are highlighted. You can programmatically read and set the name of the Emacs' icon. You can programmatically read and set the window name. You can programmatically move, resize, iconify, and uniconify the window. Emacs can be started in iconic form. Pop-up X Menus are enabled. Full interrupt (signal) driven I/O. 5. Has hooks to "ispell", the interactive spelling checker (the sources to this program is included, if you do not already have ispell -- see the file "HP-new/README.HP" for details). You can do interactive spell-checking (where you are given a list of possible correct spellings of a word) from within Emacs. 6. Gnuserv is integrated within this version. Gnuserv, along with the accompanying programs "gnuclient" and "gnudoit", is like a more powerful version of "emacsclient". By running "gnuclient" on machine "A", you can cause a copy of emacs running on machine "B" to edit a file on machine "A". You can also issue Emacs-lisp commands remotely (this feature is particularly useful with X11 window-manager pop-up menus -- you can cause Emacs to do things by selecting a window-manager menu pick). 7. Complete on-line documentation. The documentation have also been processed into PCL files, for direct feeding to an HP-Laserjet or compatible to create high-quality printed manuals (this assumes that the laser printer has enough toner and paper to print hundreds of pages). 8. The following non-standard functions are available. Note that those functions that start with `x-HP-' are functions that will no longer exist when GNU Emacs V19 is released -- they start with `x-HP-' to make it easier to identify functions that may have to be emulated when V19 appears (these functions are "stopgap" functions, functions whose functionality need to be in Emacs, but do not yet officially exist). This is not a complete list, but it is in alphabetical order: add-load-hook Add a load hook for FILE which is FORM. FORM is added to load-hooks after all load-hooks for FILE which are presently there. align-equals Make the first assignment operator on each line line up vertically. ange-crypt-mode Toggle ange-crypt mode. With arg, turn ange-crypt mode on iff arg is positive. In ange-crypt mode, when the buffer is saved, it is encrypted first, and when restored, it is decrypted first. ange-crypt-set-key Prompt the user for the key to encrypt/decrypt the current buffer. Echos a . for each character typed. End with <cr>, <lf>, or <esc>. DEL or backspace rubs out. ange-find-tag Similar to find-tag except that the tag table associated with the current buffer is used. See the documentation for find-tag and ange-tags-alist for more details. ange-find-tag-other-window Similar to find-tag-other-window except that the tag table associated with the current buffer is used. ange-tags-loop-continue Similar to tags-loop-continue except that it remembers the name of the tag file used on the last tag operation. See the documentaton for tags-loop- continue for more details. block-comment-edit Copies the comment surrounding point into a temporary buffer for editing in a more suitable major mode. See the documentation for this function for more details. browse-kill-ring Browses the kill ring in another buffer. Use C-y to yank most recent kill ring entry (number 1.) into the buffer which was current when browse-kill-ring was invoked. Use a numeric argument (M-# C-y) to yank the corresponding entry. If you move point into the browse-kill-ring buffer, the keys C-y, y, Y, or SPC will yank the entry on the current line into the previous buffer. Note that browse-kill-ring should always be called before its contents is used; its buffer is not automatically updated when kills and yanks are done. This is normally not a problem, since yanks will cause the buffer to be deleted automatically. byte-compile-file Compile a file of Lisp code named FILENAME into a file of byte code. The output file's name is made by appending "c" to the end of FILENAME. calc This is Dave Gillespie's GNU Emacs Calc program, a symbolic RPN calculator for GNU Emacs. It is similar to the HP-28S/HP-48SX calculators, and can be thought of as a "poor-man's symbolic mathematics program". c++-mode Major mode for editing C++ code. Very much like editing C code. Expression and list commands understand all C++ brackets. Turning on C++ mode calls the value of the variable c++-mode-hook with no args, if that value is non-nil. c-comment-edit Edit multi-line C comments. The comment may be indented or flush with the left margin. Similar to block-comment-edit. calendar Display a three-month calendar in another window. The three months appear side by side, with the current month in the middle surrounded by the previous and next months. The cursor is put on today's date. This function is suitable for execution in a .emacs file; appropriate setting of the variable `view-diary-entries-initially' will cause the diary entries for the current date to be displayed in another window. The value of the variable `number- of-diary-entries' controls the number of days of diary entries displayed upon initial display of the calendar. An optional prefix argument ARG causes the calendar displayed to be ARG months in the future if ARG is positive or in the past if ARG is negative; in this case the cursor goes on the first day of the month. Once in the calendar window, future or past months can be moved into view. Arbitrary months can be displayed, or the calendar can be scrolled forward or backward. The cursor can be moved forward or backward by one day, one week, one month, or one year. All of these commands take prefix arguments which, when negative, cause movement in the opposite direction. For convenience, the digit keys and the minus sign are automatically prefixes. The window is replotted as necessary to display the desired date. Diary entries can be marked on the calendar or displayed in another window. Use M-x describe-mode for details of the key bindings in the calendar window. The Gregorian calendar is assumed. After preparing the calendar window initially, the hooks given by the variable `initial-calendar- window-hook' are run. The hooks given by the variable `today-visible-calendar-hook' are run everytime the calendar window gets shifted, if the current date is visible in the window. If it is not visible, the hooks given by the variable `today-invisible-calendar-hook' are run. Thus, for example, setting `today-visible-calendar-hook' to 'calendar-star-date will cause today's date to be replaced by asterisks to highlight it whenever it is in the window. ci-buffer Performs an RCS check-in of the file associated with the current buffer. Pops up a buffer for creation of a log message then does a "ci -u file", a "ci -l file", or a "ci file". cindent Reformat the specified BUFFER using the Unix indent(1) program. Selects the specified buffer, and saves it to disk, displays new version. M-x revert-buffer and M-x undo work as expected. User may opt not to save the newly indented buffer. co-buffer Performs an RCS check-out of the file associated with the current buffer. If ARGS is not specified, the file is checked out locked. If ARGS is a string, it is used for the switches to pass to ci. Otherwise (the function was called interactively with a prefix argument) the switches to pass to ci are prompted for in the minibuffer. complete-or-expand-token Complete token before point using COMPLETION-LIST, or if complete expand to show generic declaration, and display synopsis in the echo area complete-token Complete token before point using COMPLETION-LIST. Inserts as many characters as possible, and then if multiple completions, display them in a *Completions* buffer. completing-find-tag A "better> interface to find-tag. completing-find-tag-other-window A "better" interface to find-tag, uses other window. compress-mode Automatically uncompress a compressed file when editing it. Unfortunately, there is no automatic re-compression done when the file is saved. copy-line Duplicate the current line. If repeat is non-nil, it indicates how many copys to make. copy-region Duplicate the region indicated by start and stop. If repeat is non-nil, it indicates how many copys to make. The duplicates are inserted after stop. cpp-buffer Run the current buffer through the C preprocessor. cpp-buffer-for-display Run the current buffer through the C preprocessor and make it suitable for viewing. All occurrences of successive blank lines are each compressed into one blank line. crontab Function to allow the easy editing of crontab files. date-code Update datecode and filecode of the current buffer. This function is usually bound to write-file-hooks. define-cxxr-function Given string XX, create standard function cXXr. delatex Remove LaTeX commands from the current buffer. describe-region Gives a message briefly describing the text at the beginning and end of the region. detex TeX commands from the current buffer. diff Take the diff(1) of a BUFFER and its oldest backup file. With prefix arg, (or optional flag if noninteractive) does a normal non-context diff. display-file-info Run `ls -l <filename>' whenever a file is visited, placing the output in the message line. display-line-numbers Temporarily display the line numbers of lines in a buffer. edebug-defun Evaluate the defun that point is in or before, but set it up for edebug. Print its name in the minibuffer. edit-c-prologue Edit the prologue for the C function enclosing or after point, or the file prologue if file prologue is specified. emerge-buffers Run Emerge on two buffers. emerge-files Run Emerge on two files. emerge-files-with-ancestor Run Emerge on two files, giving another file as the ancestor. etags2ctags-buffer Convert a TAGS buffer to ctags format. etags2ctags-file Convert a TAGS file to ctags format. execute-unix-command Execute a unix command, wait for it for finish, and return the exit status. This function creates a process, using BUFFER for output, and waits for the process to exit. The name of the program to run is given by PROGRAM, and any remaining arguments, ARGS, are passed as command line arguments to the program. All arguments are strings. Note that there are no provisions for giving input to the program. If you must give input to the program, create a file containing the input for the process and pass the name of this file to a shell script that starts up the program with standard input redirected to the given file. If BUFFER is nil, any output from the program will be throw away. If a ^G is pressed while the program is running, the program will be killed via a SIGINT, and this function will return nil. All other keypresses will be flushed. Upon normal completion of the program, the numeric exit status of the program will be returned. Note that this function returns either nil or a numeric value. expand-token Expand COMPLETION using COMPLETION-LIST to (momentarily) show generic declaration, and display synopsis in echo area. COMPLETION must be complete, and COMPLETION-LIST is a list of elements of the form: ("completion" . ("declaration" . "synopsis")) fast-apropos Show all symbols whose names contain matches for REGEXP. If optional argument DO-ALL is non-nil, does more (time-consuming) work such as showing key bindings. Returns list of symbols and documentation found. find-alter-file Like find-alternate-file, but presents current filename for editing. find-directory Search for DIR in the directories listed in cd-path. If found, switch to a buffer in dired-mode for that directory. find-directory-noread Search for DIR in the directories listed in cd-path. If found, switch to a buffer in dired-mode for that directory. The directory list is not read in by default; you can read it with the normal dired-mode commands. find-this-error Find the error that point is on and position to that skipping all in between. If error is before the previous error found, then reparse the errors and try again. finger Show who is on the system. free-space-checker Start a background process to display "nn% free" in the mode line. The string to be inserted into the mode-line is produced by free-space-checker-script. getris Clone of a famous Russian game program. gomoku Start a Gomoku game between you and Emacs. If a game is in progress, this command allow you to resume it. If optional arguments N and M are given, an N by M board is used. You and Emacs play in turn by marking a free square. You mark it with X and Emacs marks it with O. The winner is the first to get five contiguous marks horizontally, vertically or in diagonal. You play by moving the cursor over the square you choose and hitting RET, x, .. or whatever has been set locally. Use C-h m for more info. grep Run grep, with user-specified args, and collect output in a buffer. While grep runs asynchronously, you can use the C-x ` command to find the text that grep hits refer to. hexl-find-file A major mode for editing binary files in hex dump format. This function automatically converts the buffer to the hexl format using the function `hexlify-buffer'. Each line in the buffer has an `address' (in hexadecimal) representing the offset into the file that the characters on this line are at, 16 characters from the file in displayed as their ascii values in hexadecimal grouped every 16 bits, and (the same) 16 characters displayed as ascii characters. If any of the characters (displayed as ascii characters) are unprintable (control or meta characters) they will be replaced as periods. hide-ifdef-mode Toggle hide-ifdef-mode. Thus this is a minor mode, albeit a large one. With arg, turn hide-ifdef-mode on iff arg is positive. In hide-ifdef-mode, code within ifdefs that the C preprocessor would eliminate may be hidden from view. HP-fork-and-exit-emacs Fork the current emacs process and exit. (In case you forgot the ampersand when starting emacs.) increment-number-on-line increment-numbers-in-column insert-file-name-tail Prompts for a file or directory name and inserts that name after point. Does not include unnecessary directory path prefixes. The name may be non- existent. Useful in Shell mode. insert-prototypes Generate and insert ANSI C prototype definitions for the current buffer. The program specified by `proto-program' is used to do the actual prototype generation. The options specified by `proto- options' are also passed to the `proto-program'. ispell-buffer Check the current buffer for spelling errors interactively. The variable which should be buffer or mode specific ispell-filter-hook is called to filter out text processing commands. ispell-region Check a region for spelling errors interactively. The variable which should be buffer or mode specific ispell-filter-hook is called to filter out text processing commands. ispell-word Check spelling of word at or before dot. If word not found in dictionary, display possible corrections in a window and let user select. jbm-shell Run an inferior shell, with I/O through buffer *shell*. If buffer exists but shell process is not running, make new shell. Shell program used comes from variable SHELL, or defaults to /bin/ksh. The buffer is put in jbm-shell-mode, which gives you a number of facilities for interacting with the shell. A negative prefix argument causes further shell windows (*shell<2>*, *shell<3>*, etc.) to be created. A positive argument causes the shell window with that number to be selected, instead of *shell* (which is number 1). Shells with numbers > 1 must be created explicitly; shell 1 is created automatically if it does not already exist. The current directory of the shell is the same as the default-directory from which the command was given. kill-region-wimpy Like kill-region, this deletes the text between BEG and END, and stuffs it in the kill ring. (Interactively, uses the region.) However, it asks you if you really want to delete a large chunk of text. lint Run lint(1) on specified buffer and collect output in a buffer. While lint runs asynchronously, you can use the C-x ` command to find the text that lint gripes refer to. lint-goto-problem Show the file and line responsible for the lint output under the cursor. lispref-search Symbol is regexps. Search lispref manual, display text in other-window. make-c-prologue Make up nice software development-type prologue for the C routine enclosing or after point. If ARG is non-nil, then ask for section text, each in its own buffer (see doc for c-prologue-section-alist). Leaves point at top of C defun. Leaves mark at start of prologue. math Run Mathematica, input and output via buffer *math*. math-mode Major mode for interacting with Mathematica and editing .m files. start-math Starts a Mathematica process in the current buffer. netunam (NOTE: OBSOLETE AS OF HP-UX 8.0) Open a network connection to another machine using RFA. PATH is the full name of network special file for the machine, LOGIN is the name to login as. If netunam is called programatically, it will only prompt for a password if LOGIN contains a ':' as its last character (just like netunam(1)). If netunam is called interactively, it will always prompt for a separate login and password. If the login contains no password, the user may hit <return> right after the password prompt. To close a network connection, hit <return> after being prompted for a login. Netunam will always give a message on the status of the connection when it finishes, and will also signal an error and return nil if the connection cannot be made. Users wanting a "quiet" netunam to call programatically should call the function (sysnetunam PATH LOGIN) directly, as that function will only return t/nil to give the status of the netunam system call. The variable netunam-network-file-path may be changed (with setq) to match the configuration of your machine's network special file directory. Every successful netunam will automatically update the variable netunam-host-alist in order to allow other programs to determine which hosts/logins have been netunamed. This variable should be read only. Do M-X describe-variable netunam-host-alist for more information. new-find-file Like find-file, but does a server-netunam if necessary. next-buffer Switch editing to the next buffer. A list (in regexp form) of buffers to skip over are contained in the variable next-buffers-to-skip. occur-menu Show menu of lines containing match for REGEXP. Enters recursive edit on text showing an entry for each matching line. User can move to an entry and then exit with C-c C-e to move to the line in the original buffer described by the selected entry. Abort with C-] to avoid moving in the original buffer. If REGEXP is empty then THE EXACT SAME menu is presented again, with cursor initially at the next successive entry. This is useful for stepping through located lines rapidly in order. pascal-mode This is a mode intended to support program development in Pascal. Most control constructs and declarations of Pascal can be inserted by typing Control-C followed by a character mnemonic for the construct. Generally, the functions expect to be invoked right after typing M-x pascal-newline, except for array, record, and proc/func body which start at the end of a line. Elements of the construct to insert are prompted for. Optional elements have prompts in square brackets[]. pass-to-sh Pass the line that point is on to shell. ps-mode A major-mode for sending signals to processes. In ps-mode, you indicate signals to send to UNIX processes. Signals are marked on the left hand side of the display using an abbreviated name. query-exit-from-emacs A function to bind to C-x C-c to query the user if he really wants to exit Emacs. rcs-ci Check in a file to rcs. rcs-co Check out a file from rcs. rcs-show-diff Pops up a buffer showing the difference between FILENAME and REVISON. If FILENAME is not specified, the file for the current buffer is used. If REVISION is not specified, the revision that the user has locked is used. Failing that, the latest revision is used. re-search-backward-command Search backward from point for match for regular expression REGEXP. Set point to the beginning of the match, and return t. Optional second argument REPEAT (prefix argument interactively) is the number of times to repeat the search. re-search-forward-command Search forward from point for regular expression REGEXP. Set point to the end of the occurrence found, and return t. Optional second argument REPEAT (prefix argument interactively) is the number of times to repeat the search. read-load-hook Read the next input form and add it as a load hook for FILE. If a call to read-load-hook is in a file that is being loaded, the next lisp expression after it will be read as the load hook. read-msdos Function to strip <cr><lf> sequence out of a file when read. results-of-command Execute the given COMMAND with &rest ARGS, and return the resulting output as a string. rmb-insert-line-numbers Add line numbers to an rocky mountain basic file save-help Sometimes you want to save the *Help* buffer for later examination, e.g., when you do an apropos. save-help will rename the *Help* buffer *Help<1>*, *Help<2>*, etc., so the information won't get clobbered by further help requests. search-backward-command Search backward from point for STRING. Set point to the beginning of the occurrence found, and return t. Optional second argument REPEAT (prefix argument interactively) is the number of times to repeat the search. search-forward-command Search forward from point for STRING. Set point to the end of the occurrence found, and return t. Optional second argument REPEAT (prefix argument interactively) is the number of times to repeat the search. search-manpath-for-cat-directories Return a list of all 'cat' directories in the current path specified by the 'MANPATH' environment variable. see-chars Displays characters typed, terminated by a 3-second timeout. server-start Allow this Emacs process to be a server for client processes (this is for gnuserv(1L) -- see the corresponding man page). This starts a server communications subprocess through which client "editors" can send editing commands to this Emacs job. set-c-style Set up the c-mode style variables from STYLE if it is given, or default-c-style otherwise. It makes the c indentation style variables buffer local. show-time Display the current time repeatedly in large characters. Optional SLEEPTIME gives number of seconds between updates (default is 5 seconds). Any input pending terminates the loop, and restores the original screen configuration (if the input is a space character, it is discarded). If SLEEPTIME is negative, append a cute quote to the time. sql-mode This mode provides both for editing of sql files and for the transmittal to another process of sql statements. This later ability provides a significant functionality improvement over either 'isql' (from Sybase) or 'sqlplus' (from Oracle). strip-line-numbers Strip Basic line numbers from file sum-column Return the sum of the integers in the rectangle delimited by START and END. Interactively, it prints the sum as well, and uses the region. sum-region Adds up the numbers in the region START to END. Primitive as yet. If called interactively, uses the region and prints the sum in a message. Ignores things that aren't numbers or signs, so $1 + $4 will sum to 5, and 1 -4 will sum to -3. Of course, it's really dumb, and things like 2*3 - 5 will sum to 2+3+5 = 10 -- as will 2 + 3.5. Improvements welcomed. super-apropos Show symbols whose names/documentation contain matches for REGEXP. If optional argument DO-ALL is non-nil, does more (time-consuming) work such as showing key bindings and documentation that is not stored in the documentation file. Returns list of symbols and documentation found. symbol-around-point Return the symbol around the point as a string. template-mode Toggle template-mode, a minor mode for manipulation of text via templates. Calls 'template-mode-hook' if it is defined. tinymud Connect to a tinymud host and play tinymud. Note that the variables `tinymud-server' and `tinymud- port' may have to be set before this function can be executed. unix-apropos Display apropos for TOPIC unix-apropos-get-man Get the manual entry for the current line visit-vi-tags-table Tell tags commands to use ctags tag table file FILE. FILE should be the name of a file created with the `ctags' program. A directory name is ok too; it means file "tags" in that directory. This function should be used in conjunction with `find-vi-tag' to look for Vi tags. vkill Mode for displaying all UNIX processes owned by the current user (all the processes on the system if invoked by the superuser) and allowing the user to mark processes to be sent a certain signal. Processes are marked by moving the cursor to the line displaying information about the victim process and typing `m' to mark the process. word-around-point Return the word around the point as a string. write-msdos Function to change <lf> to <cr><lf> for MS-DOS files. x-HP-get-icon-name Get the name of the Emacs icon. If an X error occurred while trying to get the name, nil is returned. x-HP-get-window-attributes Get the window attributes of the Emacs X-window. x-HP-get-window-name Get the name of the Emacs X window. x-HP-get-window-state Get the state of the Emacs X-Window. x-HP-iconify Iconify the X window. x-HP-move-window Move the Emacs X window to a specified location (X,Y). x-HP-remap-window Map/un-iconify the Emacs X-window. This function simply calls x-remap-window and is provided only for consistency with the `x-HP-' prefix naming convention. All future functions should use `x-HP- remap-window' instead of `x-remap-window' to insure ease-of-porting if and when GNU Emacs V19 is released. x-HP-resize-window Resize the Emacs X window to the size given by WIDTH and HEIGHT. x-HP-set-icon-name Set the name of the Emacs X icon to NAME. x-HP-set-window-name Set the name of the Emacs X window to be NAME. 9. The following are some new variables: x-HP-font-height The height of the current font in pixels. x-HP-font-width The width of the current font in pixels. 10. You can now netunam from within Emacs. Just issue M-x netunam. (NOTE: THIS WILL BE OBSOLETE WITH HP-UX 8.0)
gerwitz@hpcore.Kodak.Com (Paul Gerwitz) (04/24/91)
In article <430045@hpnmdla.hp.com>, darrylo@hpnmdla.hp.com (Darryl Okahata) writes: |> In comp.sys.hp, scottm@spiderman.spider.co.uk (Scott Mackie) writes: |> |> > The GNU stuff especially has been a pain (I even got assembler patches from |> > HP Support for Emacs 18.55) although I'm told recent releases have better |> > support for HPUX. |> |> This year, people within HP contributed a number of programs to |> Interex (an HP user's group), for inclusion on their contrib tape. |> Among these was a copy of a version of GNU Emacs used within HP. It's |> basically vanilla GNU Emacs with a *lot* of added features; I've |> included a list of features at the end of this message. |> |> If anyone is bringing up GNU Emacs for the first time on an HP |> workstation, I recommend getting a copy of the Interex contrib tape with |> this version of Emacs, as it is a complete package with full |> laserjet-ready documentation. Note, however, that this is a completely |> unsupported program. |> |> -- Darryl Okahata |> UUCP: {hplabs!, hpcea!, hpfcla!} hpnmd!darrylo |> Internet: darrylo%hpnmd@relay.hp.com |> This version of EMACS is part of the 3110 release of the Interex Contributed Software Library CSL/HPUX. The CSL is one of the benefits of membership in Interex, the International Association of HP Computer Users. {I should know, I'm the CSL/HPUX committee chair} Contact Interex at (408) 738-4848, ask to talk to Steve Souza. -- +----------------------------------------------------------------------------+ | Paul F Gerwitz WA2WPI | SMTP: gerwitz@kodak.com | | Eastman Kodak Co | UUCP: ..uunet!atexnet!kodak!eastman!gerwitz | +----------------------------------------------------------------------------+ -- +----------------------------------------------------------------------------+ | Paul F Gerwitz WA2WPI | SMTP: gerwitz@kodak.com | | Eastman Kodak Co | UUCP: ..uunet!atexnet!kodak!eastman!gerwitz | +----------------------------------------------------------------------------+
mat@cdbhp1.UUCP (W Mat Waites) (04/25/91)
First - scottm@spider.co.uk writes: > The GNU stuff especially has been a pain Then - lynch@theory.tn.cornell.edu (Tim Lynch) writes: > I don't understand. We're running 18.55 on an 800 series machine and built > it with no problem. Then - scottm@spider.co.uk writes: > Well, a quick look at uname says > > $ uname -a > HP-UX spiderma 7.0 B 9000/375 spiderma AHA - Maybe it would be a good idea to include in all "problem report" postings which series of machines (300-400 or 700-800) you are having trouble with. Especially if there is assembly involved. Oh, well. I guess we won't have to worry about the 300's much longer, anyway. Mat -- W Mat Waites mat@cdbhp1.UUCP | Unlike most of you, I am not a nut. {gatech,emory}!cdbhp1!mat | -H. Simpson
jim@tiamat.fsc.com ( IT Manager) (04/25/91)
In article <101950207@hpcvlx.cv.hp.com>, jbb@hpcvlx.cv.hp.com (Jim B. Byers) writes: > > R4, Motif 1.1 libraries and servers are standard with HP-UX 8.0. Also > scalable typefacces in X11 are shipped as standard in 8.0. Motif > 1.1 libraries can be purchased for 7.0 by ordering product B1177A. > I think I'm getting confused by the fact that the same OS version number is being used for HP's workstations (300/700) and multi-user machines (800). Does the statement "R4, Motif 1.1, etc." above mean that those products are in HP-UX 8.0, or that the 8.0 version of HP's X windows contains those products? ------------- James B. O'Connor jim@tiamat.fsc.com Ahlstrom Filtration, Inc. 615/821-4022 x. 651
ian@rathe.cs.umn.edu (Ian Hogg) (04/25/91)
In article <430045@hpnmdla.hp.com> darrylo@hpnmdla.hp.com (Darryl Okahata) writes: >In comp.sys.hp, scottm@spiderman.spider.co.uk (Scott Mackie) writes: > >> The GNU stuff especially has been a pain (I even got assembler patches from >> HP Support for Emacs 18.55) although I'm told recent releases have better >> support for HPUX. > > This year, people within HP contributed a number of programs to >Interex (an HP user's group), for inclusion on their contrib tape. >Among these was a copy of a version of GNU Emacs used within HP. It's >basically vanilla GNU Emacs with a *lot* of added features; I've >included a list of features at the end of this message. > > If anyone is bringing up GNU Emacs for the first time on an HP >workstation, I recommend getting a copy of the Interex contrib tape with >this version of Emacs, as it is a complete package with full >laserjet-ready documentation. Note, however, that this is a completely >unsupported program. > When I first got Emacs running on HP-UX (my first experience whatsoever with Emacs) I followed the instructions instructions that came with it and had to think hard about certain defines and such. When we got Softbench it included detailed instructions on how to build Emacs for HPUX. Maybe someone can post that file or place it somewhere for ftp. If you are using Emacs & X, please take a look at Epoch. I prefer it to the vanilla X support. > -- Darryl Okahata > UUCP: {hplabs!, hpcea!, hpfcla!} hpnmd!darrylo > Internet: darrylo%hpnmd@relay.hp.com > >DISCLAIMER: this message is the author's personal opinion and does not >constitute the support, opinion or policy of Hewlett-Packard or of the >little green men that have been following him all day. > -- Ian Hogg email: rathe!ian@cs.umn.edu ...!umn-cs!rathe!ian Rathe, Inc ianhogg@cs.umn.edu 366 Jackson Street phone: (612) 225-1401
mjs@hpfcso.FC.HP.COM (Marc Sabatella) (04/25/91)
>The patches that I applied were to some 68XXX assembler in the src dir... I think >it was alloca.s. Obviously that doesn't apply to the 800's.... > >I can dig out the notes I got from HP Support if anyone's interested... So it's not an "assembler patch" - that would mean there was a problem with the assembler. It's a "source patch", in this case meaning the problem is with the software. alloca() and crt0 are not things that "portable" programs should be messing with. Doing so merely guarantees your software will not be portable. Software that adheres to standards (POSIX especially) should port with no problem. -------------- Marc Sabatella (marc@hpmonk.fc.hp.com) Disclaimers: 2 + 2 = 3, for suitably small values of 2 Bill and Dave may not always agree with me
harry@hpcvlx.cv.hp.com (Harry Phinney) (04/29/91)
James B. O'Connor writes: > Jim B. Byers writes: >> >> R4, Motif 1.1 libraries and servers are standard with HP-UX 8.0. Also >> scalable typefacces in X11 are shipped as standard in 8.0. Motif >> 1.1 libraries can be purchased for 7.0 by ordering product B1177A. > Does the statement "R4, Motif 1.1, etc." above mean that those products > are in HP-UX 8.0, or that the 8.0 version of HP's X windows contains HP-UX 8.0 for the series 300, 700, and 800 machines contains R4 based libraries and X servers, and Motif 1.1. Harry Phinney harry@hp-pcd.cv.hp.com
scottm@spiderman.spider.co.uk (Scott Mackie) (04/29/91)
In article <7370381@hpfcso.FC.HP.COM>, mjs@hpfcso.FC.HP.COM (Marc Sabatella) writes: |> >The patches that I applied were to some 68XXX assembler in the src dir... I think |> >it was alloca.s. Obviously that doesn't apply to the 800's.... |> > |> >I can dig out the notes I got from HP Support if anyone's interested... |> |> So it's not an "assembler patch" - that would mean there was a problem with the |> assembler. It's a "source patch", in this case meaning the problem is with |> the software. Uhh, yeah.... My fault for sloppy typing... Mind you, anyone who thinks I'd apply a patch to as(1) instead of fixing the source is being a tad silly ;-) |> alloca() and crt0 are not things that "portable" programs should be messing |> with. Doing so merely guarantees your software will not be portable. Software |> that adheres to standards (POSIX especially) should port with no problem. The patches, as I remember (I ask for artistic licence here as it's been about 9 months !), were to the pieces of code *inside* the HP assembler switches in alloca.s. I don't see that affecting future ports to other manufacturers platforms. Scott..... p.s. POSIX and Portability? Shurely Shome Mistake? ;-) -- Spider Systems Limited Net: scottm@spider.co.uk Spider Park, Stanwell Street, Edinburgh YellNet: +44 31 554 9424 #include <disclaimer.h> "Rockin' Good, Peanut!"
franks@hpuamsa.neth.hp.com (Frank Slootweg CRC) (05/03/91)
Scott Mackie writes : >|> alloca() and crt0 are not things that "portable" programs should be messing >|> with. Doing so merely guarantees your software will not be portable. Software >|> that adheres to standards (POSIX especially) should port with no problem. > >The patches, as I remember (I ask for artistic licence here as it's been about >9 months !), were to the pieces of code *inside* the HP assembler switches >in alloca.s. I don't see that affecting future ports to other manufacturers >platforms. No offense, but I think you still don't get it. I think what Marc (Sabatella) meant is that portable programs should not *use/depend_on* things like alloca() and crt0, so a patch for something you are not supposed to use is a moot point. Agreed? > Uhh, yeah.... My fault for sloppy typing... Mind you, anyone who thinks I'd apply > a patch to as(1) instead of fixing the source is being a tad silly ;-) And so is anyone using/depending_on alloca() and/or crt.0 in a "portable" program. Frank Slootweg, working but not speaking for HP.
jason@hpcndjdz.CND.HP.COM (Jason Zions) (05/13/91)
>> > Does HPUX support long filenames, symlinks, job control, sockets, NFS? >> >> Yes, since HP-UX Releases 6.5 [s300] and 2.0 [s800], respectively. > >Long file name support is there only if you've converted your file systems. >If you're writing commercial software you'll have to work with the 14 character >limit. Not all third parties have adapted to the 6.0 changes much less 6.5 >and 7.0. There are issues regarding building your own libraries too. The 14 >character limit crops up again. If you're writing commercial software you can either (a) restrict yourself to just 14-byte characters, or (b) require your customers to use LFN (long file name) filesystems. Option (a) may make life easier if you intend to port to stock SVR[23]; option (b) may cause customerw stuck with SFN to curse you. (Some applications explicitly require short-names and break if run on LFN filesystems; this is the reason for continued SFN support.) Of course, I'd personally recommend choice (c), which is to make use of the pathconf() call to ascertain the actual limit on filenames in a particular directory and adjust your operation accordingly; a little more work in the coding, but a maximally-portable solution that will please more people than either (a) or (b). >> > Does uucp work reliably at 19200 bps? Is it a solid version of uucp? > >Yes, HP runs HDB, and is solid an 19200. If you're using a series 300 machine >the internal port is a no-no (it is serviced by main CPU... and thus slows >things down). You'll want a 4 port mux... of which only one port can be >used with a modem. Argh. As has been pointed out, there is an 8-port mux board which supports modem connects on all 8 ports. Also, the built-in serial port on Series 300 machines which predate the 370 have a 1-byte FIFO which prevents reliable operation at 19200. The serial connection on the 375 (and I think 345 as well) have a much larger FIFO and have no troubles at full speed. The 98625 serial card should have no troubles. >> > Does HPUX provide lpr or does it only provide lp? >> >The printing subsystem is based on SysV. It uses a printer script and there >is no printcap (ala BSD printer subsystem). Having used both I prefer the >System V... I found it easier to bend to my will. Interesting observation! :-) Adding postscript support is pretty easy, and the lp subsystem will interoperate with lpr over the network. >> > How good (fast, robust) is HP's implementation of X? Does it include >> > all of X11R4? Is it reasonably close to X11R4? Does the MIT X11R4 >> > distribution build on a 9000/730? Is the R5 distribution likely to? >> > Do HP document their graphics hardware enough to port a window system? The 700 will run X11R4. The MIT distribution should build on the 730 with no troubles. The R5 distribution almost certainly will; there's a lot of HP gear involved at MIT. The 700 is the current X performance champion, and the HP X implementation has always had a reputation for solidity. The only remaining issue is whether HP will ship libXaw.a (i.e. the Athena widgets) or require folks to acquire the Athena goo and build it themselves. The issue HP has to wrestle with is whether shipping the Athena widgets as part of the product forces us to support them and bugfix them to the level of solidity represented by the HP-written parts of the product; we're not sure we can spare the resources to do that. Still under discussion. >> > How does HP's 19'' monitor compare to the Trinitron that Sun ships? >> > How about the keyboard and mouse? How noisy/hot would a system with >> > two internal 400Mb disks be? Would it be reasonably pleasant in an >> > office environment? Does it connect directly to thinwire ethernet? Is >> > HP hardware generally reliable? > >HP's monitors are superior in resolution and clarity to what Sun ships. I >consider Sun's monitors to be PC class in resolution, clarity, and quality. >HP's monitors are professional. Higher resolution (if you opt for it), always >clearer, and very reliable (come on now, we're talking HP here, hardware >quality is a specailty of this company). For all display models of the 700 other than the PVRX, the monitor is a 72MHz Hitachi; the PVRX family uses a 60MHz Sony Trinitron. Sonys weren't used for the other displays because of the lack of a 72MHz display, which is increasingly a mandatory requirement in Europe. Early shipments of the Hitachi suffered from some misconfiguration at the factory resulting in impaired display quality; the fixes are well understood and are covered under the 90-day warranty, if memory serves. Currently shipping Hitachis are properly adjusted. >Other things you should know. HP wrote their own networking years ago. It's >a little wierd. They're fixing it. 8.0 is the magic release again. They may >not actually be planing to call it 8.0... but it should give you the right >idea. Networking in the 8.0 releases of HP-UX (including the 700s) is based on the BSD4.3 source. In the past, the kernel portions of networking were 4.2 based with user-space commands somewhere in the continuum of patched-4.2 through patched-post-4.3. All commands are now 4.3 or better. >> *********************************************************************** >> DISCLAIMER: All of the information above is provided for informative >> purposes only, as my own opinion, and may be false. I am not authorized >> by HP to make binding statements. An HP sales representative will >> be able to give you such answers. Applies to me as well. Jason Zions
corrigan@weber.ucsd.edu (Michael J. Corrigan) (05/15/91)
In article <1730085@hpcndjdz.CND.HP.COM} jason@hpcndjdz.CND.HP.COM (Jason Zions) writes: .... } }Networking in the 8.0 releases of HP-UX (including the 700s) is based on the }BSD4.3 source. In the past, the kernel portions of networking were 4.2 based }with user-space commands somewhere in the continuum of patched-4.2 through }patched-post-4.3. All commands are now 4.3 or better. Is arp replacing probe ? (i.e. can bootpd be ported simply) } }}} *********************************************************************** }}} DISCLAIMER: All of the information above is provided for informative }}} purposes only, as my own opinion, and may be false. I am not authorized }}} by HP to make binding statements. An HP sales representative will }}} be able to give you such answers. } }Applies to me as well. } }Jason Zions
coolidge@speaker.sgi.com (Don Coolidge) (05/16/91)
In article <5351@network.ucsd.edu> corrigan@weber.ucsd.edu (Michael J. Corrigan) writes: >In article <1730085@hpcndjdz.CND.HP.COM} jason@hpcndjdz.CND.HP.COM (Jason Zions) writes: >.... >} >}Networking in the 8.0 releases of HP-UX (including the 700s) is based on the >}BSD4.3 source. In the past, the kernel portions of networking were 4.2 based >}with user-space commands somewhere in the continuum of patched-4.2 through >}patched-post-4.3. All commands are now 4.3 or better. > > Is arp replacing probe ? (i.e. can bootpd be ported simply) > >}Jason Zions (Hi, Jason!) Movie dialogue: "...and why did you come to Casablanca, Monsieur Rick?" "I came for the waters." "But there are no waters here!" "I was misinformed." ARP is not replacing Probe for the simple reason that ARP has always been there. ARP and Probe have shared an address-resolution cache since HP-UX 1.0 on the s800, and at least since 6.0 on the s300. ARPs and Probe VNAs have both been sent and received, so your question rests on a misunderstanding. (They coexisted to permit both Ethernet and IEEE 802.3 address resolution. In 8.0, parts of that relationship are redefined to focus primarily on ARP, but it still works concurrently with Probe.) If, however, you mean "...is the BSD arp command supported in 8.0?", the answer is probably "yes" - at least, it was included when I left HP. As Jason says, 8.0 networking is 4.3BSDish, and the arp command is part of that. It wasn't included in HP's 4.2ish (pre-8.0) port, though - and anyone who attempted to compile and use it would find that there was no such thing as "arptab" to examine and/or modify (and, no, I'm not about to give the pre-8.0 shared cache name. Get it from your HP rep if you want to play with it, but be aware that there are *no* guarantees you won't crash the system by doing so, since the only way to write to it is by writing directly to the kernel; there's no ioctl support for doing that with spl() protection...). - Don Coolidge coolidge@speaker.wpd.sgi.com
jwright@cfht.hawaii.edu (Jim Wright) (05/17/91)
jason@hpcndjdz.CND.HP.COM (Jason Zions) writes: >The 700 is the current X performance champion, and the >HP X implementation has always had a reputation for solidity. No question about the performance of the 700. But for the day that is was using a 720, mwm dumped core at least a dozen times. But it dumped core *fast*. >The only >remaining issue is whether HP will ship libXaw.a (i.e. the Athena widgets) >or require folks to acquire the Athena goo and build it themselves. Yes, please! Put it in /usr/unsupported if you have to, but include it. We know you use it internally, since you ship xload. -- Jim Wright jwright@cfht.hawaii.edu Canada-France-Hawaii Telescope Corp.
jason@hpcndjdz.CND.HP.COM (Jason Zions) (05/22/91)
As far as I know, bootp is shipped with HP-UX 8.0, so there's little need to port it. If you wanted to port it, it should port. The "arp" command is part of 8.0 as well. Hello to you too, Don Coolidge! Jazz #include <usual_disclaimer.h>
clarke@hpfcmgw.HP.COM (Clarke Echols) (06/03/91)
> Uh, doesn't DWB, besides n/troff, eql, tbl, also contain programs like > pic and those writing style checkers whose names I can't remember? > Anyway, none of this is in HP-UX either. Just neqn, tbl and nroff. DWB requires a separate license from AT&T. Since variants of that package is available from several third-party vendors, HP has chosen not to compete with them. DWB includes troff, eqn, pic, and tbl. HP-UX provides nroff, neqn, and tbl. We are currently using Elan's eroff package to produce the HP-UX reference with output to a LaserJet PostScript printer for post-8.0 editions with generally satisfactory results. The 8.0 edition of the "brick" as we call it was produced using an HP derivation of AT&T troff on an Autologic phototypesetter. WWB (Writers Work Bench) from AT&T is a separate package that also requires its own license. It contains numerous programs for evaluating grammar, spelling, sentence structure, and other similar tasks. We bought a copy several years ago directly from AT&T and it ran with no problems on a Series 500 system. However, it gets little use in our department today. The above statements are simply an explanation of how we do things in one department of HP. They are not intended as an endorsement nor a criticism of any given product. Clarke Echols HP-UX Learning Products
simon@liasun2.epfl.ch (Simon Leinen) (06/04/91)
In article <17780029@hpfcmgw.HP.COM> clarke@hpfcmgw.HP.COM (Clarke Echols) writes: DWB requires a separate license from AT&T. Since variants of that package is available from several third-party vendors, HP has chosen not to compete with them. DWB includes troff, eqn, pic, and tbl. HP-UX provides nroff, neqn, and tbl. All of these programs, along with drivers for PostScript(TM), TeX DVI format, X11 and ASCII/ISO Latin-1, are available in the copylefted (i.e. freely redistributable) Groff package from the Free Software Foundation. I use them to print all kinds of roff'ed documentation and I am very happy with them. You have to have a C++ compiler though (I used G++ 1.39.1, but the newer AT&T CC should work). You can get them via anonymous FTP from prep.ai.mit.edu, file /pub/gnu/groff-1.01.tar.Z (927440 bytes). -- Simon.