tim@ucf-cs.UUCP (Tim Curry) (07/08/85)
My configuration: 3B2/300 with System 5.0.5 bought Sept. 1984. Anyone outside of AT&T with a 3B2 system been upgraded to 5.2? How many 3B2 systems are there outside AT&T? I debated with myself whether to break this into several small postings or one large one. I've opted for one large article. I hope that someone with all the answers makes it through all the questions. Our software development will require many different changes depending on what a real System 5.2 looks like. My account executive is a friendly guy and attempts to be helpful but just can't get me the kind of answers I need to make programming decisions and purchase packages. I need to know what I will be seeing if I can ever get upgraded to 5.2. I am hoping that the majority of my bad reactions to system 5 vs. BSD are related to the fact that version 0.5 was a quick release to get the machines out the door and version 2 (2.1,2.2) will be a complete version of UNIX. 1) What is the current release of System 5? I see some postings in this news group with 5.2.2. Anyone answering subsequent questions please indicate what version they are refering to. 2) termcap vs. terminfo vs. curses. Our software uses screen manipulations throughout but system 5.0.5 doesn't have the -lcurses or -ltermcap I have written my own code with our own termcap-like file but would much prefer using the system standard. What will 5.2 have? What is the difference between termcap and terminfo? Will both stay around or is termcap likely to die. Does 4.3BSD stick with termcap or move to terminfo? Do the curses routines work with both termcap and terminfo? Are the curses routines compatable between 4.xBSD and S5? Does terminfo have graphic character information to allow real boxes to be drawn vs. the box(stdwin,'|','-') type. Does terminfo distinquish attribute modes by name rather than function. i.e. "Reverse video" rather than "stand-out mode" which might be "half intensity" for a different terminal. 3) nroff/troff and "man" command. I have none of these. I believe I heard that troff might be unbundled to either the documentor's workbench or writer's workbench but I would think that nroff and man should be standard on each system. If nroff is available, then are neqn, table, col etc. also available. If these aren't standard, what package do I have to buy to get them? 4) is lp (line printer spooler) an optional package or standard? Our software makes a distinction between a "display" to the terminal and a "report" to hardcopy. We popen (pipe open) our output for the report to lp but if this is not standard, then I'm not sure where to send it. Is there a "printcap" file that describes the command sequences for printers like in 4.2BSD? 5) virtual memory/record locking - I understand that 5.2 does finally incorporate virtual memory but is that part of the requirements for ALL versions of 5.2? e.g. When (or if) the XENIX version that is 5.2 compatable on the IBM PC is implemented, will it support virtual memory so that I can freely fork and execl etc. to my hearts content without worring about a 256k system running out of memory and hence having to try some overlay techniques or something?!? Likewise, will all record locking calls be supported identically on all 5.2 compatable versions? 6) is "make" standard? Does "make" still only come with the C compiler package? When distributing software (even binarys only) "make" is the usual way to do it. I suppose that shell scripts can be set up but still ... 7) What is the toolchest? I have been off of the net for 9 months and just caught the tail end of disscussions about the toolchest when I started reading USENET again. Where can I get details? 8) Is there an (MS|PC)-DOS C cross compiler available so that I can do my compile work on the 3B2 and down-load the executable to a PC machine. I would think that AT&T would have such a beast since the 6300 is a DOS machine. 9) Does anybody know if I can buy a CSH for the 3B2 from any source? I understand that the bourne shell under 5.2 has some job control finally (could somebody tell me how this interface works? Is it like the csh with ctl-z, fb, bg etc.?) and the shell functions are probably more powerfull than aliasing in the csh but I desperately miss the history, push/pop directorys, ~ expansion, and ctl-w word delete. I use these things EVERY SESSION that I'm on a BSD system. I simply don't understand why AT&T is hesitant to make the BSD software available at least as an optional package. I'm 99% sure that they use a lot of BSD software themselves!! I'd like a lot of programs that I miss. Since Berkeley is not trying to make a profit from BSD and the BSD enhancements make a considerably nicer user interface, why aren't these things available? Furthermore, I know there is a package that gives System 5.2 support under 4.2BSD but why not the reverse from AT&T? This would allow the best of both worlds to be available. 10) what is available? Perhaps the last statement was over harsh. Where can I get a list of what software is available from AT&T and its cost? The one price I got for the Documentor's Workbench was a Source price. Are binary prices available for the unbundled protions of UNIX? 11) What does it take to get a software package sanctioned by AT&T for System 5.2? Tim Curry USENET: decvax!ucf-cs!tim ARPANET: tim.ucf-cs@csnet-relay -- Tim Curry USENET: decvax!ucf-cs!tim ARPANET: tim.ucf-cs@csnet-relay
heiby@cuae2.UUCP (Ron Heiby) (07/11/85)
Here's the answer to as many of Tim's questions as I could come up with quickly. Some questions related to BSD features which I am not qualified to address. In article <2067@ucf-cs.UUCP> tim@ucf-cs.UUCP (Tim Curry) writes: >1) What is the current release of System 5? The currently available version for the 3B2/300 is: UNIX System V Release 2.0 3B2 Version 2 AT&T has announced a paging release for the 3B2 which will be available shortly and be (I believe) Release 2.1. >2) termcap vs. terminfo vs. curses. "You sure ask a lot of questions." Anyway, SVR2 has libcurses.a and uses the terminfo database. /etc/termcap is still delivered for software that has not been recompiled to use terminfo. Termcap is not likely to die completely as long as there are many systems that do not yet have terminfo. Although, new development should use terminfo, as it is technically superior. Termcap is one big file with all the terminal descriptions. Terminfo is a hierarchy of files, where each file describes one terminal, hence it's more efficient. Whether termcap or terminfo is used depends on the library code with which the application was linked. Most of the other questions can be answered by the terminfo manual. >3) nroff/troff and "man" command. The nroff and troff commands (and I think man) are in the Documenter's Workbench, which is a seperately priced package. Man files for all of the section 1-n stuff is not supplied, as AT&T feels that a machine of the class of a 3B2 does not generally have enough free disk space to devote to that kind of material and the information can be had from the paper copy, anyway. Documenter's Workbench comes with tbl, eqn, neqn, col, etc. >4) is lp (line printer spooler) an optional package or standard? Our software > makes a distinction between a "display" to the terminal and a "report" > to hardcopy. We popen (pipe open) our output for the report to lp > but if this is not standard, then I'm not sure where to send it. > Is there a "printcap" file that describes the command sequences for > printers like in 4.2BSD? The LP Spooler package is a seperately priced package. The command to queue something for printing is named "lp" and is found in /usr/bin. There is no "printcap" that I know of. Each printer has an interface file (which is generally a shell script) which handles set-up for each print job. >5) virtual memory/record locking "All" versions of UNIX have supported "virtual memory" (with exceptions like the iAPX-86 architecture, which doesn't do memory management). What is new is "Demand Paging" rather than "Swapping". System V with demand paging is currently available for the DEC VAX and AT&T 3B20 systems and has been announced for the 3B2 and 3B15, although it is not yet available there. It is extremely unlikely that any implementation of UNIX for AT&T PC 6300 machines (or compatibles) would include demand paging, as there is no hardware support for it in the iAPX-86 architecture. Those systems would continue to be swap based. AT&T has committed to supporting the /usr/group standard for file and record locking across its product line. The currently available 3B2 release supports the "Advisory" locking, which is what most applications really want. The paging release on the 3B2 supports the "Mandatory" locking, as well. >6) is "make" standard? I assume that you mean "bundled". No, I believe that "make" comes in the "Extended Software Generation" package. >7) What is the toolchest? The AT&T Software Toolchest is a system which provides electronic distribution of purchased software. Source code is licensed for a fee and delivered via uucp. The license is not for a single machine, but for every machine in (at least location, probably organization, can't remember). Binary resale licenses are available with no per-sale royalty. See your AT&T Account Executive for more information. >8) Is there an (MS|PC)-DOS C cross compiler available I know of now MS-DOS C cross compiler being sold. Sounds good though. >9) Does anybody know if I can buy a CSH for the 3B2 from any source? I know of no source for CSH. System V Release 2.0 introduced "Shell Layers", which many Berkelyites sneer at, but which I find to be quite useful. It is a facility for having multiple logical terminal sessions multiplexed over the same connection. See the User Manual for more info [shl(1)]. Regarding the csh capabilities that you want, you should look into the Korn Shell (ksh) which is available through the Toolchest. I use nothing else. My guess about csh not being available from AT&T is that it is rather a botch, changing quite a few things just for the sake of changing them and is thus quite incompatible with Bourne. Korn, being 99 and 44/100 % upward compatible with Bourne does not have this problem. As to any other BSD developments: They are all known of and looked at by AT&T developers. Some appear in System V, like "cat -v" and "ls -RadCxmnlogrtyucpFbqisf" and "mailx" (alias Mail). The thing to remember is that Berkeley is (supposed to be) in the education business. They do a good job by letting students experiment. AT&T is in the stable computing environment business. We do a good job by making darn sure that what we do doesn't break something (like a shell script or worse) and that we spend our efforts spending resources on the most important/needed enhancements first. >10) what is available? See your AT&T Account Executive. Binary packages are available for the 3B2. Prices are in the price book. >11) What does it take to get a software package sanctioned by AT&T for > System 5.2? By "sanctioned", do you mean in the catalog, AT&T co-labling, or what? For any of that, you need to talk with the AT&T Independent Software Vendor program people. Your AT&T Account Executive should be able to find the right contact. Have fun. -- Ron Heiby heiby@cuae2.UUCP (via ihnp4) AT&T-IS, /app/eng, Lisle, IL (312) 810-6109
guy@sun.uucp (Guy Harris) (07/12/85)
> 1) What is the current release of System 5? I believe it depends on what machine you're on. It's S5R2V2 for the VAX. > 2) termcap vs. terminfo vs. curses. What will 5.2 have? The VAX 5.2 has the "curses"/"terminfo" package. > What is the difference between termcap and terminfo? "terminfo" has a lot more per-terminal capabilities (i.e., control strings, characteristics, etc.) than "termcap". The database is in a "compiled" format for faster loading. > Will both stay around or is termcap likely to die. There's no technical reason for "termcap" to stick around, other than the cost of conversion. There is a "termcap"-emulator package in the "curses" library, so programs written for "termcap" should work (unless they do something *really* weird), and there is a monster "ex" script that does a lot of the work of converting "termcap" descriptions to "terminfo" descriptions. However, "terminfo" requires an S5R2 license, so... > Does 4.3BSD stick with termcap or move to terminfo? I don't think they move to "terminfo", unless Pavel Curtis' public-domain "curses" supports "terminfo" and they picked that up. > Do the curses routines work with both termcap and terminfo? The 4.2BSD curses could probably work with "terminfo" using the "termcap" emulator (but it may do "something really weird", so it may not). The S5R2 "curses" uses a lot of the added capabilities of "terminfo" so it won't work with "termcap". > Are the curses routines compatable between 4.xBSD and S5? Pretty much. I've recompiled a number of "curses"-using programs written for "old curses" with "new curses" and they work. A number of 4.2BSD programs already have hooks in them for "new curses" (like "mille" and, I think, "sysline"). > Does terminfo have graphic character information to allow real boxes > to be drawn vs. the box(stdwin,'|','-') type. Not to my knowledge. Solving that problem in general is difficult if not impossible. Not all terminals have line-drawing characters, and even those that do don't have compatible characters in the same position in the alternate character set. > Does terminfo distinquish attribute modes by name rather than function. > i.e. "Reverse video" rather than "stand-out mode" which might be "half > intensity" for a different terminal. "terminfo" has "enter_standout_mode" and "exit_standout_mode" capabilities (which are, presumably, terminal-dependent in what they actually do, just as in "termcap); it also has "enter_blink_mode", "enter_bold_mode", "enter_dim_mode", etc. capabilities and their equivalent "exit" capabilities. It also has a "set_attributes" capability which directly sets video attributes. This capability is, in effect, the ANSI X3.64 Set Graphic Rendition sequence (the "terminfo" capabilities mirror X3.64 control sequences to a large degree). > 3) nroff/troff and "man" command. I have none of these. I believe I heard > that troff might be unbundled to either the documentor's workbench > or writer's workbench... They are unbundled in S5R2 distributions (i.e., you don't get *roff with an S5R2 tape). I don't know how AT&T bundles it in binary distributions for the 3Bs. > 5) virtual memory/record locking - I understand that 5.2 does finally > incorporate virtual memory but is that part of the requirements > for ALL versions of 5.2? S5R2 Version 2 (for the VAX) has virtual memory, but S5R2 Version 1 doesn't. It is definitely NOT a requirement for all version of S5R2; what if you port to a machine which can't support virtual memory? > When (or if) the XENIX version that is 5.2 compatable on the IBM PC is > implemented, will it support virtual memory... Not bloody likely; the 808[68] can't support virtual memory (you can't trap on a missing segment) and the 80286 can't support paged virtual memory, so you won't see it on a non-AT PC and I don't think it's in the 80286 microport so you may not see it on an AT either. > Likewise, will all record locking calls be supported identically on all > 5.2 compatable versions? Again, S5R2V2 on the VAX has the record locking calls, but S5R2V1 doesn't. Any S5 based on the releases which have record locking will have compatible versions. > 9) Does anybody know if I can buy a CSH for the 3B2 from any source? I > understand that the bourne shell under 5.2 has some job control > finally (could somebody tell me how this interface works? Is it > like the csh with ctl-z, fb, bg etc.?) S5R2 doesn't have job control. It has "shell layers" which is basically a window system except for the stuff that actually manages the screen; i.e., everything except the part that actually makes windows useful. You have N (~7, I think) "layers" managed by a "window" manager called "shl". Each layer has a pseudo-tty (or its moral equivalent) as its controlling TTY. One layer is connected to your keyboard; whatever you type goes into its pseudo-tty. All layers are connected to your screen/printhead, although you can set "loblk" which is like "tostop" in that any attempt by a process whose layer isn't the "current" layer causes the process to block. You type ^Z and the TTY switches to the layer manager's pseudo-tty; you then type commands at the layer manager to switch the TTY to other layers. The TTY's tty driver modes are the modes of whatever the "current" layer's process has set them to, so if you switch from a layer in cooked mode to a layer in character-at-a-time mode the modes switch automatically. However, NO indication is given to a process that the TTY is being taken away from it or given to it, so it can't clear the screen, or move the cursor to a reasonable place, or take the terminal (as opposed to the TTY driver) out of what funny mode it's put the terminal into (i.e., setting the keypad of a VT100 to transmit numbers or escape sequences, putting a VT100 with a graphics board into Tektronix 4014 or VT100 mode, etc.). It's also not job control in that you can't stop a process and restart it later; you can only take the terminal away from it and give it back later. > but I desperately miss the history, If you can get the Korn shell (AT&T permits a sublicensor to offer the Korn shell in binary form with a system; they should take themselves up on that offer and provide it with 3Bs), you can get history and ~ expansion. Also, if you have source you can snarf the history mechanism for the S5R2 Bourne shell that Arnold Robbins posted to net.sources; I've been using it and it's very nice. > push/pop directorys, I think you can do this with shell functions - I think such a function was posted by Robbins along with his other stuff. > ~ expansion, THe Korn shell and Robbins' changes to the Bourne shell have this. (Unfortunately, I haven't been using it because it has to read the password file itself - trying to use something that does "malloc"s inside the Bourne shell is a headache, due to the way the Bourne shell manages its memory - and on a Sun workstation running 2.0 or later most of the password file is probably *not* on your machine but on a server so reading the password file yourself doesn't help. Sigh...) > and ctl-w word delete. That's not a C shell feature, it's a tty driver feature. Having added it (along with all the other Berkeley extensions) to the S5R2 driver I can testify that it could be in S5R2. Unfortunately, it isn't. Enough complains about this kind of stuff and AT&T might put it in - after all, it worked with the Coca-Cola company... Guy Harris
sambo@ukma.UUCP (Inventor of micro-S) (07/15/85)
In article <363@cuae2.UUCP> heiby@cuae2.UUCP (Ron Heiby) writes: >System V with demand paging is >currently available for the DEC VAX and AT&T 3B20 systems and has been >announced for the 3B2 and 3B15, although it is not yet available there. What is a ^^^^ ? ----------------------------------------- Samuel A. Figueroa, Dept. of CS, Univ. of KY, Lexington, KY 40506-0027 ARPA: ukma!sambo<@ANL-MCS>, or sambo%ukma.uucp@anl-mcs.arpa, or even anlams!ukma!sambo@ucbvax.arpa UUCP: {ucbvax,unmvax,boulder,oddjob}!anlams!ukma!sambo, or cbosgd!ukma!sambo "Micro-S is great, if only people would start using it."
heiby@cuae2.UUCP (Ron Heiby) (07/19/85)
In article <1958@ukma.UUCP> sambo@ukma.UUCP (Inventor of micro-S) writes: >>announced for the 3B2 and 3B15, although it is not yet available there. > What is a ^^^^ ? I don't have the product blurb in front of me at the moment, but it is basically a very high performance 3B5. As I recall, it is based on the WE 32100 instead of the WE 32000, has MAU support, high performance disk, and supports demand paging. It was announced a few weeks ago. -- Ron Heiby heiby@cuae2.UUCP (via ihnp4) AT&T-IS, /app/eng, Lisle, IL (312) 810-6109
wcs@ho95e.UUCP (x0705) (07/21/85)
> In article <363@cuae2.UUCP> heiby@cuae2.UUCP (Ron Heiby) writes: > >System V with demand paging is > >currently available for the DEC VAX and AT&T 3B20 systems and has been > >announced for the 3B2 and 3B15, although it is not yet available there. > What is a ^^^^ ? A 3B15 is a 3B5 with a faster CPU. The CPU is a 14MHz 32100, and comes with the floating point chip. It's 2-3 times as fast as the 3B5 CPU, and about 35% faster than the 3B2/400 CPU (which runs at 10 or 11 MHz) -- Bill Stewart, AT&T Bell Labs, Holmdel NJ 1-201-949-0705 ihnp4!ho95c!wcs
gwyn@brl-tgr.ARPA (Doug Gwyn <gwyn>) (07/21/85)
> What is a 3B15 ?
AT&T just announced this. It appears to be basically a 3B5 with
the new 32100 chip set, including floating-point.
I have also heard that there is a new 3B C compiler that generates
much faster code for both hardware and software implementations of
floating-point. Let's hope so..ped@mtuxo.UUCP (p.davidson) (07/23/85)
REFERENCES: <2067@ucf-cs.UUCP> <363@cuae2.UUCP> <1958@ukma.UUCP>, <47@brl-tgr.ARPA> The 3B5 vs 3B15 3B5 3B15 Max Ports 128 128 Typical # users 16-48 16-60 CPU MIPS 1.0 1.4 Word Size 32 32 Memory RAM 16MB 16MB Cache 8KB 8KB Max Disk Storage 2.7 GB 2.7 GB See the big difference is in MIPS