[comp.sys.att] bash & the att3b1

lefko@vaxwaller.UUCP (Marty Lefkowitz) (08/14/90)

I tried making bash on my 3b1 last night.  I found a few problems.
Seeing as how I bought the machine used, and I'm more used to berkeley
unix, I don't really have a good feel for whats supposed to be on
there.  Anyway I found a number of errors while making. Most were
simple.  

1. When I first ran make it couldn't figure out what machine it was so
I had to tell it.

2. alloca-files/alloca.c had to be compiled by hand because somehow
the makefile believed it was already there.  

3. NO_DUP2 had to be defined because I didn't have one.  Once I did
that NOFILE was undefined so I changed it to ENOENT (or whatever
I thought NOFILE should be)

4. siglist was undefined, and I can't find where the signal table is
so I had to tell it I didn't have one, but I would imagine that I just
couldn't find it.

This is the one that stopped me.

5. The linker could not find opendir(), readir() and creatdir().
Neither could I.  Does anyone know how to get around this problem?

kehoe@scotty.dccs.upenn.edu (Brendan Kehoe) (08/22/90)

In <4440@vaxwaller.UUCP>, lefko@vaxwaller.UUCP writes:
>2. alloca-files/alloca.c had to be compiled by hand because somehow
>the makefile believed it was already there.  

 Hmm...when I did it I didn't have that problem..

>3. NO_DUP2 had to be defined because I didn't have one.  Once I did
>that NOFILE was undefined so I changed it to ENOENT (or whatever
>I thought NOFILE should be)

 GOD was this a pain..it took forever; I sarted to write my own dup2()
(I was *so* pissed that it was in the manual but not installed), then I
finally came upon the define.

>4. siglist was undefined, and I can't find where the signal table is
>so I had to tell it I didn't have one, but I would imagine that I just
>couldn't find it.

 I had to convert this to something like:

	oldsig = signal(SIG_WHATEVERITWAS,newsig)

 Then when sigrestore() was called, I just set it back to OLDSIG; seems
to have worked.

>This is the one that stopped me.
>
>5. The linker could not find opendir(), readir() and creatdir().
>Neither could I.  Does anyone know how to get around this problem?

 Yep...type the stuff in from The C Programming Language's Unix section
(at least that's what I did).
 When I go to work tomorrow I'll go get the sources that I mod'd and
send 'em to you so you can see what's different. (For the readers of
comp.lang.c, this is one of the big reasons I started to hate the 3b1..)

Brendan Kehoe | Soon: brendan@cs.widener.edu
For now: kehoe@scotty.dccs.upenn.edu | Or: bkehoe@widener.bitnet

thad@cup.portal.com (Thad P Floryan) (08/22/90)

kehoe@scotty.dccs.upenn.edu (Brendan Kehoe) in <28707@netnews.upenn.edu> writes
:

	>This is the one that stopped me.
	>
	>5. The linker could not find opendir(), readir() and creatdir().
	>Neither could I.  Does anyone know how to get around this problem?

	 Yep...type the stuff in from The C Programming Language's Unix section
	(at least that's what I did).
	 When I go to work tomorrow I'll go get the sources that I mod'd and
	send 'em to you so you can see what's different. (For the readers of
	comp.lang.c, this is one of the big reasons I started to hate the 3b1..)

Sheesh.  10 lines of code and you start pouting like a 10-year-old kid.

If you want good stuff, get the PD "dirent" code from any archive site,
courtesy of Doug Gwyn.  Works fine on the 3B1 (and greatly helped me to port
much of the BSD "Tahoe" networking stuff to the 3B1).

If you're a real masochist and want some real torture, try using the latest
Apple abortion known as "A/UX 2.0".  Parts of SVR2 and BSD4.2 and not enough
of either to be a real UNIX.  Stuff I have that works fine on scores of other
divers UNIX systems (3B1, SVR4, HP-UX, UTS (Amdahl), etc etc) breaks right and
left under A/UX 2.0.  Just another marketing ploy of Apple (MY OPINION) to
inflict "Finder" on the government.  In many respects, A/UX 1.0 was better
than 2.0 (at least certain programs FUNCTIONED (e.g. zmodem, Emacs, ksh, etc.))
.

Perhaps you should peruse the unix-pc.* newsgroups and the archives at osu-cis,
suny, etc. before complaining.

Or try porting some of "that" code to some versions of SunOS.  Yeah, that's
the ticket!  That'll make a man out of you and put hair on your chest! :-) :-)

Seriously, over the years I've found that porting stuff to the 3B1 (esp. if
one has the latest OS version) is quite straightforward and painless.  It DOES
help to have a copy of include files from other systems to resolve some not-
well-documented peculiarites, but ...

I've also discovered that the subscribers to the comp.sys.att and unix-pc.*
newsgroup hierarchies are more than willing to help people with their problems;
unjustifiably "bad-mouthing" their systems is like biting the hand that feeds
you and will alienate those whom you wish or need to befriend.

If you have a problem: ask for help!

Thad Floryan [ thad@cup.portal.com (OR) ..!sun!portal!cup.portal.com!thad ]

mco@slimer.UUCP (Mark C. Otto) (08/23/90)

In article <28707@netnews.upenn.edu> kehoe@scotty.dccs.upenn.edu (Brendan Kehoe) writes:
>In <4440@vaxwaller.UUCP>, lefko@vaxwaller.UUCP writes:
>>2. alloca-files/alloca.c had to be compiled by hand because somehow
>>the makefile believed it was already there.  
>
> Hmm...when I did it I didn't have that problem..
>

Try picking up alloca from the osu-cis archive for the 3B1/7300.

[...other stuff deleted...]

>>This is the one that stopped me.
>>
>>5. The linker could not find opendir(), readir() and creatdir().
>>Neither could I.  Does anyone know how to get around this problem?
>
> Yep...type the stuff in from The C Programming Language's Unix section
>(at least that's what I did).

Yuck! The osu-cis archive has a POSIX upgrade kit for the 3B1 library
which contains all of these and more - debugged and working.

Your 3B1 came with System V, but you certainly shouldn't have to put up
with ALL of the limitations that implies.  Look around - there are others
who have passed this way before you and have blazed a trail.  Make good
use of the tools they have supplied and you may find the 3B1 to still
be a viable personal UNIX platform. :-)



-- 
Mark C. Otto   EMail: mco@slimer, {teemc | hpftc}!slimer!mco
Voice: 1-313-441-4264    USnail: 5133 Heather #208, Dearborn, MI. 48126
Quote: "Yeah. Right. Kermit my a*s." - Mark C. Otto, '90