Delivery-Date: Sat, 19 Mar 94 09:08:53 -0800
Message-Id: <199403191708.JAA17589@nobozo.CS.Berkeley.EDU>
To: Marc Teitelbaum <marc@vangogh.CS.Berkeley.EDU>
Cc: aoki@postgres.Berkeley.EDU
Subject: Re: Preliminary release of POSTGRES V4R2 
In-Reply-To: Your message of "Fri, 18 Mar 1994 14:08:37 MST."
             <199403182208.OAA02241@vangogh.CS.Berkeley.EDU> 
Date: Sat, 19 Mar 1994 10:08:43 -0700
From: Daniel Simmons <simmdan@kenya.isu.edu>


Mark and Paul,

Here are a few quick responses to your questions about my small
difficulties with the HP port of Postgres V4R2:

(BTW, I must apologize if there were some stupid/obvious mistakes in
my earlier message reporting difficulties, because I was far too tired
when I wrote that message.  I've slept quite a bit since then, and my
comments should be at least a LITTLE more sensible.)

Marc writes:
>The "include" problem is because there were header files left
>in the include directory.  Normally, BSD install is perfectly
>happy to install over a read-only target (in fact, that's the common
>route - recompile some programs and re-install over the 0555 programs
>currently in /bin, /usr/bin or whereever).  I wonder what the
>HP system makefiles look like (actually - no, I don't wonder... :)
>So the HP install I guess doesn't want to.  So why isn't this usually
>a problem?  Hmm - second barf - seems that the HP install installs my
>0444 header files as mode 0755 !!! (gag) - so I guess you won't
>run into this as long as I don't leave any junk around...

Well, as you have apparently figured out, the problem seems to be with
the bsdinst script.  I was using the one that came with X11R5.
Naturally, this wasn't really a big deal.  As I mentioned before I
fixed it by manually changing permissions on the include files and
then rerunning the install.

Marc writes:
>I'm re-installing on an HP here to check out the shell script
>problems.  All of the testing on the HP's was done by Paul, and it's
>possible he had stuff setup in his environment such that he
>didn't run into problems, but right off the bat I see problems.
>Like $USER isn't set on HP's, and we use it all over our scripts.
>And some unprotected shell variables that _should_be a problem
>on our other platforms as well, but I guess aren't for some as of
>yet unknown reason.  I'll give it a going over.

Apparently the fact that I didn't have USER set in my "postgres"
account is why I had problems getting createuser to work with that
account but did not have porblems with createuser from another account
(which I now realize DID have USER set).  

Paul writes:
>ok, so bsdinst doesn't work well if you run it twice.  were there
>other problems,

Well, the biggest concern was with the regression tests failing.  I
did run them from an account with USER set and which was not my
"postgres" account.  At first they failed because they couldn't create
the regression output file because of file permissions.  Then, I ran
them again, and they ran for a long time before croaking on something
having to do with the inversion file system.  The regression output
file was not less than half as large as the sample output file.  So, I
am assuming the tests were not nearly done.

Also, when I ran the checkdiff script it barfed all over the place.  I
mean, I don't think it was just complaining about the tests not
finishing.  I think it was reporting errors for virtually every query.
I must say, though, that I didn't look too closely.

At any rate, it seems to be working for the basics.  So, I'm going to
start working on my project.  I'll know more as my project progresses
and I start using it for "real" work.  Again, let me know if there is
anything I can do to help get it all cleaned up.  It would make me
(and my "customers" for this project--they aren't customers in the
strictist sense of the word, but they are the people around the
university who will be using the software when we are done...)  a
little more comfortable if my installed copy passed the regression
tests.

Thanks again for all your work on this and for quickly getting back to
me.

Danny

-------
Daniel Simmons          electronic mail : simmdan@isu.edu
Idaho State University       voice mail : (208) 236-3199
Computer Center              snail mail : Box 8037, Pocatello