[comp.lang.perl] PL36: Is this a bug ???

mleiber@hpwad.HP.COM (#Michael Leiber) (10/22/90)

Hi,

I just applied patches 29 thru 36 to perl(1) and ran the regression tests.
It looks as though there's a bug, as reported by t/op.read when running
that test on an nfs mounted file. 

Here's the result:

----snip--------------------------------------------------------------snip----
>perl op.read
1..4
ok 1
ok 2
not ok 3
not ok 4
----snip--------------------------------------------------------------snip----


Here's the context:

both machines are running under HP-UX 7.0

nfs server:

HP-UX hpwadrd 7.0 B 9000/370 hpwadrd

nfs client:

HP-UX hpwadlk 7.0 B 9000/360 hpwadlk

checklist:
# logical (nfs) file systems
hpwadrd:/usr/local /usr/local nfs defaults 0 0

the tests reside in </usr/local/src/perl/pl36/t>

When running the test on the server machine, it works fine, but running
it on the client fails.

I think the problem has to be searched within seek/read on unix rather
than in perl, but I just wanted to report that problem. Unfortunately
I didn't have the time right know to track it down, so if someone else ...

to Larry: perl's a really incredible tool, keep on working and enhancing
	  it. I'm also happy to see people working on MeSsy-DOS ports, 
	  that makes this beasty machine a bit more comfortable. Great stuff !

			regards   michael

------------------------------------------------------------------------------
internet: mleiber@hpwadlk
phone:	  Germany 7243 602527
address:  Michael Leiber, Hewlett Packard, Waldbronn Analytic Division,
	  Hewlett Packard Str. 1, 7517 Waldbronn, West Germany
------------------------------------------------------------------------------

pcn@daimi.aau.dk (Peder Chr. N|rgaard) (10/23/90)

mleiber@hpwad.HP.COM (#Michael Leiber) writes:


>Hi,

>I just applied patches 29 thru 36 to perl(1) and ran the regression tests.
>It looks as though there's a bug, as reported by t/op.read when running
>that test on an nfs mounted file. 

>both machines are running under HP-UX 7.0

>When running the test on the server machine, it works fine, but running
>it on the client fails.

>I think the problem has to be searched within seek/read on unix rather
>than in perl, but I just wanted to report that problem. Unfortunately
>I didn't have the time right know to track it down, so if someone else ...

We tracked that one down long ago, so just to save other people the 
trouble:  This is an error in HP-UX 7.0, NOT in perl.  Your regression
test on perl.3.28 and any earlier version of perl (and HP-UX) would
have provoked the same error.

Larry's regression test tests a rather obscure part of UNIX file semantics
- what happens if you seek far below end of file on a file and
try to read.  It should, of course, fail.  Well, on a HP-UX system
working on an NFS file, it doesn't.  No big problem for anyone,
(I mean, to do that in a program, you would have to commit a logical
programming error first!) but of course one might suggest that Hewlett
Packard ought to correct their NFS code some day.

Greetings
				--peder chr.

--
Peder Chr. Norgaard              ****              pcnorgaard@daimi.aau.dk 
              Computer Science Department,  Aarhus University
               Ny Munkegade 116,  DK-8000 Aarhus C,  DENMARK
phone: +45 86 12 71 88 / telefax: +45 86 13 57 25 / telex: 64767 aausci dk