Mail Archives: cygwin/2005/07/04/17:56:03
--------------enigBA538C078FBBDBC4A0F5243D
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
>>>After upgrading cygwin yesterday I get the following reproducible hang
>>>in a perl script starting an external program.
>>>
>>>This is the perl script that works with the 1.5.17 cygwin dll and hangs
>>>with 1.5.18:
>>
>>Did you also see this with snapshots?
I didn't try. I will first try to condense the problem to a simpler testcase.
If that fails I can do a binary search backward.
> While waiting for the answer to the above rhetorical question, I thought
> I'd add an observation: Rather than go to a lot of effort running things
> and getting straces which may or may not illustrate the problem, it is
> ALWAYS a much better plan to reduce things to a simple test case which
> can be reproduced by people who may or may not use strace to debug the
> problem.
I know. I reduced the 10h build of OpenOffice.org to the 5 lines of perl
script that fails. I didn't expect an immediate solution for the problem,
but I attached the end of the strace anyway in the hope that it might
trigger an idea for someone.
> The division of labor should be like this:
>
> The reporter more or less understands the code for which they are
> reporting problems so they should be able to reduce things, as much as
> possible, to a test case which reproduces the problem. With a test case
> in hand, you can then hand off the problem to someone who understands
> cygwin and who will be able to use the simple test case to debug the
> problem.
Fair enough.
> Attempting to bypass the test case step and do the cygwin maintainer's
> "job" of generating strace output is not as likely to be a worthwhile
> endeavor as producing the test case itself. You are not likely to jump
> start the debugging process by doing this. Even if the strace output
> was useful, the debugger would have no real way of knowing if they fixed
> the problem without a test case.
Well, I could have waited reporting this problem until I have extracted
a 2 line testcase, but I wanted to post what I have so far.
> The theory here is that a bug reporter shouldn't need to spend any time
> trying to debug cygwin (via strace) unless they really are interested in
> learning about cygwin internals. That means that the most profitable
> thing a bug reporter can do is create a test case.
Quote from the original mail:
> * Attention! Stop reading if you're offended by unsolicited strace outputs. *
Stop wasting your precious time complaining about my useless strace. I didn't
waste much time producing it and it shows that perl is waiting for
something, not more.
Regards
Volker
--
PGP/GPG key (ID: 0x9F8A785D) available from wwwkeys.de.pgp.net
key-fingerprint 550D F17E B082 A3E9 F913 9E53 3D35 C9BA 9F8A 785D
--------------enigBA538C078FBBDBC4A0F5243D
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename="signature.asc"
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (MingW32)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
iD8DBQFCybDtPTXJup+KeF0RAsXuAJ4gUv0SebhbvezBbQmFgispA/RilwCfUKFW
Uq1aYbolYVxFqpS8EF6e538=
=4TXL
-----END PGP SIGNATURE-----
--------------enigBA538C078FBBDBC4A0F5243D--
- Raw text -