Re: sh.test failed

[Ralf Wildenhues]

Bug report #1:

* Dmitri I Kononchuk wrote on Tue, Nov 23, 2004 at 04:19:30PM CET:
> FAIL: sh.test
*snip*
> `/var/tmp/portage/libtool-1.5.2-r7/work/libtool-1.5.2/tests'
> make: *** [check-recursive] Error 1
> 
> !!! ERROR: sys-devel/libtool-1.5.2-r7 failed.
> !!! Function src_test, Line 520, Exitcode 0
> !!! Make check failed. See above for details.
> !!! If you need support, post the topmost build error, NOT this status
> message.


Bug report #2:

* Lukasz Sowinski wrote on Tue, Nov 23, 2004 at 04:19:07PM CET:
> 
> Application: libtool-1.5.2-r7
> Linux: gentoo-sources-2.4.25
> Problem: sh.test failed

OK guys, here's the deal:
libtool-1.5 was not very good at telling people what to do when the test
does not succeed.  It *should* have said something like this:

CHECKLIST FOR BUG REPORTS:
- Do you have the newest stable version?
  If not, please retry with that (or bug your distro to update).
- Did you take a look at the mailing list archives whether the bug has$
  already been reported? 
  If so, please only report if you have additional information which
  might be of value, or the developers have apparently skipped the
  report (which can happen).
- So you want to report a bug.  How about reading [1]?
- Provide operating system used, versions of all involved software used
  (in this case, especially the shell).
- Re-run the failed tests with VERBOSE=x (this should have been noted
  very prominently in the docs).  Please be aware that the libtool test
  suite has inter- and order-dependent tests, groups of which should all
  be run together.
- Report as much information as you can.  Don't report irrelevant stuff
  (if you know it's irrelevan; if not, more is better than less).

Then your bug reports will become valuable.

I'm sorry to pound on this, I don't want to blame you personally for the
shortcomings of the software, but these reports are next to worthless.

[1] http://www.chiark.greenend.org.uk/~sgtatham/bugs.html

> By the way, why is there so much spam on this list?

Well, because people post unnecessary bug reports for old software.

Oh, you meant the other stuff.  That's because of GNU policy of not
requiring a subscription for posts to bug lists.  No, it does not help
to discuss this on this list.  I think there are special GNU discussion
lists for the mailing list handling, where the people listen that can
change this.

BTW, sorry for everyone who *has* so far provided useful bug reports.
We are very grateful for them, and there have been quite a lot of them.

Regards,
Ralf