tags 483655 + patch
severity 483655 minor
Raphael Hertzog writes ("Re: Bug#483655: queue.length assertion failure"):
> Ping. (Yes I believe one should always upload relevant information
> immediately, it's a pain to have to remember to ping people later on)
Yes, sorry, but I really didn't have time then (I was already late and
was bashing out my mails in a hurry while fixing lots of other stuff
here) and then of course I forgot about it :-(.
I've uploaded it now FWIW at
http://www.chiark.greenend.org.uk/~ian/2008-07-24/varlibdpkg-483655.tar.gz
but the fix is trivial, and attached.
The assertion (and hence the bug) was introduced by me in my triggers
patch, and it causes this assertion failure whenever (and only when)
dpkg is stopping processing packages (for configure or remove) due to
too many errors. The assertion is overzealous and should be bypassed
in that case.
Since the only effect is to turn a correct nonzero exit into a
coredump (and incidentally to suppress the intended display of the
list of unprocessed packages) I think this bug is minor. But it
should be fixed for lenny; it is easy to see that the fix is correct.
Ian.
--- orig2/dpkg-1.14.20/src/packages.c 2008-06-28 15:54:24.000000000 +0100
+++ dpkg-1.14.20/src/packages.c 2008-07-24 00:55:17.000000000 +0100
@@ -233,7 +233,7 @@
/* give up on it from the point of view of other packages, ie reset
istobe */
pkg->clientdata->istobe= itb_normal;
error_unwind(ehflag_bombout);
- if (onerr_abort > 0) break;
+ if (onerr_abort > 0) return;
continue;
}
push_error_handler(&ejbuf,print_error_perpackage,pkg->name);
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
