Bug#303687: Dumps core on start

[Steve Langasek]

--wac7ysb48OaltWcw
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Fri, Apr 08, 2005 at 01:02:43AM -0700, Thomas Bushnell BSG wrote:
> Steve Langasek <vorlon@debian.org> writes:

> > Well, I think it's equally valid to say that this is a libgconf11 bug, =
for
> > trying to launch the gnome2 version of gconfd and *not* falling back to
> > launching the gnome1.2 version.  If people want to fix this via gconf2,
> > that's fine; the important thing then is that this version of gnucash
> > doesn't get into testing without the fixed version of gconf2 -- which it
> > won't.

> Actually, my understanding is that libgconf11 is trying to launch the
> normal gnome-1 thing, and the normal gnome-1 thing is not being
> supported correctly by gnome2, though gnome2 promises to.  But my
> understanding may well be inaccurate in this respect.

An strace shows that libgconf11 is trying to start gconfd-2, not gconfd-1.
It succeeds in starting it, then fails to find the control file in the
desired gnome 1.2 location.

IMHO, the sensible thing for libgconf11 to do in such a case is to fall back
to starting gconfd-1, which is in fact the version in depends on and
therefore the one that is guaranteed to work.  ("Guaranteed" in the sense
that you have a legitimate grievance with the maintainer if it doesn't, at
least.)

> > I'll set the severity to grave, because this *is* an RC issue one way or
> > another; but it also doesn't affect any packages currently in testing. =
 The
> > defensive thing to do, if you care about getting this version of gnucash
> > into sarge, would be to sort out the libgconf11/gtkhtml ungraceful beha=
vior
> > rather than waiting on the gconf2 bug to be fixed...

> Can you please explain why this is RC?  There is a bug, yes, but it
> isn't critical or severe or even serious by the normal standards of
> such bugs, since it has such a trivial workaround.

It's because it's trivially fixable that I believe it's unreasonable to ask
the user to have to employ such workarounds by hand.  Saying "you must crea=
te
this symlink before the package is usable" is a *major* usability gaffe.
Documenting it is not a substitute for providing a suitably usable package;
it's far more trivial for you to kludge around this problem in the package
than to ask each user to do so individually.

> Even if it prevented reports from being generated entirely, it would
> only be important in my opinion, though it might be RC.  But normal
> bugs with trivial workarounds aren't RC.

Actually, in my case I have some reports open by default in my
configuration; this causes gnucash to crash on startup if I don't already
have gconfd-1 running, or haven't created the compat symlink.  Since there's
no GNOME 1.2 environment available in sarge, I'm certainly not going to have
gconfd-1 running by default.

> > Does gnucash work with the woody version of gconf2, btw? =20

> I don't know the history of it well enough to be able to say.

Someone will have to determine that before I'll consider this version of
gnucash releasable, then.

--=20
Steve Langasek
postmodern programmer

--wac7ysb48OaltWcw
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: Digital signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.5 (GNU/Linux)

iD8DBQFCVkSdKN6ufymYLloRAokNAJwPr2y2BJwSow8UNrVaBbQvcGxBoQCgx4Uv
/vlksyhYgoFFyUkAhkWRNyw=
=OE/y
-----END PGP SIGNATURE-----

--wac7ysb48OaltWcw--