Error output from rpm pre/post scripts (was Re: Florian La Roche please stop mass filing bugs)
Michael Thomas
wart at kobold.org
Thu Jul 5 20:15:02 UTC 2007
Ville Skyttä wrote:
> On Thursday 05 July 2007, Hans de Goede wrote:
>> Florian La Roche wrote:
>>
>>> The error output is a real bug, so just closing this is not very
>>> good.
>> I agree that its a bug. However as said the scriptlet recipe was changed
>> from checking to not checking, I'm sure this was discussed and there were
>> reasons. So lets first discuss this with the people who made this change.
>
> Well, I don't remember if I was involved in that, but if we don't want the
> scriptlet to fail and don't want to display the error output (no matter
> whether it's because of a missing executable or some other error running it),
> dunno if checking first adds any value.
>
> Not adding the dependency is there because it'd pull in GTK to non-GTK setups
> for no gain. This is fine and should be preserved, but actually, I think the
> snippet/guideline should be changed so that the Requires(post) and
> Requires(postun) dependencies *should* be added in cases where the packaged
> app requires GTK itself anyway, and left out just like currently if it
> doesn't. Opinions?
What about adding 'Requires(post): coreutils' to bring in 'touch' for
the gtk+ icon cache scriptlet:
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=246770
It seems excessive to me, because I can't imagine a situation where
coreutils would not get installed. But I also haven't looked that
closely at the dependency tree either.
--Wart
More information about the Fedora-maintainers
mailing list