Mailing List Archive: Final remarks about RHEL6

Final remarks about RHEL6

Feb 11, 2011, 3:22 AM

Post #1 of 3 (1988 views)

Hi, Axel

I think we have now a minimum set of packages to turn rhel6 usable from a
desktop point of view.

However, I have a question: what is the official qt4 ATrpms version?

When I installed avidemux, it had an unresolved symbol because I was using
the
stock qt 4.6.2. That is, the package installed, but the executable did not
run (not very nice..).

avidemux: symbol lookup error: avidemux: undefined symbol:
_ZN9QListData11detach_growEPii

also, see:

http://forums.opensuse.org/english/get-technical-help-here/applications/446948-symbol-lookup-errors-after-updating-libkdecore4.html

Well, if qt47 is mandatory, because you seem to be linking everything
against it,
it can not be on testing. It has to go to stable.

The other option is forgetting qt47, and start using qtwebkit, possibly
including
it as a BR in some specs (e.g., mythtv). The absence of qtwebkit was the
only reason for providing qt47 on rhel6. Qt seems to be compatible top down
only, I mean, it works when having a higher version replacing
a lower version, and not the other way around.

Finally, will we have a rhel6 kernel-tuxonice?

Thanks.

--
Paulo Roma Cavalcanti
LCG - UFRJ

Re: Final remarks about RHEL6 [ In reply to ]

Axel.Thimm at ATrpms

Feb 24, 2011, 3:16 PM

Post #2 of 3 (1925 views)

Permalink

On Fri, 2011-02-11 at 09:22 -0200, Paulo Cavalcanti wrote:
> I think we have now a minimum set of packages to turn rhel6 usable
> from a desktop point of view.

:)

> However, I have a question: what is the official qt4 ATrpms version?
>
> When I installed avidemux, it had an unresolved symbol because I was
> using the
> stock qt 4.6.2. That is, the package installed, but the executable did
> not run (not very nice..).
>
> avidemux: symbol lookup error: avidemux: undefined symbol:
> _ZN9QListData11detach_growEPii
>
> also, see:
>
> http://forums.opensuse.org/english/get-technical-help-here/applications/446948-symbol-lookup-errors-after-updating-libkdecore4.html
>
> Well, if qt47 is mandatory, because you seem to be linking everything
> against it,
> it can not be on testing. It has to go to stable.

:/
The repo need to define proper names for the bist used. qt47 will
probably never make it to what we currently call "stable" as that is
supposed to not touch upstream packages.

> The other option is forgetting qt47, and start using qtwebkit,
> possibly including
> it as a BR in some specs (e.g., mythtv). The absence of qtwebkit was
> the only reason for providing qt47 on rhel6.

I don't yet trust qtwebkit as much as I'd like to. The reason atwebkit
was dropped from RHEL builds were too many security issues. ATM I feel
more confident tracking qt from Fedora (and thus the security updates to
webkit) than staying with an older version where future security issues
will require backporting of fixes.

> Qt seems to be compatible top down only, I mean, it works when having
> a higher version replacing
> a lower version, and not the other way around.

Yes, that's true of all libs. If done correctly the soname should change
even for just adding new symbols to libraries for exactly this reason.
The same has happened with glib/gtk etc. over the years.

The remedy for such issues are rpaths, but these are not in favour in
Fedora/RHEL world.

> Finally, will we have a rhel6 kernel-tuxonice?

Probably a question to ask at Matthias while offering access to RHEL6
systems/licenses.
--
http://thimm.gr/ - http://ATrpms.net/

Re: Final remarks about RHEL6 [ In reply to ]

promac at gmail

Feb 25, 2011, 3:45 AM

Post #3 of 3 (1914 views)

Permalink

On Thu, Feb 24, 2011 at 8:16 PM, Axel Thimm <Axel.Thimm@atrpms.net> wrote:

> On Fri, 2011-02-11 at 09:22 -0200, Paulo Cavalcanti wrote:
> > I think we have now a minimum set of packages to turn rhel6 usable
> > from a desktop point of view.
>
> :)
>
> > However, I have a question: what is the official qt4 ATrpms version?
> >
> > When I installed avidemux, it had an unresolved symbol because I was
> > using the
> > stock qt 4.6.2. That is, the package installed, but the executable did
> > not run (not very nice..).
> >
> > avidemux: symbol lookup error: avidemux: undefined symbol:
> > _ZN9QListData11detach_growEPii
> >
> > also, see:
> >
> >
> http://forums.opensuse.org/english/get-technical-help-here/applications/446948-symbol-lookup-errors-after-updating-libkdecore4.html
> >
> > Well, if qt47 is mandatory, because you seem to be linking everything
> > against it,
> > it can not be on testing. It has to go to stable.
>
> :/
> The repo need to define proper names for the bist used. qt47 will
> probably never make it to what we currently call "stable" as that is
> supposed to not touch upstream packages.
>
>
In this specific case, avidemux did not pull qt 47, and this is the problem.
How one would know what to do to satisfy the unresolved reference?

May I suggest a simple work around?

Why not using the green colour this way:

"good packages that have been working very well, but depend on the testing
repo being activated."

The green colour has never been used in ATrpms site before, and it could
inform all of us what packages have been linked
against the testing repo.

--
Paulo Roma Cavalcanti
LCG - UFRJ

Mailing List Archive

Mailing List Archive

Attached Files: