After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 535524 - Codegen install dir "violates" FHS
Codegen install dir "violates" FHS
Status: RESOLVED WONTFIX
Product: pygobject
Classification: Bindings
Component: codegen
2.15.x
Other Linux
: Normal normal
: ---
Assigned To: Nobody's working on this now (help wanted and appreciated)
Python bindings maintainers
: 528066 (view as bug list)
Depends on:
Blocks:
 
 
Reported: 2008-05-29 16:02 UTC by Rémi Cardona
Modified: 2011-04-20 20:15 UTC
See Also:
GNOME target: ---
GNOME version: ---


Attachments
patch to fix codegen install location (1.62 KB, patch)
2008-05-29 16:08 UTC, Rémi Cardona
none Details | Review
patch to fix codegen install location (1.60 KB, patch)
2008-08-19 19:48 UTC, Rémi Cardona
none Details | Review

Description Rémi Cardona 2008-05-29 16:02:00 UTC
By default, codegen is installed in /usr/share/pygtk/2.0/codegen/ which breaks FHS recommendations.

Patch to follow puts codegen in ${pyexecdir}/gtk-2.0/codegen with the rest of pygtk. It of course also updates the pygtk pkg-config file to point to the new location.

All the gnome apps and libraries that depend on pygtk I had on my system rebuilt just fine without any modifications : gnome-mag, pessulus, gedit, gnome-python*, gst-python, pygtkglext, pygtksourceview, d-feet, git, gnome-applets, gnome-menus, deskbar-applet, gnome-games, eog, gimp, rythmbox, totem, avahi, gtk-vnc, epiphany, ccsm, vte and alacarte.
Comment 1 Rémi Cardona 2008-05-29 16:08:18 UTC
Created attachment 111741 [details] [review]
patch to fix codegen install location

Patch that is currently applied in Gentoo to fix the issue
Comment 2 Johan (not receiving bugmail) Dahlin 2008-05-29 17:47:07 UTC
Personally I don't care a lot about FHS compilance, but this will probably break quite some apps.
Comment 3 Mart Raudsepp 2008-05-29 21:06:28 UTC
Yes, having them in /usr/share breaks things. We end up having *.pyc and *.pyo files in /usr/share, and they can get left behind on the system or mixed up or any number of weirdnesses on python upgrades that require things to be rebuilt... all because the tools used don't find them, because they don't look into /usr/share and rightly so - it's for data files, not executable stuff like *.py, *.pyc or *.pyo.
Comment 4 Rémi Cardona 2008-05-30 08:02:28 UTC
(In reply to comment #2)
> Personally I don't care a lot about FHS compilance

Well, I don't really care for it either, but this specific recommendation seems to make sense and the cost to fix it seems really low.

On the other hand, our QA tools really don't expect arch-specific files in /usr/share. So overall, it just seems like good idea to change it.

> but this will probably break quite some apps.

It'll only break those that don't use pkg-config. Which they should :)

So far, we haven't found any that didn't use pkg-config. If we do, we'll of course fix and report bugs to the concerned project. And those apps will still work with unpatched versions of pygtk.

Like the Numeric/Numpy migration, maybe this can be done for the next big release?

Thanks
Comment 5 Rémi Cardona 2008-07-21 13:08:49 UTC
Now that codegen has moved from pygtk to pygobject, I see that the current trunk still installs it in /usr/share.

Now looks like the best time to install it in the correct place.

If not, what's holding this back?

Thanks
Comment 6 Rémi Cardona 2008-08-19 19:48:59 UTC
Created attachment 116998 [details] [review]
patch to fix codegen install location

Ping again :)

Any chance for this patch to land? Since codegen's name has been changed in pygobject 2.15.2 (breaking backwards compatibility), I believe it makes sense to put it in a better location.

Thanks for considering this updated patch
Comment 7 Paul Pogonyshev 2008-08-25 19:12:32 UTC
It doesn't break backward compatibility.  We now have PyGTK install a dummy script that tries to invoke new script (pygobject-codegen-2.0), so that using old name still works.

I'm inclined to postpone this for 3.0 as I don't see much improvement, but there is likely a possibility to break something.
Comment 8 Arun Raghavan 2008-10-19 18:39:46 UTC
*** Bug 528066 has been marked as a duplicate of this bug. ***
Comment 9 Pacho Ramos 2011-04-20 13:04:32 UTC
We are still carrying this with 2.28 and looks to work ok
Comment 10 johnp 2011-04-20 20:15:56 UTC
No point in fixing this now that codegen is deprecated and in maintenance mode.  Closing this.