[MIR] Main inclusion request for evas

Bug #488923 reported by Jamie Bennett
10
This bug affects 1 person
Affects Status Importance Assigned to Milestone
evas (Ubuntu)
Fix Released
Undecided
Unassigned

Bug Description

This package is needed by the mobile team to make the 2D non-accelerated netbook launcher environment available to ARM architectures.

Report:
    https://wiki.ubuntu.com/MainInclusionReportEvas

affects: ubuntu → evas (Ubuntu)
Martin Pitt (pitti)
Changed in evas (Ubuntu):
assignee: nobody → Loïc Minier (lool)
Revision history for this message
Loïc Minier (lool) wrote :

Issues I found while reviewing evas with evas:
* configure: WARNING: unrecognized options: --disable-maintainer-mode, --disable-silent-rules, --disable-rpath
--disable-rpath is set in debian/rules and should be dropped, you might want to send that fix to Debian; not worth diverging here though
* there are various warnings during build which would be worth forwarding upstream, samples:
evas_object_box.c:86: warning: 'opt' may be used uninitialized in this function
evas_object_text.c:1269: warning: pointer targets in passing argument 1 of 'evas_common_font_utf8_get_len' differ in signedness
evas_convert_rgb_8.c:28: warning: right shift count is negative
evas_image_load.c:106: warning: passing argument 1 of 'evas_module_foreach_image_loader' from incompatible pointer type
* this class of warnings is actually serious and should definitely be reported upstream; it's likely to cause segfaults on 64-bits architectures if this code path is under use:
evas_image_main.c:156: warning: implicit declaration of function 'evas_cserve_image_free'

I didn't review ALL warnings, since there are way too many of these: I think the source deserves some cleanup

* I'm a bit worried that fb_setvt() calls chown(vtname, getuid(), getgid()); on a /dev/tty node; I don't think that's the expected use case for Debian/Ubuntu where we usually manage device permissions via udev only
* the build should disable valgrind support explicitly as it might be picked up automatically if installed in the build env and that might end up in the evas.pc file

Overall, the only thing I'm concerned with are the serious warnings above, but these code pathes might or might not be used at runtime.

Since the package is copied verbatim from Debian and the Debian maintainer is also an Ubuntu developer subscribed to the evas bugs in Ubuntu, I'm fine with promoting evas to main. However I think the serious warnings and the chmod() question should be considered important bugs to look at quickly.

Thanks,

Changed in evas (Ubuntu):
status: New → In Progress
status: In Progress → Fix Committed
Revision history for this message
Loïc Minier (lool) wrote :

Hmm it would also be nice to set DEB_DH_MAKESHLIBS_ARGS or DEB_DH_MAKESHLIBS_ARGS_pkg instead of copying the SONAME and versioned package name in debian/libevas-svn-05.shlibs.

Revision history for this message
Albin Tonnerre (lutin) wrote :

Hi there,
I'll fix the minor packaging bits in the next debian upload.
I'll try to fix some compilation warnings upstream so that they get fixed in the next snapshot (which will surely get in lucid), but TBH the evas codebase is moving really fast these days, and it's hard to keep track of things, so there's really no promise that next release will have less warnings

Regards,
Albin

Revision history for this message
Jamie Bennett (jamiebennett) wrote :

Bugs filed upstream.

Martin Pitt (pitti)
Changed in evas (Ubuntu):
assignee: Loïc Minier (lool) → nobody
Revision history for this message
Steve Kowalik (stevenk) wrote :

Promoted to main.

Changed in evas (Ubuntu):
status: Fix Committed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.