[virt-tools-list] virt-manager gtk3 branch now available

[Cole Robinson]

Hi all,

I've pushed the my work-in-progress gtk3 port of virt-manager to git, under
the 'gtk3' branch:

http://git.fedorahosted.org/cgit/virt-manager.git/log/?h=gtk3

It's generally working fine but there are still some big chunks of work I'd
like to finish before cutting a release, outlined in todo.txt in the repo
(also attached):

http://git.fedorahosted.org/cgit/virt-manager.git/tree/todo.txt?h=gtk3&id=dd15447580c42e017beac05e16466317a01a6649

Patches and testing appreciated.

Thanks,
Cole
-------------- next part --------------
  TODO before upstream merge
  ==========================

drop autotools:
- at least investigate it, distutils-extra should give us what we need
- will make the virtinst merge much easier

merge virtinst:
- no big clean up, just get it all + unit tests working
- merge history or just add code?



  TODO before first release
  =========================

check all XXX/TODO in the code, make sure nothing important is missing
verify gettext still works
is tui working? maybe just remove it, check with mcpierce
gsettings port (or fix gconf bindings)
gtkapplication + drop dbus uniqueness stuff? at least investigate it
python3? consider it



  Misc issues
  ===========

If libvirt connection is dropped, app segfaults
segfault if open test:///default, then try and open testdriver that libvirt can't parse XML


test:///default->test->Details: Switch to processor page and 'apply' is activated, shouldn't happen. I think it affects master branch as well


removing hardware can give:
Traceback (most recent call last):
  File "/home/crobinso/src/virt-manager/src/virtManager/details.py", line 1254, in hw_changed
    if self.compare_hw_rows(newrow, oldrow):
  File "/home/crobinso/src/virt-manager/src/virtManager/details.py", line 1226, in compare_hw_rows
    if row1[idx] != row2[idx]:
  File "/usr/lib64/python2.7/site-packages/gi/overrides/Gtk.py", line 1136, in __getitem__
    return self.model.get_value(self.iter, key)
  File "/usr/lib64/python2.7/site-packages/gi/types.py", line 48, in function
    return info.invoke(*args, **kwargs)
TypeError: unknown type (null)


Traceback (most recent call last):
  File "/home/crobinso/src/virt-manager/src/virtManager/console.py", line 1270, in scroll_size_allocate
    self.viewer.display.size_allocate(viewer_alloc)
  File "/usr/lib64/python2.7/site-packages/gi/types.py", line 48, in function
    return info.invoke(*args, **kwargs)
TypeError: argument allocation: Expected cairo.RectangleInt, but got tuple


virt-manager: drop all the HAL code, we could convert to udev but it's not
    really worth it given interface + nodedev APIs


start a spice guest, shrink the window to smaller than guest resolution, scrollbars work but don't have any actual 'bar'.



  After virtinst merge
  ====================

enable all pep8 bits
pylint: drop shell script, stick it in setup.py
drop old stuff from both spec files
update README
pull out all python-virtinst selinux stuff, drop spec dep
virtinst initrd test: drop the big files? just stub em out


  Problems in other packages
  ==========================

gconf set_list: https://bugzilla.gnome.org/show_bug.cgi?id=681433
gtk tooltip: https://bugzilla.gnome.org/show_bug.cgi?id=691639
Keyring bindings broken: https://bugzilla.gnome.org/show_bug.cgi?id=691638

pygobject ctrl-c issue: https://bugzilla.gnome.org/show_bug.cgi?id=691658
get option list for --help: https://bugzilla.gnome.org/show_bug.cgi?id=691659
pygobject row[idx] = None: https://bugzilla.gnome.org/show_bug.cgi?id=691660

rcm-tools koji virt-install patch: http://post-office.corp.redhat.com/archives/rcm-tools/2012-February/msg00054.html


  Manager spacing bug
  ===================

spacing is weird in manager, prefs window. cant have unequal tree rows
- have 3 rows: qemu, test default, test custom
-- connect to test:///default, watch the parent rows expand
- doing col.set_sizing(Gtk.TreeViewColumnSizing.AUTOSIZE) for the stats
    column helps but doesn't fix it entirely, the bottom row will still
    expand weirdly.
- save this till the end, narrow it down to a simple reproducer and try and
    fix, probably a gtk3 issue.
- list size issues: master actually has bigger list size than collapsed gtk3 rows, maybe that is part of the issue.


  Object leaks
  ============

objects leak on cleanup, pygobject has reference counting issues
- https://bugzilla.gnome.org/show_bug.cgi?id=692044
- https://bugzilla.gnome.org/show_bug.cgi?id=693111
- leak detection disabled in the manager.py for now
- see if gtk.grab_get_current autodrawer leak fixed by gtk3