Michal Pryc wrote: > Shawn, > I was trying to reproduce this bug, but I can't publish any package to > my repository as > Shawn Walker wrote: >> Michal Pryc wrote: >>> Shawn Walker wrote: >>>> Michal Pryc wrote: >>>>> Hello, >>>>> New webrev available at: >>>>> 2/ >>>>> >>>>> Comparing to the previous webrev, this changed the way of handling >>>>> some >>>>> widgets (some of them didn't have gobject.idle_add and should have, so >>>>> now all that need this - the ones that are inside Thread contains >>>>> this) >>>>> >>>>> Also when the user clicks cancel during install/update evaluate stage >>>>> the CancelException is raised and handled properly. Before the >>>>> function >>>>> just returned which caused that the user couldn't quit properly >>>>> application. >>>> >>>> Your code changes look fine, but I'm still testing. This 2nd version >>>> resolved a segmentation fault that was present with the 1st version of >>>> the patch. >>> >>> Shawn, >>> Most of the seg faults in the pygtk are because the gtk functions were >>> called from another Thread without gobject.idle_add. I have reviewed >>> all widgets in the remove.py and installupdate.py and made proper >>> changes prior to the 1st version as John pointed in his review. >>> >>> Please let me know when you will decide "go/no go" for those changes :) >> >> A few more breakages: >> >> 0) applied v2 of your patch to my workspace, and changed is_admin to >> always return 1 >> 1) started copy of build 96 repository on port 10000 >> 2) pkg image-create -a authname=http://localhost:10000 ~/devel/test-image >> 3) pkg -R ~/devel/test-image install pkg:/[EMAIL PROTECTED] >> 4) packagemanager -R ~/devel/test-image >> >> Received this warning: >> /export/home/swalker/devel/pkg-1449-gui/proto/root_i386/usr/bin/packagemanager:1372: >> >> >> GtkWarning: file gtkliststore.c: line 1441: assertion `new_order != >> NULL' failed >> self.category_list.reorder([r[-1] for r in rows]) > > This is because the category file is not present for authname on the > system and we are trying to alphabetical resort list which is 0 element > list. The new webrev fixes this by checking if the category list lenght > is greater then 0: > > http://cr.opensolaris.org/~migi/04_09_2008_sync_with_ips_bug_3193_v5 > >> 5) Clicked "Update All" >> >> Exception in thread Thread-5: >> Traceback (most recent call last): >> File "/usr/lib/python2.4/threading.py", line 442, in __bootstrap >> self.run() >> File "/usr/lib/python2.4/threading.py", line 422, in run >> self.__target(*self.__args, **self.__kwargs) >> File >> "/export/home/swalker/devel/pkg-1449-gui/proto/root_i386/usr/bin/packagemanager", >> >> >> line 1023, in __do_ips_uptodate_check >> self.ips_uptodate = self.__ipkg_ipkgui_uptodate(self.image_o) >> File >> "/export/home/swalker/devel/pkg-1449-gui/proto/root_i386/usr/bin/packagemanager", >> >> >> line 1050, in __ipkg_ipkgui_uptodate >> img.retrieve_catalogs() >> File "/usr/lib/python2.4/vendor-packages/pkg/client/image.py", line >> 1145, in retrieve_catalogs >> auth = self.installed_file_authority(path) >> RuntimeError: ([({'origin': 'http://pkg.opensolaris.org:80/', 'ssl_key': >> 'None', 'prefix': 'opensolaris.org', 'ssl_cert': 'None', 'mirrors': []}, >> <exceptions.IOError instance at 0xa9faeac>)], 1, 0) >> >> I had to kill the client at this point as it was "hung". > > Were you running together with the newest SUNWipkg bits from the gate? > > We were missing part for checking if the image is the opensolaris one > and every time we were trying to update SUNWipkg-gui/SUNWipkg which was > not good. Now together with this webrev I've added the bits to check > this, in the same way as is in the client.py. > > I don't want to make much more changes in this webrev, since they don't > necessary are fixes for bug 3193 and I am starting to be lost :) :) > > If you think that this webrev addresses the previous issues, I will > commit this and then simply we can fix other (if any) bugs.
My feeling is that your webrev addresses the previous issues adequately. There are likely other, separate issues to be addressed here. The only thing I would ask is whether you have tested the packagemanager gui using "udpate all" yet? If so, and it works, than my personal view is that "you are good to go." Cheers, -- Shawn Walker _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
