Should we also add StartupWMClass=someting if StartupNotify=true doesn't work?
在2010-03-15 22:33:58，"Colin Walters" <walters@xxxxxxxxxx> 写道： >On Sun, Mar 14, 2010 at 5:40 PM, Ville Skyttä <ville.skytta@xxxxxx> wrote: >> >> If an app uses GTK+ or Qt, does that alone always imply that it satisfies the >> desktop entry spec's requirements for StartupNotify=true, i.e. no further >> examination of the app's behavior is necessary? > >The main tricky situation comes when the app implements >single-instance behavior internally, and does some sort of IPC (dbus, >whatever) to talk to an existing instance. In GNOME 3 this doesn't >matter as much because we do single-instance by default, but otherwise >it's definitely possible to get the stale "Starting foo..." until it >times out. Actually handling this correctly is tricky[1], and I just >noticed one of my apps doesn't. Maybe I should really take the plunge >and backport app tracking to GNOME 2 which would obviate this. > >But as a general rule: add it. Even for the IPC case, it only occurs >if the user tries to relaunch an existing app, which (albeit without >data, but with some educated background) is unusual. Without IPC, >having it has a 99.9% chance of being correct. > >[1] libunique handles this, but see also my fix for my app here: >http://git.gnome.org/browse/hotssh/commit/?id=57c43b39413c5128983286f5c09cbaba6b397103 > >We have plans to basically make all this work out of the box when >using GTK+ but it blocks on gdbus. >-- >devel mailing list >devel@xxxxxxxxxxxxxxxxxxxxxxx >https://admin.fedoraproject.org/mailman/listinfo/devel
-- devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/devel
