[plasmashell] [Bug 315488] icon-only task manager groups chrome/chromium web apps with chrome/chromium

2016-03-21 Thread GaryM via KDE Bugzilla
https://bugs.kde.org/show_bug.cgi?id=315488

--- Comment #27 from GaryM  ---
Forgot to say.  When using the launcher rules you can have multiple launchers
which work as desired but then vanish when grouping kicks in only to then
reappear when their linked application is closed. 

I understand what you were saying about Google changing the ClassName for
webapps, but it seems the current situation is that it works for launchers, but
not grouping, resulting in some quite interesting behaviours when the launchers
also stack into  the group and then pop back out again.

If the ClassName is never used in grouping, I'm unclear what benefit there is
to linking a launcher to a specific ClassClass and ClassName.  It would seem to
only make sense to link a launcher to a specific ClassClass otherwise you get
this slightly unexpected launcher stacking/hiding behaviour.

-- 
You are receiving this mail because:
You are watching all bug changes.


[plasmashell] [Bug 315488] icon-only task manager groups chrome/chromium web apps with chrome/chromium

2016-03-21 Thread GaryM via KDE Bugzilla
https://bugs.kde.org/show_bug.cgi?id=315488

GaryM  changed:

   What|Removed |Added

 CC||garymarty...@gmail.com

--- Comment #26 from GaryM  ---
It seems to some extent the code is in place.  If you create a launcher
matching rule for a Chrome WebApp launcher it looks at the Windows Class and
Window Name in order to match them.  So you can create a series of launchers
for web apps and if you run one of them it correctly matches to the launcher. 
It's only when you run a second and it 'groups' them that they stack under a
single icon.

If the code is already reading both Class & Name fields to match to a launcher,
it would seem that this is already an indication that the item would make sense
to not be grouped with others of the same class, or indeed that a flag in the
launcher matching rule to 'group independant of core class' (or some better
term) would not add any significant additional complexity for users (given
they've already gone so far as to make the rules).   

Could this logic then be fed through to the grouping algorithm to keep them
separated out when grouping kicks in?

-- 
You are receiving this mail because:
You are watching all bug changes.