3.0 testing priorities

[Dirk Hohndel]

On February 25, 2013 4:49:00 AM "Lubomir I. Ivanov" 
<neolit123 at gmail.com> wrote:
> On 25 February 2013 10:05, Amit Chaudhuri <amit.k.chaudhuri at gmail.com> wrote:
> > Testing OS X
> >
> > Deleted all previous applications (drag to trash).
> >
> > Confirmed I have the latest ( - 26 - sha etc)
> >
> > Started to test various basic menu options.  Response times are pretty
> > rubbish...
> >
> > On Linux (admittedly half decent 64 bit desktop machine) menu responses are
> > immediate.  On the Mac Book Pro, they are so slow I'm thinking "this app is
> > going to drive me nuts"....
> >
> > Returning to the app on subsequent tests things seem to get a bit snappier.
> > It's as if the first time it runs something it really can't be bothered but
> > then it learns how to go a bit quicker.  But these are things I'd expect to
> > happen really fast.
> >
>
> well GTK is a layer on top of the Windows and OSX native API, so with
> abstraction there should be a a slight performance drop, but noting
> that noticeable on a multi-core.
> also involved are the internal ways GTK handles signals, layouts
> updates and invalidation. on Windows i started noticing some slight
> visual glitching related to the selection cursor hacks we are doing,
> but i don't think it will affect the end user much.
>
> given you mention a laptop designed for OSX , things should be quite
> snappy on that with GTK, or least this is the first ever report
> related to a performance issue, if i recall.
> perhaps something on your end.

No, this is a known issue with gtk on Mac.
One more reason to switch to qt

> > The map widget seems to come up fast.  The about dialog not so.  Planner
> > also rather sluggish.  Does gtk on mac go around the houses somehow? And
> > does the map widget by pass this stuff?
> >
>
> that's surprising considering the widget is quite "incomplete" (trying
> _really_ hard to structure the sentence in a polite way here).
> on the visual/performance side, a personal annoyance is the obvious
> lack of off-screen rendering for the OSD.
>
> > To test the web service, do I need some sort of id?
> >
>
> a "Get UID" button that opens the browser for the user to register
> might not be a bad idea.
> i can add that if requested. does anyone think this is a good addition?

Yes, definitely

/D