On Wednesday, February 10, 2016 03:59:52 PM Jan Mulder wrote: > On 10-02-16 15:28, Sebastian Kügler wrote: > > On Wednesday, February 10, 2016 03:24:38 PM Jan Mulder wrote: > >> As has been noticed by a number of people that are active in the android > >> mobile development, the scrolling/swiping from dive-detail to > >> dive-detail is not "smooth". It stops and moves forward at (lets say) > >> 80% of the width of the profile window, and up to now it is unclear if > >> it is a QML/Qt thing or something in our code. > > > > This is most likely the profile being rendered, which is quite slow. Ideas > > to make it faster: > > > > - cache the dive profiles, so only render them one for a given size, > > re-read> > > from disk; this would of course only help after a dive has been shown > >at least once > > > > - use an imageprovider *1 for the dive profile; this allows to load them > > > > asynchronous, could be combined with the above cache idea > > > > *1 http://doc.qt.io/qt-5/qquickimageprovider.html > > Tempted by the hypothesis of the slow profile rendering being the the > issue of the "strange scrolling behavior", I did the following > experiment. Commented the entire QMLProfile block (and the references to > it) from the DiveDetails.qml. This does result in a page without the > profile, so, with knowing much about the internals of QML, I assume that > this suppresses any dive profile rendering related computations. > > This does not change the strange scrolling behavior. > > I agree that the profile rendering is an computational serious effort, > and that asynchronous computation, and caching are very relevant ideas > to tackle this, but I doubt that it touches the "80% scrolling" behavior.
Hm, I tried this a while ago myself, and back then it had a noticeable effect... I wonder what changed. (It's of course awesome that the profile rendering doesn't seem the culprit anymore.) Next thing I can think of is a too complex details page. -- sebas http://www.kde.org | http://vizZzion.org _______________________________________________ subsurface mailing list [email protected] http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface
