On 8 June 2011 04:33, David Lowe <doctorjl...@verizon.net> wrote:
> On 7 Jun, 2011, at 9:32 AM, Marko Lindqvist wrote:
>> Qt-client development
>        Until somebody better comes along, i can at least at least keep this 
> ball rolling.  Caveat: i've only worked through the TrollTech tutorials last 
> semester while learning C++, so i'm far from being an expert.  Since i've 
> never done anything this big before, i could benefit if somebody with "the 
> big picture" gave a breakdown of the tasks.

 By the time I had to stop Qt-client development due to time
restrictions, I had also encountered design problem in fitting client
common code and Qt models of graphics canvas together. This probably
needs to be resolved before any other work on Qt-client makes sense.

 Client common code just overdraws old sprites with new ones, possibly
even with completely black ones, when screen needs update. In fact it
doesn't even remember exact sprites it has previously drawn.
 Qt-classes I looked at the time worked on the idea that they store
graphical items to be shown. While you can place such items on top of
each other (overdraw), old ones would still need to be removed to free
memory and not to use processing time when refreshing screen.

 I'd start this task by investigating if there's some other way in Qt
to implement canvas than one requiring explicit removal of items, as
it would require bug changes to client common code to handle the

 I did implement canvas that has this memory usage problem. So if you
want to see what Qt-client looks like with it, you can test it. But
usable for longer use it is not as client gets slower with every
screen refresh (more items to handle). See patch #2443.

 - ML

Freeciv-dev mailing list

Reply via email to