Re: [AD] Big BeOS patch

[ Thread Index | Date Index | More lists.liballeg.org/allegro-developers Archives ]


First of all, I'll let you know I already submitted to sourceforge another 
BeOS patch that fixes some things and adds hardware acceleration support to 
the fullscreen driver. I uploaded it to the Allegro project page last night, 
but had no time to post a message here... Anyway, the new patch require the 
other to be already applied (i.e. as Allegro in CVS right now).
About CVS, I've downloaded the lib as is there, and it compiles under BeOS. 
The only problem is with dependencies: from a fresh CVS installation, I 
simply do fixbe.sh, make and make install, but this way during the make 
process I have a problem with text.c, searching for font.h (not there 
anymore). If I run "make depend" before make, the dependencies are fixed, but 
something gets broken, and I have to manually run mmxtest, create plugins.h 
and other things like this I don't remember... On the other hand, if I don't 
run "make depend", but I simply edit obj/beos/alleg/makefile.dep and I toggle 
the font.h dependency in text.c, everything works fine.
As I said, I don't know how to generate dependencies, so I'd really like to 
know what you did to compile the thing under your Be box.

Now, about the new patch I submitted, here's a list of things it adds:

- hardware accelerated rect, hline, vline, clear_to_color and blit.
- mouse fix in windowed mode: if cursor is not on program window, the Allegro 
cursor disappears from it.
- added desktop_color_depth() support and finally yield_timeslice() (not 
having it under a multithread system as BeOS was a shame)

Ok, now on the comments about the old patch:

> - some program backgrounds don't get drawn properly in 8 bit desktop, e.g.
>   exhello, extimer, etc.
> - expal doesn't work properly on 8 bit desktop

Humm, I've worked on 16bit mode only, and now I see how dumb I've been not 
testing the thing under other resolutions/depths...

> - after using windowed mode, cursor is hidden when over allegro_message
>   window (test program: demo)

Hiding/showing the cursor under Be is done at BApplication class level only 
AFAIK, so this is an expected side effect... Message boxes are still part of 
the Allegro app.

> - "phony" mouse cursor should move even if window unfocused

Don't know what you mean here; try the new patch and see if it fixes this...

> - occasionally leaves mess on the desktop after quitting

?? Never had such problems. Can you tell me exactly what you did? (and what 
was the configuration of your desktop)

> - claims to support hardward scrolling, but it doesn't (AFAICS)

Humm, I also think scrolling in windowed mode should be disabled, as it makes 
no sense.

> - switching desktop color depth while program running doesn't work
> perfectly

I'll check this out.

I've another question for the BeOS fullscreen driver: currently it seems to 
support only small virtual screen sizes (1024x1024 using test program in 
640x480 mode for me, and I have 32MB of video RAM - btw I've also tried to 
disable BeOS workspaces), and I wonder if it could be possible to set larger 
ones... Now that hw acceleration is possible, it would be cool to have larger 
video ram portions available to the user.
My tests with the BWindowScreen class had no luck in setting larger 
framebuffers, so I wonder if the BScreen class could help. Any thoughts?

-- 
Angelo Mottola
a.mottola@xxxxxxxxxx
ICQ UIN 66972680



Mail converted by MHonArc 2.6.19+ http://listengine.tuxfamily.org/