[Pysdl-devel] Re: A little trivia

[Martijn F. <fa...@ve...>]

Hi there,

Argh..having some trouble getting my mail sent to the list; sent this
off to Mark alone by mistake. Here it is for you all:
[also see my later notes for the followup]

Mark Baker wrote:
> On Thu, 20 Apr 2000, Martijn Faassen wrote:
> > Odd, I thought I did use HW in my code. Also I recall something in the
> > docs on the color keyed blits bit. I don't have it here but at 
> > home, but I'll look tonight if I get a chance. I'll mail a report
> > on that tonight, hopefully. That'll be more constructive than my
> > vague underinformed bitching of late. :)
> 
> You can specify HWSURFACE, when you create the surface, and if it can't be
> allocated in hardware memory, it'll be allocated in system memory.

That must be what's happening then; I just checked and I do use HWSURFACE.
So I wonder how David found out it wasn't possible; I suppose he did
a query on the surface after creation?

> Color keying works in software, as well as in hardware. 

Must be, then! I think I misread this in the SDL reference:

  SDL_SRCCOLORKEY means that the surface will be used for colorkey blits and
  if the hardware supports hardware acceleration of colorkey blits between two
  surfaces in video memory, to place the surface in video memory if possible,
  otherwise it will be placed in system memory.

Our own PySDL says this:

SRCCOLORKEY

The surface will be used for colorkey blits and if the hardware
supports hardware acceleration of colorkey blits between two surfaces
in video memory, to place the surface in video memory if possible,
otherwise it will be placed in system memory. 

Both are certainly a bit clunky as written, though it's obvious now that I
didn't read them right. I would patch the texts to something like:

  SDL_SRCCOLORKEY indicates that the surface will be used for colorkey blits. 
  If the hardware supports acceleration of colorkey blits between
  two surfaces in video memory, SDL will try to place the surface in 
  video memory. If this isn't possible or if there is no hardware
  acceleration available, the surface will be placed in system memory.

Or something like that. Of course I should send it off to Sam Lantinga, so
I'll do that now. :) David Clark is also informed of this minor documentation
patch. Though I was probably the only one who got confused by the original.

Oooh, I finished off this mail, and *now* I found what confused me:

SDL_SRCCOLORKEY and SDL_SRCALPHA will only be honored if the display surface
is in video video memory (the surface returned by SDL_SetVideoMode() has the
SDL_HWSURFACE flag set.) 

I didn't read the *display* part in display surface, and also managed to
ignore what was between the brackets. Silly me!

Regards,

Martijn