Hi,

Several cleanups. Gtk people, please test your compilation, I don't have cross-compile ...

Are there test cases / examples for fpimage support?

Regards,

Micha.

Heh, tried an updated version of speedtest.pp. To penguin turns up blue+pink :). Is RRGGBB swapped incorrectly then?

Regards,

Micha.

Hi, Several cleanups. Gtk people, please test your compilation, I don't have cross-compile ...

seems to work

Are there test cases / examples for fpimage support?

None yet.
Simply load some bmp, xpm or png files. Or you can try the jpeglazarus package on lazarus-ccr.

Mattias

Do the raw images support ByteOrder already? I changed it, but it doesn't seem to have an effect. I also can't find any mention of ByteOrder intfgraphics.pas.

Micha.

I am trying this with the rawimages :). I now have the colours of the main bitmap ok, but the alpha bitmap is still a problem. It looks weird, not the outline of the penguin, but...can't really tell. I have specified alphabitperpixel = 1, is that valid?

Micha.

I have a very strange troubles with current lazarus and current fpc development branches. I have full cvs of FPC installed, and /usr/src/fpc is written in "fpc path". But no fpc unit (including system) is found by codetools.

You can see what the codetools found in your FPC source directory: Source Editor -> popup Menu -> Unit Info -> Show Codetools Values -> #UnitLinks

If there are no units, maybe the cache needs to be rebuild: Change the FPC Src Dir to something , leave the environment options dialog (ignore warning), then set the correct path.

Mattias

That's because intel is the 'illogical' way, the bytes are reversed relative what you think it should be, like: 3 2 1 0 7 6 5 4 ... 3 is least significant; i use low significance = low memory address :). Powerpc is the 'right' way.

Micha.

The penguin is like any other penguin black and white. If you got colors, then
- the xpm color map loading is buggy
- the creation of the pixels is buggy.

Please check the CreatePixmapIndirect function in the win32 intf.

Or: wait till we find a clean solution for rawimages and you and I will have implemented it.

Mattias

Attached patch seems to make things a little better, but still the right and left half of the image are swapped. Mattias, maybe it gives you a hint? Alpha bitmaps are monochrome on win32 afaik, so this functionality is crucial.

Micha.

Index: intfgraphics.pas
=================================================================== RCS file: /FPC/CVS/projects/lazarus/lcl/intfgraphics.pas,v retrieving revision 1.10
diff -u -b -r1.10 intfgraphics.pas
--- intfgraphics.pas 7 Nov 2003 22:50:44 -0000 1.10 +++ intfgraphics.pas 13 Nov 2003 21:20:05 -0000
@@ -803,6 +803,8 @@

   BitOffset:=FDataDescription.AlphaBitsPerPixel*cardinal(x)+Position.Bit;
   Position.Bit:=BitOffset and 7;
   inc(Position.Byte,BitOffset shr 3);

+ if FDataDescription.ByteOrder = riboLSBFirst then
+ Position.Byte := (Position.Byte and not $3) + (3 - (Position.Byte and $3)); end;

 procedure TLazIntfImage.LoadFromDevice(DC: HDC);

I have a very strange troubles with current lazarus and current fpc development branches. I have full cvs of FPC installed, and /usr/src/fpc is written in "fpc path". But no fpc unit (including system) is found by codetools.

Now I think I see it: it seems the bytes are swapped within the dwords or so. Weird that the bitmap looks good, but the alpha mask does not.

Micha.

No, it does not yet. I thought, we need it for powerpc, not intel. Feel free to implement.

Mattias