[comp.windows.x] 16 bit server natural format

dale@amiga.UUCP (Dale Luck) (06/07/88)

   After about a week of heavy text editing and debugging I've now have
a mfb that uses 16 bits as it's natural working unit instead of 32 bits.
A lot of 'ints','unsigned','unsigned ints', doesn't anyone want to use
the same defination?

   Also places where 31,32,5,4,2 constants are used were changed to
WORK_MAXX,WORKSIZEBITS,LOG2_WORKSIZEBITS,WORKSIZEBYTES,LOG2_WORKSIZEBYTES
Anyway, what I have now is a new type def 'worktype' which I specify
in server/include/servermd.h. So to generate a 16 or 32 bit server all I
need to do is change a couple #defines in servermd.h and recompile about
half the modules. This has definately not been checked out with the other
device specific drivers. Hopefully going to 64 bits will be easier now.

   Is anyone interested in this work? I think using one typedef for the
working unit makes much cleaner code as well as you then know what variables
are actually going to deal with the raster as opposed to what variables are
just poking at width values to set up a fillspan.

   In my opinion it makes it more maintainable and flexible.

   It has been tested with the latest PURDUE patches and everything seems to
work just fine.

   I would like to contribute this effort to the Mit standard distribution if
this is appropriate. If not I could just post the diffs and those interested
could grab them themselves.

Dale Luck