Re: Musing about the new API

[Just van Rossum]

At 7:21 PM +0100 19-03-2000, Pavel Kankovsky wrote:
[ ... ]
All good points...

>5. Is it necessary to have one implementation of FT_Outline_Decompose() in
>ftraster.c and another one in ftobjs.c?

I wouldn't call it neccesary, but I know the reason: David designed
ftraster.c so that it can also be used as a standalone unit, without
needing *any* other FT sources. While I appreciate the standalone-ness
(probably being its only user ;-), I wouldn't mind at all if I had to
include some additional source files to be able to use it. To have it in
the rasterizer module seems even sillier in the context of David writing a
new one every other day ;-). The implementation in ftraster should/could
disappear. It also appears that the *true* FT_Outline_Decompose should live
in ftoutln.c, and not in ftobjs.c. But apparently the entire ftoutln.c file
is optional, so that might explain why it *isn't* there:
FT_Outline_Decompose is not optional after all... Still, I would've made
ftoutln.c mandatory, and just use some #ifdefs around the optional bits.
Just my 2 cents.

Just