Re: fucked in gnus
Ben Wing <ben(a)666.com> writes:
the tricky thing with the x code is that the selection data can be
one of three types: 8, 16, or 32 bit data; under windows, we only
have 8-bit data. 16 and 32-bit data automatically gets byte-swapped
as necessary to deal with endian-ness issues, necessary because
clients on different architectures might request data from the same
X server. the c code tries to support this by allowing conses of
16-bit values [because xemacs lisp integers are less than 32 bits]
Could we use doubles instead of the silly conses? It seems more
obvious and more reliable, at least in the [0, 2**32) range.