If this is the source of the problem, then moving os_subtype to w32.h
might fix it. But first, we must understand why this problem has
occurred. Is the new implementation of w32select.c using a C library
function that internally mallocs memory using the C library malloc
instead of Emacs' one? Or is it the calls to GlobalAlloc that is
causing this (in which case there is no conflict).
Tak Ota <address@hidden> writes:
This patch introduces the following linking problem.
Thanks for trying it out ;-)
libc.lib(heapinit.obj) : error LNK2005: __heap_init already defined in
The patch shouldn't have touched those two. The only connection is
that w32select.c uses "os_subtype", which I is defined in w32heap.c.
For that w32select.c includes "w32heap.h", that is all the conneciton