Why aren't mingw local patches in FSF tree?

Danny Smith dannysmith@clear.net.nz
Tue Mar 4 22:31:00 GMT 2003

I'm not subscribed to this list but I do scan. The subject line was
ughed at lately

Many of the cygwin/mingw local patches (in particular,  __fastcall   and
native bitfield support) have been committed to trunk but not to 3.3
branch.  The main exceptions are:

1) Allow exceptions and termination handlers to cross over exe/dll
boundaries.  This needs a lot of work to get rid of  many, many ugly
#if defined (__MINGW32__) || defined (__CYGWIN___) conditionals in
generic eh code.  There is also a copyright assignment issue.

2) Rationalize cygwin and mingw commonalities and provide new mechanism
for cygwin -mno-cygwin.  This patch will  be submitted soon (ie after
the cricket World Cup).

The sjlj vs dwarf  decision is in limbo until we can find a solution to
a w32api callback problem, but currently mingw binaries use the sjlj
mechanism, which is the default for FSF sources.

Also of interest may to you may be the ReactOs patchset for SEH handling
by GCC.  I think it was submitted (and ignored) but not sure.

It "would be nice" (tm) If the mingw project had more developers with
Ranjit's energy and perseverence.  Ranjit, surely you will accept CVS
write approval after all the flowers you have been receiving.


More information about the Java mailing list