On Thu, 18 Jun 2009, NightStrike wrote:
So it seems that boehm-gc is in the exact state as libffi.
This is yet another example of why we shouldn't duplicate sources...
Hans, would you be willing to bring boehm-gc up to speed so that we
can start getting it to work for Win64? Without this, we obviously
cannot add gcj to our list of supported compilers.
I think win64 in the upstream sources is in reasonable shape.
I generally haven't been doing a great job in keeping up with patches to
the upstream source, because I often just haven't had time. But I think
I have mostly kept up with patches originating from gcc. I'm not sure the
upstream sources are missing much.
What has been a problem is that while the 6.8 -> 7.0 changes cleaned
up the code substantially, and a lot of contributed patches since then
have done a lot more of that, that step also introduced a fair amount of
instability. I think we're largely over that now. But 7.1 is fairly old,
so the merge would either have to use the CVS trunk or wait for 7.2 (which
I'm trying to find time to get out real soon now).
On the positive side, 7.1+ also have a bunch of substantial functional
improvements.
One additional minor complication is that I have a bunch of changes that
haven't yet made it into the upstream CVS repository, that provide initial
support for the C++0x GC support API, but touch some of the core GC
datastructures. Maybe those should wait for a leter merge anyway?
Hans
