[MLton] cygwoes

[Stephen Weeks]

> Any word on the cygwin problem? 

I made a little progress.  The way I usually debug stuff on Cygwin is
by cross compiling the regressions.  I don't see the signal B error
that you do.  Rather, I see 

  VirtualFree release failed

Hopefully they are related, although who knows.

MLton has two different ways of grabbing memory from the OS on Cygwin:
Virtual{Alloc,Free} or (the Cygwin-emulated) mmap.  In the past,
Virtual{Alloc,Free} has been more stable and gave access to more
memory, so we made it the default.  Perhaps that is no longer true.
We have a runtime flag, use-mmap, which forces the runtime to use
mmap.  I am able to get all of our regressions to pass on Cygwin,
provided that I run them with @MLton use-mmap --.

As a stopgap, you could cross-compile your program with 

  -runtime use-mmap

to force the executable to use mmap.  You might even try a full
bootstrap with use-mmap, possibly even tweaking GC_init in gc.c to set

  MLton_Platform_CygwinUseMmap = TRUE;

You could also try experiments with Virtual{Alloc,Free} by compiling
the runtime with DEBUG_MEM = TRUE in platform.h.  There are a couple
of other DEBUG_* flags in gc.c that might shed some light:
DEBUG_RESIZING comes to mind.

> I'm willing to try to fix this if someone can give me a tip on what it
> might be. Or is the release right around the corner?

The release has been right around the corner for over a month, the
delay being due only to the Cygwin problem.  It would be great if you
could fix it.  MLton might even be able to release!