On Fri, 23 Sep 2011 18:28:21 +0200, dsimcha <dsim...@yahoo.com> wrote:
== Quote from Derek (ddparn...@bigpond.com)'s article
I've been out of the 'loop' with D for quite awhile now so I haven't
been
keeping up with current developments.
I will have a need for a decent 64-bit compiled language soon and I was
wondering how close D is away from this.
DMD now has a 64-bit codegen that works quite well (well enough for
serious use,
no showstopper bugs) on Linux. This has been around since 2.052,
released last
February. It also supports 64-bit on FreeBSD, though I don't know how
well that
works because I've never tried it. On Windows and Mac OS, 64-bit isn't
supported.
On Windows this is because Optlink (the linker DMD uses on Windows)
doesn't work
with 64-bit objects. On Mac OS, I don't know what the bottleneck is.
GDC has also been improving rapidly and is, as far as I can tell, one
showstopper
bug
(https://bitbucket.org/goshawk/gdc/issue/226/gc-from-spawned-threads-segfaults-on-64)
away from being usable for serious work on Linux x64. It already seems
usable in
32-bit mode. There's been some work towards porting it to Windows, but
that's
somewhat behind the curve. The Windows version only supports an ancient
front-end
release and 64-bit Windows support is buggy.
LDC is a release behind DMD and GDC, so I haven't been following it.
It's
supposed to support 64-bit Linux, though.
FreeBSD is pretty fine since we've figured out how to correctly add the
static root to the garbage collector.
DMD currently has one known nasty bug for 64-bit codegen though it happens
rarely (plenty of function arguments increase the probability).
http://d.puremagic.com/issues/show_bug.cgi?id=6189.
Another one involves extern(C) functions.
http://d.puremagic.com/issues/show_bug.cgi?id=5570.
You still should always take the possibility of a codegen bug into account
but I haven't run into one for months.
martin
