This release has some fixes for some important compiler bugs, and a
couple of changes which will aid Phobos development (most notably,
nothrow becomes usable, and some severe issues relating to const are
fixed).
Unless we expect to have a fix for the killer bug 4854
"Regression(2.047, Mac 10.5 only) writefln Segmentation fault if no
globals", I think we should release a beta immediately.

In particular, I think std.datetime should wait for the next cycle.

If there are any known blockers, please list them now!

On 13 dec 2010, at 08:36, Don Clugston wrote:

> This release has some fixes for some important compiler bugs, and a
> couple of changes which will aid Phobos development (most notably,
> nothrow becomes usable, and some severe issues relating to const are
> fixed).
> Unless we expect to have a fix for the killer bug 4854
> "Regression(2.047, Mac 10.5 only) writefln Segmentation fault if no
> globals", I think we should release a beta immediately.

I was hoping that someone could have a look at the patch I've attached to that ticket. That patch fixes the reported error (about the __tls_get_addr function) but it still segfaults, now in another place. I don't know if I introduced the n new segfault, if it's another bug or the solution we though would fix the problem wasn't enough.

> In particular, I think std.datetime should wait for the next cycle.
> 
> If there are any known blockers, please list them now!
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos

-- 
/Jacob Carlborg

Because the bug that should be repaired was discovered, I report it.

dmd(trunk r804) crashes by the compilation of Phobos with following flags:

DFLAGS=-unittest -d -g -debug

See also(Japanese):
  http://twitter.com/#!/Rayerd/status/13261697154293760
  http://hibari.2ch.net/test/read.cgi/tech/1281826764/885

There may be the last bug report and connection, but I do not understand
it in detail.
  http://d.puremagic.com/issues/show_bug.cgi?id=5343

--
SHOO

(2010/12/13 16:36), Don Clugston wrote:
> This release has some fixes for some important compiler bugs, and a
> couple of changes which will aid Phobos development (most notably,
> nothrow becomes usable, and some severe issues relating to const are
> fixed).
> Unless we expect to have a fix for the killer bug 4854
> "Regression(2.047, Mac 10.5 only) writefln Segmentation fault if no
> globals", I think we should release a beta immediately.
>
> In particular, I think std.datetime should wait for the next cycle.
>
> If there are any known blockers, please list them now!

On 13 December 2010 19:05, SHOO <zan77137 at nifty.com> wrote:
> Because the bug that should be repaired was discovered, I report it.
>
> dmd(trunk r804) crashes by the compilation of Phobos with following flags:
>
> DFLAGS=-unittest -d -g -debug

I can confirm that on Windows. But, it's not a regression. Using the
DMD2.050 compiler and the current phobos,  it does exactly the same
thing. I believe it's the same capacity bug we've seen before.
Is it the same as bug 4904 / bug 3372 ?
http://lists.puremagic.com/pipermail/phobos/2010-September/002741.html