[phobos] Is the build broken?

[Dmitry Olshansky]

On 13.11.2011 9:35, Jonathan M Davis wrote:
> On Saturday, November 12, 2011 19:21:41 Andrei Alexandrescu wrote:
>> I just updated everything (dmd, druntime, phobos) and make unittest
>> fails for phobos on OSX/32 with the error below. Is it reproducible?
>>
>> core.exception.AssertError at std/regex.d(7138): bmatch(R,RegEx) if
>> (is(RegEx == Regex!(BasicElementOf!(R)))): mismatch pattern #204:
>> \b\w+\b expected: abde4 vs de
>> ----------------
>> 5   regex                               0x001086b6 _d_assert_msg + 26
>> 6   regex                               0x0008a25d void
>> std.regex.__unittest7().void
>> __T9run_testsS173std5regex6bmatchZ.run_tests() + 82613
>> 7   regex                               0x00005fd5 void
>> std.regex.__unittest7() + 13
>> 8   regex                               0x00001fb1 void
>> std.regex.__modtest() + 21
>> 9   regex                               0x000ff8b1 extern (C) bool
>> core.runtime.runModuleUnitTests().int __foreachbody265(ref
>> object.ModuleInfo*) + 45
>> 10  regex                               0x000fa617 int
>> object.ModuleInfo.opApply(scope int delegate(ref object.ModuleInfo*)) + 79
>> 11  regex                               0x000ff7a2 runModuleUnitTests + 134
>> 12  regex                               0x00108e2f extern (C) int
>> rt.dmain2.main(int, char**).void runAll() + 43
>> 13  regex                               0x001089a1 extern (C) int
>> rt.dmain2.main(int, char**).void tryExec(scope void delegate()) + 29
>> 14  regex                               0x0010893b main + 179
>> 15  regex                               0x00001f89 start + 53
>> 16  ???                                 0x00000001 0x0 + 1
> The OSX build has been broken ever since the new std.regex was merged in:
> http://d.puremagic.com/test-results/
>
> You can read the discussion about it on the pull request:
> https://github.com/D-Programming-Language/phobos/pull/310  It sounds like it's
> a compiler bug, but I don't think that they've quite figured it out yet.
>
>
Sorry about that, but so far I failed to get a grip on what's actually 
wrong. For what it's worth I expected it to be completely platform 
agnostic. There is nothing special except malloc|free|memchr here.
If anyone on Mac OS X can be bothered to try it with earlier versions of 
compiler, then  please do.

-- 
Dmitry Olshansky