windows head build and tests

[Claus Reinke]

>> the utf8_* are probably due to distclean not removing the *Lexer.hs
>> files, so the upgraded alex had no effect. conc068 is still affected by
>> that resolver warning. ThreadDelay001 reports:
> 
> Lexer.hs is not supposed to be removed by distclean, because it is part of 
> a source distribution.  distclean is supposed to clean the tree of 
> everything that shouldn't be in a source distribution, that does not 
> include autoconf, happy and alex-generated files.

yes, i was just noting that the strictest clean that works isn't 
sufficient for testing. in other words, we need a stricter clean
in addition to distclean, and need to use that for testing.

> Could you run ThreadDelay in all ways and tell me which ways fail? (make 
> stage=2 TEST=ThreadDelay001)

sure, output attached. all of them fail; as there's no 
ThreadDelay001.stdout, none of the output is expected?

$ ls testsuite/tests/ghc-regress/lib/Concurrent/
Chan001.exe.manifest  MVar001.stdout        QSemN001.exe.manifest      SampleVar001.stdout
Chan001.hs            Makefile              QSemN001.hs                ThreadDelay001.exe.manifest
Chan001.stdout        QSem001.exe.manifest  QSemN001.stdout            ThreadDelay001.exe.tix
MVar001.exe.manifest  QSem001.hs            SampleVar001.exe.manifest  ThreadDelay001.hs
MVar001.hs            QSem001.stdout        SampleVar001.hs            all.T

btw, all those manifest files should only be needed for vista,
right? and even there, the idea of having to think of 2 files
for each executable sounds like a small nightmare - is there
no way to integrate the .manifest into the .exe, or at least
not to produce those .manifests for non-vista?

claus
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ThreadDelay001.log
Type: application/octet-stream
Size: 23071 bytes
Desc: not available
Url : http://www.haskell.org/pipermail/cvs-ghc/attachments/20070914/eafc5032/ThreadDelay001-0001.obj