Hello, On Mon, Mar 07, 2011 at 11:55:48PM +0530, Joachim Breitner wrote:
Hi, Am Montag, den 07.03.2011, 15:02 +0530 schrieb Joachim Breitner:Hmm, is this useful? I don’t see a real pattern. We see failures with both versions of libc, with all four different versions of ghc available, across different architectures...I now gave ghc or haskell-transformers back on some of the failing architectures. If some of them pass through we can hopefully identify the cause. Maybe different buildd configuration. Strangely, the armel build of 7.0.2-1 went through, while 7.0.1-1 did not. Build logs looks very similar. Help! :-)
Well, it looks like good news! I just compared libraries/unix/configure.ac between 7.0.1 and 7.0.2 and 7.0.2 does indeed look like it contains a fix for this issue. The configure script now uses AC_SEARCH_LIBS to find sem_close and set up the linking appropriately. I don't understand why the build went through — thought it'd need my bootstrapping optl-pthread hacks as the previous version was mislinked — but it seems like it did. :-) Try building *and running* something on one of the failing arches. I suspect it will work now and we can close this bug. Cheers, Iain
Attachment:
signature.asc
Description: Digital signature