Description of problem: After yesterday's build of glibc-2.11.90-7 in dist-f13 the ghc package no longer builds in koji. :-( eg http://koji.fedoraproject.org/koji/taskinfo?taskID=1916172 Version-Release number of selected component (if applicable): glibc-2.11.90-7 How reproducible: every time Steps to Reproduce: 1. try to build ghc/devel in rawhide with glibc-2.11.90-7 2. revert to glibc-2.11.90-4 and retry Actual results: 1. build fails like above buildlog: http://koji.fedoraproject.org/koji/getfile?taskID=1916175&name=build.log&offset=-2000 2. build succeeds Expected results: 1. build to succeed Additional info: Perhaps it is better to untag -7 ?
Please attach config.log.
Created attachment 383215 [details] config.log This is the config.log just before the error if it helps. Note that HsBase.h is part of the ghc source and not generated. HsBaseConfig.hs is unchanged.
Should be easy to reproduce locally: $ fedora-cvs ghc $ cd ghc/devel $ rpm -q glibc-headers $ make local
*** Bug 554679 has been marked as a duplicate of this bug. ***
When will be the build system updated, or is it already?
Presumably. I see glibc-2.11.90-8: http://koji.fedoraproject.org/koji/buildinfo?buildID=150617
Thanks Resubmitted ghc job: http://koji.fedoraproject.org/koji/taskinfo?taskID=1917588
Thanks, evolution-data-server was built, though it failed on evolution itself now, with an error: > e-mktemp.c: In function 'e_mktemp': > e-mktemp.c:182: error: implicit declaration of function 'mktemp' Why is that? I thought the online doc [1] will mention its dropping or deprecation or something, but it isn't. Thus I consider it another bug in a new version. Reopening. [1] http://www.kernel.org/doc/man-pages/online/pages/man3/mktemp.3.html
POSIX.1-2008 removes the specification of mktemp().
Sure, the question is whether we need to be this pedantic even when no strict POSIX 2008 conformance is requested. In particular, couldn't the condition be #if defined __USE_MISC || (defined __USE_XOPEN_EXTENDED && !defined __USE_XOPEN2K) ? (__USE_GNU implies __USE_MISC I think). So, for strict -D_POSIX_C_SOURCE=200809L mktemp wouldn't be defined, but for no feature test macros at all it would?
*** Bug 555060 has been marked as a duplicate of this bug. ***
(In reply to comment #10) > Sure, the question is whether we need to be this pedantic even when no strict > POSIX 2008 conformance is requested.... I do not care that much. My concern was rather whether it was a typo, like with the above, or intentional change. If it's intentional, then I'm fine with that. And I'm going to rewrite necessary parts with mktemp in evolution anyway.
I'm confused why bug 555060 having to do with 'struct wait' is flagged as a dup of this one which seems to be about mktemp?
(In reply to comment #13) > I'm confused why bug 555060 having to do with 'struct wait' is flagged as a dup > of this one which seems to be about mktemp? No, mktemp is different topic for evo. Grep attachment in comment 2 for __WAIT_STATUS.