The attached packaging is a first attempt at building mysql-unified-5.0.45 with clang. The resulting build doesn't completely pass InfoTest but neither does the current build against llvm-gcc.
Proposed packaging passes 'fink -m' on 10.7 and 10.8 fink.
Passes here as well and it should be checked in to the 10.7 tree tomorrow if Todai doesn't respond (the usual 3 day wait).
One thing about the TestScript, with the removal of the UID test, it assumes that the default --build-as-nobody flag hasn't been overridden by the user. The UID test should probably go back in, but either fail tests with a message if UID equals 0 (or just bypass the tests). Forcing a BuildDepends on fink-0.33 only means that --build-as-nobody is the default, but doesn't prevent being overridden by the user.
if [ $UID = 0 ]; then
echo "The mysql tests can not be run as root."
make -j1 test-force || exit 2
Committed to 10.7 tree. The same packaging builds against an llvm32 svn package's clang compilers on x86_64 10.6 fink using Xcode 4.2's cctools and passes all of InfoTest. The current llvm31 clang compilers are insufficient and fail with compilation errors of...
longlong2str_asm.c:29:7: error: expected parameter declarator
char *longlong2str(longlong val,char *dst,int radix)
I decline assignment of this.
I don't know or care a damn thing about mysql. Nor do I have time right now to deal with random packages that I don't maintain.
I only assigned you because the cvs logs showed you did the last two commits to mysql-unified in the 10.7 tree.
Note that the new 10.7 packaging might need an xcode dependency. If we get any reports of failures for building with the clang from older Xcode's than 4.5, one can be added. However I suspect the new xcode dependency in gcc47 will have shifted the user base to Xcode 4.5 for the 10.7 tree.
Closed as noted by submitter.
Log in to post a comment.
Sign up for the SourceForge newsletter:
You seem to have CSS turned off.
Please don't fill out this field.