Oswald Buddenhagen wrote: > feedback to them from the features. if the effect you're describing is > genuine, there is something seriously screwed up in the qmake internals; > you may get useful reports from valgrind or asan.
HAH! Would you believe I can no longer reproduce the error I copied earlier in this thread? I'm not going to swear I didn't make or dream it up but I almost feel like doing just that... It does suggest that there's something fishy in qmake's internals. It actually felt like looking at a Heisenbug, so to be certain I've unpacked to a fresh build dir, but same there. All of a sudden -no-pulseaudio is happily accepted, provided the qtmultimedia directory is present. (That bit is a tad annoying but I guess we can live with that. Would be nice though if `configure --help` scoured the available */config_help.txt files and printed the additional known options.} I have a trace from valgrind but there's nothing in there that suggests foul play anywhere. A couple of memory leaks (see below) but nothing surprising nor particularly shocking for something like qmake. ==56998== HEAP SUMMARY: ==56998== in use at exit: 656,533 bytes in 4,188 blocks ==56998== total heap usage: 1,047,206 allocs, 1,043,018 frees, 173,627,879 bytes allocated ==56998== ==56998== Searching for pointers to 4,188 not-freed blocks ==56998== Checked 20,540,472 bytes ==56998== ==56998== LEAK SUMMARY: ==56998== definitely lost: 4,240 bytes in 73 blocks ==56998== indirectly lost: 30,816 bytes in 84 blocks ==56998== possibly lost: 265,256 bytes in 70 blocks ==56998== still reachable: 295,509 bytes in 3,496 blocks ==56998== suppressed: 60,712 bytes in 465 blocks _______________________________________________ Development mailing list [email protected] http://lists.qt-project.org/mailman/listinfo/development
