On Sun, Nov 14, 2010 at 10:05:18AM +1000, Allan McRae wrote:
On 14/11/10 09:03, Nezmer wrote:
On Sun, Nov 14, 2010 at 02:40:10AM +1000, Allan McRae wrote:
On 13/11/10 23:16, Nezmer wrote:
On Sat, Nov 13, 2010 at 10:48:33PM +1000, Allan McRae wrote:
Hi Nezmer,
Can you run test the updated autoconf stuff on my diskspace branch? Specifically, I would like to see the output from configure to do with getmntinfo to see if it detects the correct version to use.
The actual building will still be broken...
Thanks, Allan
checking for getmntinfo... yes checking parameter style for getmntinfo... statfs
Thanks. I have installed a FreeBSD VM and the diskspace checking is now working... sort of (the calculation of blocks used has issues exposed due to types being used and using ceil() requires a -lm added). But the basics are fine.
I forgot to mention that I used "LIBS='-lm'". My bad.
As an aside, it was quite annoying to find that even the current pacman does not compile on FreeBSD but no bug has ever been reported here. So I had to hack around those issues before testing. (for those that are interested, FreeBSD does its own thing with time_t definitions, so the formatting for printf is not quite right...).
What? pacman's C code *never* failed to compile on FreeBSD since I started using it (in January I think). I use gcc45 though so my environment is not exactly the same as the official one. Maybe GCC's include-fixed headers are doing magic.
I can share ISOs, images If anyone needs a nice working FreeBSD/pacman environment.
You do not compile with --enable-debug which turns on some more strict error checking (turns warnings into errors).
Allan
With gcc45, I get: (1) --enable-debug: success (2) CFLAGS+=" -Werror": error util.c:883:15: error: static declaration of ‘strnlen’ follows non-static declaration /usr/include/string.h:105:9: note: previous declaration of ‘strnlen’ was here gmake[2]: *** [util.o] Error 1 gmake[2]: Leaving directory `<pacman-tree>/src/pacman' (3) CFLAGS+=" -Werror -DHAVE_STRNDUP": success