On 05/24/2011 12:33 PM, Matthias Bolte wrote:
2011/5/24 Eric Blake <eblake@redhat.com>:
We don't use the gnulib vsnprintf replacement, which means that on mingw, vsnprintf doesn't support %sn.
I meant %zn, not %sn; and I'll also mention %llu which is another one mingw doesn't natively know.
But as it turns out, VIR_GET_VAR_STR was a rather inefficient reimplementation of virVasprintf.
* src/util/logging.c (VIR_GET_VAR_STR): Drop. (virLogMessage): Inline a simpler version here. * src/util/virterror.c (VIR_GET_VAR_STR, virRaiseErrorFull): Likewise. Reported by Matthias Bolte. --- src/util/logging.c | 45 ++++++--------------------------------------- src/util/virterror.c | 44 +++++--------------------------------------- 2 files changed, 11 insertions(+), 78 deletions(-)
ACK.
Thanks; pushed.
This actually fixed a debug output that confused me. Before this patch the output claimed
19:14:15.365: 2140: debug : virEventPollCalculateTimeout:345 : Timeout at 0 due in 0 ms
But this 0 0 case is impossible from the logic in the code. After the patch it is reporting correctly
20:15:25.157: 2140: debug : virEventPollCalculateTimeout:345 : Timeout at 0 due in -1 ms
Cool. That probably means that mingw treated %llu as %lu and only parsed 32-bits of a 64-bit stack entity, leaving the %d for the remaining 32 bits; whereas post-patch you get the right argument mapping. -- Eric Blake eblake@redhat.com +1-801-349-2682 Libvirt virtualization library http://libvirt.org