On 19.07.2012 16:08, Eric Blake wrote:
On 07/19/2012 01:13 AM, Hendrik Schwartke wrote:
I reconsidered the way timestamps are represented. I think that an event at 100.91 happened before 100.200 is misleading. Yep, definite bug - you have to zero-pad the subsecond resolution, and also consider whether to strip trailing zeros.
So I changed that. Furthermore I would appreciate if the timestamps are available in 0.10.0, so I splitted the patch. The first patch doesn't use stat-time and the second patches the first and does use stat-time. Not necessary to do two implementations. stat-time has now been relicensed in gnulib, so all we are waiting on now is for updated automake to hit Fedora 17 (it has already hit rawhide): https://bugzilla.redhat.com/show_bug.cgi?id=838660 Wow, that was fast. Many thanks for your help.
Given the security bug in current automake, I will actively protest releasing the next version of libvirt until fixed automake is available.
Don't worry - I plan to include this feature in the next libvirt build; just a few more days of waiting.