On Mar 8, 2009, at 4:51 PM, Junio C Hamano wrote:
I think this patch moves things in the right direction, but there are
other uses of "st_[cm]tim.tv_nsec" that do not use the
ST_[CM]TIME_NSEC
macro.
$ git grep -n -e 'st_[cm]tim\.' --cached -- '*.[ch]'
builtin-fetch-pack.c:810: || st.st_mtim.tv_nsec != mtime.nsec
git-compat-util.h:396:#define ST_CTIME_NSEC(st) ((unsigned int)
((st).st_ctim.tv_nsec))
git-compat-util.h:397:#define ST_MTIME_NSEC(st) ((unsigned int)
((st).st_mtim.tv_nsec))
read-cache.c:207: if (ce->ce_mtime.nsec != (unsigned int)st-
>st_mtim.tv_nsec)
read-cache.c:209: if (trust_ctime && ce->ce_ctime.nsec != (unsigned
int)st->st_ctim.tv_nsec)
Interesting. I couldn't use git-grep due to other problems, but
thought any other tim/timespec issues would have stopped my
compilation. Looking at the code, this is because everything other
than the #defines in git-compat-util.h are surrounded by USE_NSEC
which is not defined on my machine.
However, I noticed another breakage entirely. Namely, that USE_NSEC
is not defined anywhere. There's a comment in the Makefile that says
"define USE_NSEC below", but there is no code that checks for USE_NSEC
and sets the appropriate compiler switch. Trivial patch to follow.
Probably we should apply the following patch as a fix, and then
apply your
enhancement to support st_[cm]timespec systems?
Your patch looks sane to me.
~~ Brian
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
