On 15-03-14 11:14, Samuel Thibault wrote: > Paul Gevers, le Sat 15 Mar 2014 11:01:14 +0100, a écrit : >> On 15-03-14 10:43, Samuel Thibault wrote: >>> Paul Gevers, le Sat 15 Mar 2014 10:32:54 +0100, a écrit : >>>> If yes, should be change the way we build or do we need a different >>>> solution for this problem? >>> >>> I'd say we can simply set LC_ALL to C. >> >> That doesn't solve the problem of the timestamp being different in the >> git checkout than in the tarball. > > Aren't we regenerating the files? To recap: the problem that we need to solve is the timestamp in the man pages that is different on different architectures if we don't take care to force it to something stable. My original solution was to take the timestamp of the source file, but apparently that doesn't work for some work flows (like the one of Mario), as the timestamp in the file system might not match the timestamp in the tar ball. Paul
Attachment:
signature.asc
Description: OpenPGP digital signature