info-cvs
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: CVS Server times


From: Jim.Hyslop
Subject: RE: CVS Server times
Date: Fri, 25 Feb 2005 09:35:01 -0500

David A. Bartmess wrote:
> But if a local PC was used to work on files, having a current 
> time at checkin 
> of 11:30 MDT, and the server at time of checkin is 13:35 EDT, 
> this could 
> seriously screw with any automation that relies on the 
> datetime stamp to know 
> if a build has already included that change (i.e., cruisecontrol).
> 
> Just FYI...

So, are you saying that cruisecontrol relies only on the timestamp of the
repository file to determine what has changed? This seems a little crude -
if I apply a tag to a file, that will change its timestamp, but not the
contents. No rebuild should be required.

Anyway, let me elaborate on your scenario slightly to see if I understand
you. Client-server protocol is not being used (if a client/server protocol
is being used, then all timestamps would be relative to the server's clock,
and the client's clock is irrelevant). Clients mount a server's directory
locally, and the mounting system sets the timestamp on the file according to
the client's system clock. Cruisecontrol looks only for newer files (not
just files whose timestamp changes).

Sounds like the answer is: if you're using cruisecontrol, or if you have
workstations in multiple time zones, use client/server protocols instead of
mounting the repository locally. This would be good security advice in any
case - I really don't like users having direct access to the repository.

-- 
Jim Hyslop
Senior Software Designer
Leitch Technology International Inc. ( http://www.leitch.com )
Columnist, C/C++ Users Journal ( http://www.cuj.com/experts )




reply via email to

[Prev in Thread] Current Thread [Next in Thread]