info-cvs
[Top][All Lists]
Advanced

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

RE: File date and timestamp


From: Flagg, David
Subject: RE: File date and timestamp
Date: Tue, 29 Jun 2004 09:38:56 -0400

Thanks for the info.

The reason the timestamp is important is because the file is part of an
external delivery.  the customer has previous deliveries and flags files
whose timestamps have changed.  As a result, they wonder why the
timestamp changed on this binary file, yet we didn't inform them of any
changes with that particular capability.

-----Original Message-----
From: Jim.Hyslop [mailto:address@hidden
Sent: Monday, June 28, 2004 11:45 AM
To: Flagg, David; address@hidden
Subject: RE: File date and timestamp


Flagg, David wrote:
> This is a re-post.  This seems like it could possibly be a 
> common problem someone might have.  Is there a more 
> appropriate forum somewhere else?  Thanks.
Hmmm.. I thought I saw a response to your original message from someone
else
already, but I don't see it in my 'trash' or in the archive.

> >  -----Original Message-----
> I have a binary file that is in my repository, libtemp.a, 
> which has the following timestamp in cvsroot (i.e., timestamp 
> in the Unix filesystem): Jun 18 09:42.  Apparently, this is 
> the time when I last tagged the file.
That would make sense; applying a tag modifies the RCS file.

> When I check out the file (with WinCVS, by the way), the 
> date on the file is (as viewed by Win 2K): 5/3/2004 6:47 PM.  
> This is the date/time when I last committed the file.  This 
> also agrees with the timestamp in the Entries file.
> > 
> When I committed the file, the timestamp on the file was 
> 12/24/2003 2:23 PM.  It is important that this be the 
> timestamp that is associated with the binary file, as it is 
> pulled in from another source.  When I check this file out, I 
> want the timestamp to continue to be 12/24/2003 2:23 PM.
Timestamps are an interesting issue. The problem is, how do you decide
which
timestamp to use on the checked-out file?

Even if you succeed, there is the issue of updating to a specific tag.
If
you issue the command 'cvs update -r1.1' the timestamp of the file will
become the current date and time. While this may seem strange and
unintuitive at first, you must consider that CVS also interacts with
other
systems - such as build systems. If CVS used the original modification
timestamp or the check-in timestamp, the dependency system would not
flag
the dependencies as out-of-date.

> (1) Is there a way that I should have committed this file 
> such that I can get the behavior I want?
I believe that the rcs command allows options to specify the timestamp
to
apply to the file. Or, you can hack the CVS source to get the behaviour
you
want.

This introduces another issue, though - 'cvs log' and anything else that
uses the date option works on the date in the RCS file. You could
introduce
subtle, unexpected behaviour changes with this modification.

> (2) How can I correct the situation for this particular file?
You could manually edit the RCS file.

It sounds to me, though, like you need a configuration management system
that is aware of things beyond the current scope. Why is it so important
to
maintain the exact timestamp of the binary file?

-- 
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]