In article <01nD7.2790$[EMAIL PROTECTED]>, Michael Howes wrote:
>
> I have  a strange problem.
>
> I use BOTH the command line version of cvs and wincvs (I like some of the
>features in both). I've been using both for a while now just fine. What I
>often do is cvs update -d from the command line but do commits from wincvs.
>
> Afer the time change something is strange between the two (command line and
>wincvs).
>
> If I cvs update from the command line, then wincvs thinks ALL files are
>dirty (marking the icon red).

This is a problem that happens somewhere in the date comparison. WinCVS
uses the microsoft stat() function to retrieve the file timestamps from
the filesystem. I suspect that stat() is doing a bogus adjustment
for DST rather than retrieving universal values.

I have an unconfirmed hypothesis that changing the routine to use some
Win32 function to access the file time stamps could fix the problem.

One way to reproduce the problem is simply to check out the files using
WinCVS, and then flip your timezone setting.  Such a flip should have
no effect at all.

-- 
. .
 .  .
  .   .  <- Mysterious Powdery Substance
.
_______________________________________________
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs

Reply via email to