[prev in list] [next in list] [prev in thread] [next in thread] 

List:       dcms-dev
Subject:    Re: [opencm-dev] cm update file time behaviour
From:       "Jonathan S. Shapiro" <shap () eros-os ! org>
Date:       2003-05-28 14:12:42
[Download RAW message or body]

Michael:

Timestamps are a mess for reasons that have nothing to do with OpenCM.
OpenCM merely reveals the problem.

First, your clock and mine may not be in sync. What time is it? We
observe cases in our lab occasionally where the time drift is enough to
cause weirdness if the update command preserves timestamps here is how:

   1. I edit and commit at my time T0
   2. You later edit at your time T1 < T0
   3. You update. If update respects time stamps, then file reverts
      to time T0. Make breaks.

Generally, an update should yield a file that is newer than anything
currently in your workspace. This guarantees that make works most of the
time. The only case in which this is a problem is when people screw up
their build system by mis-using something like automake.

In that case, the thing to do is run 'cm revert', which restores things
to *exactly* the state they were in at checkout (i.e. including
timestamps).

You will note that 'cm revert' output advises that you immediately do a
'make clean', because builds following a cm revert can have
unpredictable consequences.

The long and short of it is that it's not possible to get timestamps
right unless people sync their clocks, and real machines don't do that.
The current behavior is the best compromise we could find.

shap

_______________________________________________
opencm-dev mailing list
opencm-dev@smtp.opencm.org
http://www.opencm.org/mailman/listinfo/opencm-dev
[prev in list] [next in list] [prev in thread] [next in thread]