I'm trying to finish 0.30 issues, and #1445 talks about 'svn export'
and 'svn checkout' setting last-commit-timestamps on files.
I don't want to repeat the huge conversation we had on this list about
how timestamps should work; rather, I just want to affirm the
short-term strategy for 1.0. Here's what I remember:
* Everyone agreed that 'svn export' should set last-commit-times on
files. That feature was already committed a while back; we're
simply waiting for httpd-2.0.48 to be released (to get the new APR
timestamp-setting API).
* At least one person had a proposal to write a fancy system for
storing and retrieving mtimes; this was deemed too complex for
1.0, and put off for discussion as a post-1.0 feature.
* The issue says that we want 'svn checkout' to set
last-commit-times, but'svn update' to continue setting now-times,
as it already does (to ensure that 'make' works correctly).
Greg Hudson tells me (in IRC) that the last point is an incorrect
memory of a consensus, that we never agreed to make 'svn checkout' set
commit-times. He and Philip didn't want their working copies to be a
weird mixture of commit-times and now-times.
However, Greg did seem amenable to having a run-time option (in
~/.subversion/config) that causes checkouts to either always use
now-times or always use commit-times.
I've almost finished a patch which makes 'svn checkout' set
commit-times, but not 'svn update'. I can easily add a config option
to toggle the behavior of 'svn checkout'. Is that acceptable to most
developers?
In summary, once I commit this change, the behavior will be:
* 'svn export' always sets commit-times
* 'svn update' always sets now-times
* 'svn checkout' can behave either way, based on config variable.
Speak now, good developers.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Sep 5 00:56:26 2003