Re: stable git commit ids over time

Michael Haggerty <[email protected]>
Newsgroups gmane.comp.version-control.subversion.cvs2svn.user
Message-ID <[email protected]>
On 10/08/2010 11:57 AM, James Blackburn wrote:
> On 8 October 2010 03:18, Daniel Hagerty <[email protected]
> <mailto:[email protected]>> wrote:
> 
>     > Given that the difference is only one second, it is probably
>     related to
>      > the code that tries to ensure that all timestamps are distinct.
>      This is
>      > not really a requirement for git, so if you want to try again without
>      > that code, apply the attached patch (vs. trunk).
> 
>        To follow up, this code seemed to do the trick immediately.  I've
>     been running this code for a week now and haven't seen any further
>     instances of commit id churn.
> 
> 
> This is very neat, and a major feature that git cvsimport holds over
> cvs2git...

Agreed.

> Will this patch make it into a cvs2git release?

The effect of this patch is that timestamps are no longer guaranteed to
be unique, though they are still guaranteed not to be in backwards order.

* 2git: I'm pretty sure that the change is fine.

* 2bzr, 2hg: I would be surprised if they mind the change, but I'm not
certain.

* 2svn: I always assumed that the 1-second spacer is required by
Subversion for some reason.  I'm pretty sure it's been in the cvs2svn
code since before I started working on the project.  But now that I
think about it, it is quite likely that the gap was added to help the
old pre-2.0 cvs2svn algorithm figure out changesets, not because
Subversion requires it.  Does anybody know about this?  So I'm guessing
that the one-second spacer could be taken out of the 2svn conversion, too.


But I don't think we should use this change to claim that cvs2git
supports incremental conversions.  It only works by coincidence, not by
design.  (Not that "git cvsimport" works by design, but we have higher
standards over here :-)   .)

Michael

-- 
Michael Haggerty
[email protected]
http://softwareswirl.blogspot.com/

------------------------------------------------------
http://cvs2svn.tigris.org/ds/viewMessage.do?dsForumId=1670&dsMessageId=2669282

To unsubscribe from this discussion, e-mail: [[email protected]].
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.