On Sep 5, 2006, at 08:45, Gavin Lambert wrote:
> I'm just curious if anyone knows how well Subversion handles unordered
> files (eg. XML files where the order of 'records' is unimportant, and
> tend to be written out in fairly random fashion, meaning that every
> time
> the file is modified & committed the diff consists of nearly every
> line
> in weird patterns).
>
> I'm specifically thinking of Visual Studio's .resx files at the
> moment;
> I suspect that VS loads them into a hashtable (randomising their
> order)
> and then writes them back out, leading to large diffs.
>
> Will this result in the entire file contents (or more, with
> bookkeeping
> overhead) being duplicated for each revision it's changed in? Or
> is it
> smarter than that?
I don't think it's smarter than that. I don't think Subversion has
any choice here other than to include every changed line in the
revision.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Tue Sep 5 11:44:00 2006