>> >>> Once you can push a button to determine if you have the correct
> set
>> >> of
>> >>> files, the desire for accurate timestamps diminishes greatly.
>> >> And svn?
>> >
>> >
>> > Eh?
>>
>> I was under the impression that if the timestamps and perhaps length
>> matched on the working/pristine copies, svn assumed the contents were
>> identical. Perhaps this isn't correct.
It's correct. And for many projects the only way to get reasonable
performance (by not scanning the complete file).
> I'm 99% sure it's using checksums, with a 1% chance of black magic.
>
> Change one character in a file. Run 'svn status' and see that it is
> flagged as 'M'odified. Change the character back to its original value.
> Run 'svn status' and you'll see that it no longer appears as modified.
> It is completely ignoring the timestamp (which has changed twice) and
> the file size hasn't changed, so it must be using checksums.
Umm, well, I hate to spoil your fun, but: actually, it's using
checksums, but only if either the timestamp or the file size have
changed...
Bye,
Erik.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: users-help_at_subversion.tigris.org
Received on 2008-06-10 22:57:56 CEST