John Peacock wrote:
> Barry Scott wrote:
>
>>> If you come up with a brilliant way to determine and populate that
>>> copyto info, you just let me know.
>>>
>> So its a good idea, but its hard to implement?
>>
>> How do you populate the copy_from list? Isn't the copy_to list the same
>> problem? On a copy you have to record the copy_to info on the src
>> and copy_from in the dest.
>
>
> I think part of the issue is that copy_from is a one-to-one
> relationship (a file can only be copied from one file). copy_to,
> however, is potentially one-to-many, which complicates the storage on
> the source (since you have to track each and every copy_to). A single
> file could potentially be used as the source for every other file in
> the repository, so you'd have to have an way of storing a limitless
> number of copy_to pointers in a way that wouldn't be slow to add new
> values.
That's not so hard to do if you use indexes.
> It's also something that would mean the repository format would change
> completely,
Not "completely", but incompatibly. Adding forward pointers to the
schema is on my list for 2.0.
> plus I don't know how difficult it would be to recreate the copy_to
> list during conversion. It would require at least one extra pass
> through the data to read the copy_from's and intuit the copy_to's.
Now that's simple. If you can reproduce every copy during a dump/load,
then obviously you can populate all copy-from and copy-to data.
-- Brane
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat May 1 21:58:23 2004