Re: copy_revs ?

Barrie Slaymaker <[email protected]> Tue, 16 Dec 2003 07:14:03 -0500
Newsgroups gmane.comp.version-control.revml
Message-ID <[email protected]>
On Tue, Dec 16, 2003 at 12:15:38PM +0100, Timothee Besset wrote:
> On Mon, 15 Dec 2003 23:05:02 +0100
> Timothee Besset <[email protected]> wrote:
> > 
> > This all seems to evolve around copy_revs. It looks like the VSS and the
> > P4 sources are doing the job in copy_revs .. which I don't see mentioned
> > in the documentation. Actually, there is one spot in the source where it
> > says copy_revs is deprecated..
> > 
> 
> Silly me. You *did* mention copy_revs. I'll look at CVS or VSS source
> driver to learn about the data it should return.

It is supposed to scan the metadata and then send it to the destination.
Generally, this is done by calling queue_rev for each VCP::Rev created.
Because most repositories emit log from most recent to oldest, you can
get at the last rev that have been queued and the last rev in each
filebranch that has been queued in order to set its previous_id().  See
last_rev() and last_rev_for_filebranch().

It's also deprecated as in I haven't had time to replace it, but I need
to.  If you put all your metadata scanning code in a subroutine called
scan_metadata(), it'll be forward compatible.  I'll change the existing
sources' APIs to that soon.

> I used my devnet account to ask the NxN people about their perl API, will
> see what they have to say about it.
> 
> Also asked for a guest account on public.perforce.com

Excellent on all accounts!  Thanks.

- Barrie