Re: Externals handling...

Joerg Hessdoerfer <Joerg.Hessdoerfer@sea-gmbh.com> writes:

> They worked fine on checkout, and I had a nice source tree. The problem came
> later, when I began coding on the tree and committed changes. I noticed that
> the 'external' directories where not recursively committed, too, as I had
> expected.
>
> Looking in the code, there seems to be no handling of this. Is this correct?
> If so, why? I didn't find any mention of the intended behaviour in the book,
> and no filed issue of this.

That's because svn:externals is a half-implemented feature, IMO. I
think it needs to be finished by 1.0 for people to take it seriously.

It works for checkout and update, but not commits or status. The
status bug you mention already is already filed as a beta issue:

  http://subversion.tigris.org/issues/show_bug.cgi?id=959

The commit case is a thornier issue. On the one hand, our
commit-crawler already knows how to gather changed files and group
them by repository. So once everyone has repository UUIDs sitting in
their entries files, there's no theoretical reason commits couldn't
descend into 'externals' directories. If the externals are from other
repositories, then you may end up with several independent commits,
which is fine and expected.

But kfogel has worries about this, and I don't understand what they
are yet.

Perhaps we should file a companion issue to #959, about commits
following svn:externals?

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Mar 5 16:02:19 2003