I had my local sprint branch set up with:
  pull remote-repo http://code.haskell.org/darcs/sprint
in its _darcs/prefs/defaults

The default repo is kowey@code.haskell.org:/srv/code/darcs/sprint

But when I pulled, I kept getting ssh prompts.  If I remove the
_darcs/prefs/sources file, it does the right thing (pulling via ssh).

I think it should be fairly easy for somebody to write up a test case for this,
using only local repos.  Setting this to need-volunteer so that somebody can
write the test

Hi Eric, maybe I'm missing something here but :

> I had my local sprint branch set up with:
>
>   pull remote-repo http://code.haskell.org/darcs/sprint
> in its _darcs/prefs/defaults
>
> The default repo is kowey@code.haskell.org:/srv/code/darcs/sprint
>
> But when I pulled, I kept getting ssh prompts.  If I remove the
> _darcs/prefs/sources file, it does the right thing (pulling via ssh).
>
> Wouldn't be the right thing  here  pulling via http from http://code.haskell.org/darcs/sprint
? <http://code.haskell.org/darcs/sprint>  not via ssh.
When you said ssh prompts, was it prompting from
kowey@code.haskell.org:/srv/code/darcs/sprint
?

Although --remote-repo  "rewrites" the defaultrepo, it always contains the
defaultrepo also in the sources list, that's why it was trying to connect to
"kowey@..." (I'm just guessing it was to that one), and darcs always tries
the repos from the cache rather than the default or the given one.

Now thinking about it, I don't see how could probably be useful having the
defaultrepo also replied in the cache if it is remote, obviously the local
ones would be really useful, but having remotes repos in sources, force us
to establish a connection ( which I could have done with the given repo or
defaultrepo ), not really contributing that much to performance, so it
sounds to me like we shouldn't save ( by default) remote repositories in the
cache.

Cheers,

Adolfo

Attachments

• unnamed