On Tue, Sep 09, 2008 at 12:33:46PM +0100, Eric Kow wrote:
> So previously, we said
>
>    Eric only considers code comments, documentation and test suite
>    modifications to be obvious
>
> I think this sort of change is something I'm willing to consider obvious
> and start fast-tracking as well.  Objections?

It's already applied and is being pushed, and I see no reason to
short-circuit the normal testing framework for this sort of thing.
Speaking of which, we really should get zooko (or can you do this?) to
switch the buildbots over to darcs-unstable.  Otherwise they can't
give you much help, as you won't know until after you've applied a bad
patch...

> On Mon, Sep 08, 2008 at 23:41:05 -0700, Dmitry Kurochkin wrote:
> > Tue Sep  9 10:24:29 MSD 2008  Dmitry Kurochkin <[EMAIL PROTECTED]>
> >   * Move network related options to network_options.
> >
> > Tue Sep  9 10:28:24 MSD 2008  Dmitry Kurochkin <[EMAIL PROTECTED]>
> >   * Space in Darcs.Arguments module.
> >
> > Tue Sep  9 10:37:43 MSD 2008  Dmitry Kurochkin <[EMAIL PROTECTED]>
> >   * Documentation for -[-no]-http-pipelining options.
>
> Move network related options to network_options.
> ------------------------------------------------
> This looks good to me

Yes indeed.

> > -ssh_cm = DarcsMultipleChoiceOption
> > -         [DarcsNoArgOption [] ["no-ssh-cm"] NoSSHControlMaster
> > -                  "don't use SSH ControlMaster feature [DEFAULT]",
> > -          DarcsNoArgOption [] ["ssh-cm"] SSHControlMaster
> > -                  "use SSH ControlMaster feature"]
> > +network_options :: [DarcsOption]
> > +network_options =
> > +    [DarcsMultipleChoiceOption
> > +     [DarcsNoArgOption [] ["ssh-cm"] SSHControlMaster
> > +                           "use SSH ControlMaster feature",
> > +      DarcsNoArgOption [] ["no-ssh-cm"] NoSSHControlMaster
> > +                           "don't use SSH ControlMaster feature 
> > [DEFAULT]"],
> > +     DarcsMultipleChoiceOption
> > +     [DarcsNoArgOption [] ["http-pipelining"] HTTPPipelining "enable HTTP 
> > pipelining",
> > +      DarcsNoArgOption [] ["no-http-pipelining"] NoHTTPPipelining "disable 
> > HTTP pipelining"]]
>
> I would tack [DEFAULT] on one of these (or maybe that would be too messy
> in the code since we have a different default depending on the libcurl
> version)

That'd be a good improvement, but as you say, it's a bit tricky.  It
should just involve some cpp work, but that work ought to be in
URL.hs, not here, so it'd involve an extra export.

> A sort of random note: do lazy repositories mean that we have to let
> pretty every darcs command (e.g. apply, check) have networking options?
> I recall having an apply hang on me once until I hacked up darcs apply
> to accept --disable-pipelining (now --no-http-pipelining)

Yeah, we might want to move network_options into the "all" category so
that all commands automatically get it.  Of course, the same arguments
apply to the ssh_cm stuff, which is why it's good to have these two
together.

David
_______________________________________________
darcs-users mailing list
[email protected]
http://lists.osuosl.org/mailman/listinfo/darcs-users

Reply via email to