Re: [Monotone-devel] sync defaults

[Timothy Brownawell]

On Tue, 2008-03-18 at 05:30 -0400, Stephen Leake wrote:
> Timothy Brownawell <address@hidden> writes:
> 
> > On Sun, 2008-03-16 at 19:46 -0400, Stephen Leake wrote:
> >> I'm proposing the following patch to cmd_netsync.cc:
> > [...]
> >> The first change (passing 'true' to find_key_if_needed) requires a key
> >> (needed for writing to the server database); that means it will look
> >> in _MTN/options if present. That was the behavior in mtn 0.39, and is
> >> the same as 'mtn push'. 'mtn pull' sets this to false, since it can
> >> work anonymously.
> >>
> >> The second change writes the workspace options if --set-default is
> >> specified and a workspace is present. That allows fixing an error
> >> reported by sync when there are multiple keys in the keydir.
> >>
> >> Hmm. 'push' should also write the
> >>
> >> Most commands write the workspace options even without --set-default,
> >> but it seems appropriate to require that for sync.
> >>
> >> The current help for '--set-default' says:
> >>
> >>   --set-default        use the current arguments as the future default
> >>
> >> I initially interpreted that to mean "arguments and options", and
> >> wondered why it wasn't writing 'key' to _MTN/options. This change
> >> makes it mean that.
> >>
> >> Any objections?
> >
> > I like it, but the first half will conflict with some changes I'm in the
> > middle of. Do we have a test that checks this behavior?
> 
> I've committed the above changes, and a test 'sync_use_ws_options'.
> 
> When do you need to pass 'false' to find_key_if_needed?

It's more that I folded find_key_if_needed into other (new) functions,
as part of changing how netsync args are handled to allow for 'mtn://'
urls. Should be pushed soon-ish, after I make sure I didn't break the
new test.