Re: [Monotone-devel] partial pull #2 - gaps instead of a single horizon

[Zbynek Winkler]

On 5/28/07, Markus Schiltknecht <address@hidden> wrote:
> As the netsync protocol needs to change anyway, to support partial pull,
> I'm in favor of adding those node ids (note that they don't necessarily
> need to be the same as any one peer's local node ids).

While we are at it (changing netsync protocol) - why not to make it
history based as was proposed some time ago? I really liked the
idea... and it would allow things like "give me all descendants of
this revision that match this branch pattern" on pull which is IMHO
exactly what is needed for partial pull. To me it is the definition of
partial pull... how else would you want to reliably restrict what you
pull?

And for the trust thing - as far as I understand it - you would only
have to trust the hash of the revision at the beginning of the gap
(history wise) and if this hash can be verified by other means, it is
fine, isn't it?

Regarding the missing revisions - I would not allow to do anything
across a gap because it would not be reliable - ie if you want to
merge something you'd have to get enough revisions to have all the
information needed. For other commands I'd happy even if they just
stopped where they start missing revisions (log, annotate etc).

Zbynek

--
http://robotika.cz/