Plans for Loggerhead

[Martin Albisetti]

On Wed, Jan 7, 2009 at 12:15 AM, Robert Collins
<robertc at robertcollins.net> wrote:
> High level questions:
>  - json objects are usable by browser side code as-in ?

Used to render the actual html via javascript.

>  - You seem to suggest a _lot_ of new caching activity in loggerhead,
>   which scares me badly. Have I misunderstood?

I suggest the possibility to cache  :)
Slapping squid in front of Loggerhead with this way of doing things
should make it quite scalable.


> Random thoughts:
>  - loggerheadlib should be 'bzrlib.plugins.loggerhead

Sounds reasonable. I don't exactly know what the consequences of that
is, but it sounds reasonable :)


>  - how does this help with working with multiple branches at once in
>   loggerhead?

The biggest difference would be that we would *just* generate the
jsons server-side, and the rendering would be done client-side. This
will relieve us of a *lot* of processing. It gets more powerful with
multiple related branches if we cache the json files (however we
choose to do it), as you would only need to get the information for a
revision once.


-- 
Martin