I'm trying to tackle the update of the repository cache to the multiple
repository situation. I have one question re.
MultipleRepositorySupport/Cache.

In the "Model" section, the "repository" table should bind an "id" to a
repository UUID (probably the .name attribute of the SVN repos) and some
metadata. What should that "id" be? The name used in the [repositories]
section as the config key? This would mean that changing the "mount
point" of a repository would require resyncing the cache.

I would propose enhancing the [repositories] section as follows:

[repositories]
my_repo_a.dir = <path to repository>
my_repo_a.mount = <mount point of repository>
my_repo_a.type = <type of repository>

and to use "my_repo_a" as the repository key for the cache. This would
allow changing the mount point without changing the key (i.e. without
invalidating the cache), and would also allow having a mount point with
caps or the "." character, as requested at the bottom of
MultipleRepositorySupport. We could even make .mount optional, and use
the key as the mount point by default.

Does that make sense?

-- Remy


PS: Should I continue asking questions about multirepos on trac-dev (for
"development documentation"), or should I mail Christian directly?

Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to