The registry endpoints are all about modifying the index. These git
endpoints are all about fetching.
Cargo uses git to sync index data incrementally, so in order to give it
what it wants we need to implement the "fetch" portions of the "smart"
git server protocol as detailed here:
With these endpoints in place, folks will configure cargo to use the
registry with:
[registries]
estuary = { index = "<scheme>://<host>:<port>/git/index" }
The Procfile was introduced to the project because we (prior to this
diff) needed to run git daemon to serve the git repo in addition to
the estuary server.
I've left the Procfile where it is, but removed the git server from it now
that it's redundant.
Who knows, the Procfile may still be of interest later on if/when other
services are required for estuary to operate.
The registry endpoints are all about modifying the index. These git endpoints are all about fetching.
Cargo uses git to sync index data incrementally, so in order to give it what it wants we need to implement the "fetch" portions of the "smart" git server protocol as detailed here:
With these endpoints in place, folks will configure cargo to use the registry with:
The
Procfile
was introduced to the project because we (prior to this diff) needed to rungit daemon
to serve the git repo in addition to the estuary server.I've left the
Procfile
where it is, but removed the git server from it now that it's redundant.Who knows, the
Procfile
may still be of interest later on if/when other services are required for estuary to operate.