On Thu, Jun 09 2022, Junio C Hamano wrote: > Derrick Stolee <derrickstolee@xxxxxxxxxx> writes: > [...] >> I hope I am going in the right direction here, but I likely >> misunderstood some of your proposed alternatives. > > I wasn't seriously "proposing" an alternative. It was just that it > looked wasteful to go to a separate format (i.e. bundle) when packfiles > should suffice, as you would be adding extra information that is not > in bundles via the table-of-contents anyway, and what is given by a > bundle that is missing in a packfile is only the refs information, > which should be trivial to add to the table-of-contents. One thing that got pointed out to me by someone interested in this feature is that they'd be interested in serving up historical git repositories with a repo.bundle file. We're much better off with a format that includes refs for those use cases, even though no version of the patches that have been kicked around support this particular use-case yet (but it would be relatively easy as a follow-up). I also daresay that bundles will tend to integrate better into existing infrastructure, since if you're doing incremental backups with them it's a natural extension to throw those same incremental updates at a CDN and serve them up with bundle-uri. I also think we can come up with much better tooling for collections of bundles than random packfiles, since they declare what "tip" they want, and we're able to add arbitrary key-values to the header format in the future (and Stolee already added one such key-value).
