> I think this is an OK level of detail. I'm not sure quite sure about the > goal of the project, though. In particular: > > - I'm not clear what we'd hope to gain. I.e., what richer information > would we want to pass back and forth between index-pack and the > other processes? It might also be more efficient, but I'm not sure > it's measurably so (we save a single process, and we save some pipe > traffic, but the sideband demuxer would probably end up passing it > over a self-pipe anyway).
I didn't have any concrete ideas so I didn't include those, but some unrefined ideas: - index-pack has the CLI option to specify a message to be written into the .promisor file, but in my patch to write fetched refs to .promisor [1], I ended up making fetch-pack.c write the information because I didn't know how many refs were going to be written (and I didn't want to bump into CLI argument length limits). If we had this feature, I might have been able to pass a callback to index-pack that writes the list of refs once we have the fd into .promisor, eliminating some code duplication (but I haven't verified this). - In your reply [2] to the above [1], you mentioned the possibility of keeping a list of cutoff points. One way of doing this, as I state in [3], is my original suggestion back in 2017 of one such repository-wide list. If we do this, it would be better for fetch-pack to handle this instead of index-pack, and it seems more efficient to me to have index-pack be able to pass objects to fetch-pack as they are inflated instead of fetch-pack rereading the compressed forms on disk (but again, I haven't verified this). [1] https://public-inbox.org/git/20190826214737.164132-1-jonathanta...@google.com/ [2] https://public-inbox.org/git/20190905070153.ge21...@sigill.intra.peff.net/ [3] https://public-inbox.org/git/20190905183926.137490-1-jonathanta...@google.com/ There are also the debuggability improvements of not having to deal with 2 processes. > - index-pack is prone to dying on bad input, and we wouldn't want it > to take down the outer fetch-pack or receive-pack, which are what > produce useful messages to the user. That's something that could be > fixed as part of the libification, but I suspect the control flow > might be a little tricky. Good point. > - we don't always call index-pack, but sometimes call unpack-objects. > I suppose we could continue to call an external unpack-objects in > that path, but that eliminates the utility of having richer > communication if we sometimes have to take the "dumb" path. A while > ago I took a stab at teaching index-pack to unpack. It works, but > there are a few ugly bits, as discussed in: > > > https://github.com/peff/git/commit/7df82454a855281e9c147f3023225f8a6f72e303 > > Maybe that would be worth making part of the project? I'm reluctant to do so because I don't want to increase the scope too much - although if my project has relatively narrow scope for an Outreachy project, we can do so. As for eliminating the utility of having richer communication, I don't think so, because in the situations where we require richer communication (right now, situations to do with partial clone), we specifically run index-pack anyway.
