Andrew Olsen

Addressing comments - @rcoup's suggested cases: 1. `git_commit_tree()` - done 2. `git_reference_peel()` - done (via `git_object_peel()`) 3. `git_tag_peel()` - done (via `git_object_peel()`) (Note that case 2 and 3 only return...

I had another idea - making a separate "error-subtype" int or bitfield or whatever in the thread-local, same as is used to store the last error message. That way all...

Trying to design the behaviour - sorry it got a bit long. I will start by saying that these different types of errors have to be sort of best-effort or...

Yeah sorry I wrote a little bit about this in https://github.com/libgit2/libgit2/pull/5776 but I should have been more clear about it in this PR which is actually implementing it. See https://git-scm.com/docs/partial-clone/2.19.0#_handling_missing_objects...

Hi @ethomson - sorry if the above got a bit philosophical. Is there anything I can do to help this along, or is there anything in particular where you feel...

I took the liberty of adding another commit that conveys the extra information we have so far as an error "subcode" - I can revert it if it doesn't make...

#5776 has been closed since it is superceded by #5993 (and follow-ups)

My current plan - 1. New error code EMISSING for when objects are referenced by the ODB, but missing from the ODB. #5993 2. Detect when packfiles are promisor packfiles...

@ethomson just looking at recent commits, it seems like you are the best at making things happen in libgit2 - can you comment on if this is a reasonable plan,...

Hi @erikmchut and @ethomson am I still working on it? Not really - I have a fork of libgit2 that seems to do what I want it to for now....