On Sun, Oct 27, 2024 at 05:23:24PM -0400, karthik nayak wrote:

While thinking about this over the last few days and also getting some
advice from Patrick, I realized that we don't need to be this disruptive
by simply adding the 'repository' variable to the already existing
'packed_git' struct. This allows us to leverage this information more
easily, since most of the functions already have access to the
'packed_git' struct.

Great idea!

This, plus the series by Jeff 'jk/dumb-http-finalize' which also removes
some existing functions. We reduce the impact to only 3 functions being
modified.

I think with such low impact, it might make more sense to not go with
the Coccinelle approach, since it is a lot simpler without it.

I'll post a new version tomorrow showcasing this approach, but I'll
leave the final decision to you whether it is still disruptive, and if
the approach you mentioned would be better.

I'll have to see the end result to know for sure, but it sounds like
this would be a good way to move it forward without being too
disruptive.

I wonder how it interacts with alternates, though...

Thanks,
Taylor