In D11212#174951, @valentin.gatienbaron wrote:

The two wrappers make sense, although it seems random to encode the query as {} in one case and None in other case

In D11212#174786, @danchr wrote:

In D11212#174524, @Alphare wrote:

Thanks for bringing this up. Does that represent a large amount of work for you?

Well, my main challenge here is that I have absolutely no idea what's going on — I don't have much experience with the very low levels of networking in Mercurial, or the (relatively) new peer API. It seems to me that the point of that code is, essentially, to provide a minimal repository or peer that does nothing. Could something like that be moved to Mercurial core?

In D11212#174524, @Alphare wrote:

In D11212#174426, @danchr wrote:

In D11212#173796, @Alphare wrote:

I don't think the backwards compatibility concerns are too bad, but I might be wrong.

Please note that this breaks hg-git, as it currently relies on the future for wrapping a repository. We'll probably have to come up with a wrapper regardless.
See e.g. https://foss.heptapod.net/mercurial/hg-git/-/blob/branch/default/hggit/gitrepo.py#L113

Thanks for bringing this up. Does that represent a large amount of work for you?

In D11212#173796, @Alphare wrote:

I don't think the backwards compatibility concerns are too bad, but I might be wrong.

In D9725#147983, @marmoute wrote:

This looks fine, but could you describe a bit what this is suppose to achieve and maybe provide a link to an example ?

This breaks the build; could you please unapply it while I investigate?

In D9670#146245, @Alphare wrote:

In D9670#146220, @marmoute wrote:

What is this patch doing ? Should we add some comment in the code to explain the rationnal around the skip thingy ?

Thinking about it some more, this should just use https://doc.rust-lang.org/std/path/struct.Path.html#method.strip_prefix

In D9671#146004, @Alphare wrote:

I'm not a big fan of the idea of people using this version with incomplete normalization by lifting the compiler error since it may cause bugs, wrong repo states or even security issues.

In D9640#146337, @marmoute wrote:

This looks good.

As a minor addendum: With this change, the Rust dirstate code almost works with case folding, but has issues detecting folded directory names. This appears mostly unimplemented AFAICT.

In D8937#134080, @marmoute wrote:

Looks like we should take the change to dirstate_map.rs (in its own changesets) and delay the change to rust/hg-core/src/lib.rs, right? @danchr can you split this?

I actually hadn't noticed that you kept discussing this prior to abandoning it. I'm busy with other things, though, so I won't be able to follow up on this.

In D9256#139031, @durin42 wrote:

Does this fix a defect? The test makes it seem like it does?

In D8937#133741, @Alphare wrote:

What do you mean by "mishandling removed files"? The fold map is there to give a mapping between the normalized (internal) representation of file paths and the actual ones on disk. As much as I hate the dirstate code, this is a useful part of the software that is triggered depending on your filesystem and paths used.