Mark Hammond

> So, if this is the case, the GUID could be generated by applying a suitable hash function to the concatenation of those attributes. It also occurred to me that...

To be quite honest, I don't think this addon should bother trying to support Sync in this way (ie, if it was my addon, I'd probably chose to not support...

I've added a simple test case for this. I also added the `Option` to `FileState`, but I'm not sure when it would get set other than `None` :) I'll have...

Thanks for the comment. I suspect that "full" support for case-insensitive file-systems will be a multi-pronged effort. In this first instance I'd be very happy with just allowing a tree...

> Maybe it's better to walk the directory tree on disk and then check what's already in the base tree to decide which path to write to the new tree....

> What do you think about skipping entries in a `Tree` object if we've already visited an entry that maps to the same path? I think there are 2 questions...

> (and the normalization rules apply even on some case-sensitive filesystems). I guess those case-sensitive file-systems will have the same problem - if walking the fs provides paths which aren't...

> Matching Git is a bit more limiting but probably fine. When I said "matching git" I meant in a very loose way. git doesn't necessarily do the *best* thing,...

> I'll try and get some insight into what git actually does with the `core.ignorecase` setting too, but I suspect I'll find that code-base fairly intractable. heh, I was wrong....

FWIW, https://hg.mozilla.org/mozilla-central/rev/3b6c10d28209 just landed, which means the current tip of gecko no longer exhibits this problem (although obviously it may come back at any time and still exists for older...