Randell Jesup

Note also that resolve() (or whatever we rename it to) is also impacted, of course. With a ref-based approach, resolve() would return the new path; with a pure path-based approach,...

I prefer reference-based. It also simplifies thinking about SyncAccessHandles, since in a reference world, it's ok if an ancestor changes name, so there's no need to lock directories above it...

If FileSystemHandle is a path, then the option to create the file/dir is odd and doesn't make sense; in that case the option to create should be on createSyncAccessHandle/createWritable. If...

If there's specific action an application might want to take for a specific failure mode, then tightly specifying errors would be useful. Otherwise tightly specifying errors doesn't have much benefit...

I'd suggest considering changing the function name/signature when you correct this, as that will breakage due to API change -- either remove the old one (compiler will notify you to...

It looks like no real work has been done (and hardly any comments) to any of the issues filed against this over 3 years ago. It has the appearance of...

h3 is about 25% of all traffic currently, and rising (and H2 is near 50% and stable). WebTransport on H2 is coming. Please keep this aligned, api-wise, with WebTransport. Is...

I assume the issue here is the "path separator [] on the underlying filesystem". It's problematic for this allowed filenames to depend on the platform; if we want to block...

From https://man7.org/linux/man-pages/man2/rename.2.html POSIX rename() atomically replaces 'newfile' on rename() if it exists, so you're correct that posix wants that behavior. I think at least some non-normative mention of WebLocks to...

Agreed. This is no longer relevant since we modified SyncAccessHandle