Daniel Alley

That seems reasonable, but do you see a reason not to mirror the `BytesText` API here? e.g. * `Attribute::from_escaped_bytes` * `Attribute::from_bytes` * `Attribute::from_escaped_str` * `Attribute::from_str` (I don't love the `_plain_`...

`borrowed_name()` doesn't exist anymore (in the current unreleased master branch)

For some complex situations `_with_custom_entities` won't be enough. Example: as per the specificiation, [entity replacement in text contents can impact XML parsing](https://www.w3.org/TR/xml/#sec-entexpand): > If the DTD contains the declaration >...

Figuring out how much of an improvement this is (or even if it is one, on average) might be challenging because I get different results on my AMD based desktop...

Code-wise, this all looks fine to me, I don't have any complaints that block the PR. I also checked out the branch, ran tests, went through the code for a...

So for the stuff I'm about to say, wait on an acknowledgement and :+1: from @Mingun before putting any effort into addressing it. I have a patch that you can...

I just meant the overall idea, rather than any specific PR. Anyway, #426 ought to address the concerns about commit #3, which just leaves #2. I really do appreciate how...

I promised some patches: * Add some debug_asserts to protect invariants - https://pastebin.com/mLP5wfat * Only allow `&str` for `SliceReader` - https://pastebin.com/Gxd6xcMj With the second commit, you can see the error...

That's true - but it's partly why I'm OK with letting the issues I mentioned be addressed as followups. I'd rather not force @999eagle to go through a bunch of...

@Mingun I think if you were to compare the files side-by-side, manually, like this:  it would quickly be apparent how little has actually changed. The new...