Rowan Tommins

This appears to be the reference in the current spec: https://html.spec.whatwg.org/multipage/parsing.html#adoption-agency-algorithm

Possibly relevant is that calling `feof($r)` after the first read returns `true`, whereas with a real file it returns `false` until after an additional call to `fgets`: ```php $file =...

I believe the test cases in https://github.com/bovigo/vfsStream/pull/223 reproduce my issue and describe the behaviour that would match native files.

Ah, I think I see what's happening then. `fgets` is doing the following under the hood: 1. On the first call, ask the stream wrapper for _up to_ 8192 bytes...

Based on the above, I have raised a PR which tracks offset and EOF separately: https://github.com/bovigo/vfsStream/pull/224/files The possibly non-obvious part is that `fseek` always resets `feof` to false, even if...

> The upside of this approach is normal users won't have to run the real file tests on their system and CI can/will. Whereas in my approach I can't turn...

Alternative implementation that applies on top of #221: https://github.com/bovigo/vfsStream/compare/refactor/rename_and_restructure...IMSoP:issue-222-rebase Oddly, a bunch of the tests that my change caused to fail don't seem to exist on that branch.

Yeah, that makes sense. I think to backport to 1.x, we'd need to also backport #220 (otherwise the streams all have the same offset anyway), so it needs a call...

I am very much interested in this, because I would like to be able to document APIs that already exist, even if they don't conform to common designs. As such,...

[Sealed Classes](https://wiki.php.net/rfc/sealed_classes) is also listed as Accepted, but it's actually Declined.