Tidying up 2.0

[ChrisShank]

I think there are some things left to explore before we publish 2.0. I just successfully updated my arrow library to use it and things are looking good! I would love to do some side by side comparisons to make sure perf and memory consumption have improved.

Todo List

• [ ] Update README/docs
• [x] think about adding a takeRecords method (This could be a 2.1 task)
• [ ] investigate perf/memory consumption
• [ ] think about removing private methods (see https://github.com/samthor/viz-observer/pull/3#discussion_r1575417324) (This could be a 2.1 task)
• [ ] optimize root refreshes (see https://github.com/samthor/viz-observer/pull/3#discussion_r1568076902) (This could be a 2.1 task, although it might be consider breaking 🤔)

[samthor]

I honestly can't remember exactly my thought process years ago. But...

I'm a bit stuck on - absolute elements aren't reported correctly by the IntersectionObserver with document.documentElement except if body is set to position: relative. The other option is to use document as the root, which works, but then makes scrolling have the IntersectionObserver need to be recreated. There could be a dual path, but working out what positioning a random element is under is kind of tricky...

On Fri, Apr 26, 2024, 14:54 Chris Shank @.***> wrote:

I think there are some things left to explore before we publish 2.0. I just successfully updated https://github.com/ChrisShank/perfect-arrow/commit/0d25037a2bb083c2ffb626b819ae668dd16a340f my arrow library to use it and things are looking good! I would love to do some side by side comparisons to make sure perf and memory consumption have improved.

Todo List

• Update README/docs
• think about adding a takeRecords method
• investigate perf/memory consumption
• think about removing private methods (see #3 (comment) https://github.com/samthor/viz-observer/pull/3#discussion_r1575417324 )
• optimize root refreshes (see #3 (comment) https://github.com/samthor/viz-observer/pull/3#discussion_r1568076902 )

— Reply to this email directly, view it on GitHub https://github.com/samthor/viz-observer/issues/4, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAA5DEFVISNYYG6WO3G7BOLY7HMXFAVCNFSM6AAAAABG2DAOIGVHI2DSMVQWIX3LMV43ASLTON2WKOZSGI3DIOJVHE4TKMQ . You are receiving this because you are subscribed to this thread.Message ID: @.***>

[ChrisShank]

except if body is set to position: relative.

Ya its a tough problem and I wonder if it's a browser bug or just an unintended quirk. I currently lean towards very explicitly documenting this edge case so consumers can make the necessary change.

[ChrisShank]

Addressing some things here in #5

[samthor]

It's probably a reasonable compromise if this API is sort of a pseudo-polyfill for a feature that we personally want to exist. that is, imagine a world where one browser has this, and this was a polyfill—you could say you need to do this odd thing that mostly has little effect on the page.

[ChrisShank]

fun fact this approach is transiently being used in the css anchoring pollyfill https://github.com/oddbird/css-anchor-positioning/blob/main/src/polyfill.ts#L2