- From: L. David Baron via GitHub <sysbot+gh@w3.org>
- Date: Wed, 14 Aug 2024 19:22:46 +0000
- To: public-css-archive@w3.org
I came back to this issue to make the edits for #10369, and I realized that (somewhat related to the two comments immediately above) the edits made here don't quite match what I had implemented before the edits were made. In particular, I hadn't implemented the eager canonicalization (or the new `percentage` keyword). Instead, I had only done the canonicalization to substitute away the calc-size basis when trying to interpolate. I also think that the discussion in #10369 assumed a different conclusion here, in other words, it assumed that it would be possible to have two valid endpoints of an interpolation (because they weren't simplified) but a midpoint that would be IACVT (invalid at computed value time) if interpolation was attempted. The edits here made the endpoints IACVT, which means that I think the situation in #10369 (that we resolved to handle as described there, and which I implemented) can't actually occur. Do you think we should instead switch the Chromium implementation to what's in the spec now -- eager canonicalization -- which I think means discarding the resolution to #10369, and which also means somewhat more frequent occurrence of IACVT behavior? Or do you think we should switch the spec to follow the slightly less eager canonicalization combined with the resolution in #10369, which I think in practice means there's *never* IACVT behavior because the cases where it would happen lead to failure to interpolate instead. (I think avoiding IACVT is good!) -- GitHub Notification of comment by dbaron Please view or discuss this issue at https://github.com/w3c/csswg-drafts/issues/10220#issuecomment-2289680171 using your GitHub account -- Sent via github-notify-ml as configured in https://github.com/w3c/github-notify-ml-config
Received on Wednesday, 14 August 2024 19:22:47 UTC