Re: [csswg-drafts] [css-fonts-5] Text Fitting: Default scaling limit (#12886)

The CSS Working Group just discussed `[css-fonts-5] Text Fitting: Default scaling limit`.

<details><summary>The full IRC log of that discussion</summary>
&lt;bramus> tkent: the problem here is that when the user zooms in the page they expects that the text size in ??? and that the content size is viewport oblivious<br>
&lt;bramus> … not changed by operation, so text-size is not changed<br>
&lt;astearns> s/in ???/will increase/<br>
&lt;bramus> … so when the user ??? they need to avoid changing the text size<br>
&lt;bramus> … in the issue kizu proposed to add a scaling limit of 1500%<br>
&lt;kizu> q+<br>
&lt;astearns> s/1500/200/<br>
&lt;TabAtkins> q+<br>
&lt;bramus> … and people including me proposed to start layout a a point of which physical size is ??? in 100% zoom level<br>
&lt;bramus> … so that is we think our way to resolve the issue<br>
&lt;bramus> astearns: the summary is int he last comment as well<br>
&lt;astearns> ack kizu<br>
&lt;bramus> kizu: would be nice if that method worked.<br>
&lt;bramus> … I have 3 points<br>
&lt;bramus> … 1. is not implemented in the prototype, has updated text-fit but doe snot yet have this implemented<br>
&lt;bramus> … would be nice to ahve a prototype to test this<br>
&lt;bramus> … will try to prepare a page for the ???limit<br>
&lt;bramus> … would be nice if chrome could provide an alternative to look at both cases together and decided<br>
&lt;bramus> … then 2 questions about the method<br>
&lt;bramus> … 1. the proposed formula for the google method mentions 100% of zoom, so q is when is this calculated? page load or not?<br>
&lt;bramus> … once you load a page, then you zoom in and refresh ; will the font size change<br>
&lt;bramus> … or once you zoom in, the page has to layout twice to get the 100% zoom value for the font-size<br>
&lt;bramus> … 2. related to the proposal about the post-??? font-size<br>
&lt;bramus> … see problem: the way text fitting works we first find the breaks with orig font size, then do the text fit and fit all the lines<br>
&lt;bramus> … if we takes this value and multiply by zoom factor of the page, the text could overflow<br>
&lt;bramus> … will this mean we have an extr aline breaking happening if it doesnt fit, or will it overflow the container and potentially create overflow?<br>
&lt;bramus> … if it creates overflow, it is problematic, which goes against  SC 1.4.4<br>
&lt;bramus> … zooming shouldnot create unnecessary overflow<br>
&lt;bramus> … contention between repsonsive typography but you dont want it to overflow<br>
&lt;bramus> tkent: for q1<br>
&lt;bramus> … intention is to do layout twice<br>
&lt;bramus> … one at 100% of zoom<br>
&lt;bramus> … for q2, we directly apply the updated font-size before line breaking so it could change line breaking<br>
&lt;bramus> astearns: so you could get overflow in that 2nd lahyout depending on the container<br>
&lt;bramus> TabAtkins: not inline overflow, would break to a new line<br>
&lt;bramus> kizu: if there is 1 word with no breaking oppportunities, then it could<br>
&lt;bramus> astearns: which is also the case for the 100% layout as well<br>
&lt;astearns> ack TabAtkins<br>
&lt;bramus> TabAtkins: dont have strong opinions. I like kent’s proposal to use the larger of the two. seems to satisfy the a11y guidelines while still letting text fitting do its thing<br>
&lt;astearns> ack fantasai<br>
&lt;Zakim> fantasai, you wanted to ask about font size limits and nowrap<br>
&lt;bramus> … so sgtm<br>
&lt;bramus> fantasai: Where is the proposal?<br>
&lt;bramus> astearns: last comment<br>
&lt;bramus> fantasai: but the feature in general? which spec?<br>
&lt;bramus> TabAtkins: in PR<br>
&lt;bramus> fantasai: not huge fan of writing a spec and then &lt;missed><br>
&lt;kizu> q+<br>
&lt;fantasai> s/&lt;missed>/iterating over them over months in the form of a PR/<br>
&lt;astearns> ack kizu<br>
&lt;bramus> kizu: this could work from what I understand first if th euser is zoomed we do layout twice and then do an extra pass to first break lines, break again if overflow and then replay text fitting<br>
&lt;bramus> … only concern how performant this is because the potential problem I see is that if we dont do all that work at 100% zoom, users who use different zoom have perf impact on the page in general<br>
&lt;bramus> … if it is ??? I think this is good, but would b enice to look at a prototype and play with it<br>
&lt;bramus> … and to see it in action<br>
&lt;astearns> ack fantasai<br>
&lt;Zakim> fantasai, you wanted to ask about font size limits and nowrap<br>
&lt;bramus> astearns: agree that having a demo of all solutions to this problem is going to be key here, bc there are people with a11y expertise that might want to evaluate<br>
&lt;bramus> fantasai: we talked a lot about re=-breaking the line and that makes sense for a good chunk of problems<br>
&lt;bramus> … but what is the text is nowrap? does it shrin kdown and fit 1 single line?<br>
&lt;bramus> … what if its too small … what do we do in that case?<br>
&lt;bramus> iank_: the WCAG reqs are written in such a way that clipping of content doesnt exist.<br>
&lt;bramus> … so basically its a rock and hard place situation, WCAG say you must grow always but if you do that you might clip<br>
&lt;bramus> astearns: im hearing people express the opinion tha tthis seems promising but that we would like to see it in action<br>
&lt;bramus> … so kent, would you be able to prototype this resizign behavior on zoom?<br>
&lt;bramus> tkent: yes<br>
&lt;TabAtkins> here we go, the feature is currently living in an explainer https://github.com/explainers-by-googlers/css-fit-text/blob/main/README.md<br>
&lt;bramus> astearns: that would be the next step<br>
&lt;bramus> … and roman to prototype – though at some point it might get implemnted with the hacks<br>
&lt;bramus> kizu: I can do it in chrome canary  as it has suppport for max values<br>
&lt;bramus> … and shows all those situations<br>
&lt;bramus> … should be simple enough<br>
&lt;bramus> astearns: so lets get that done, and then evaluate (and perhaps iterate), and then show a11y folks<br>
&lt;bramus> … does that sound like enough for today, kent?<br>
&lt;bramus> tkent: that’s nice, thank you<br>
&lt;astearns> ack fantasai<br>
&lt;bramus> fantasai: in terms of making this easier to review, looks like these are all against css-text-5 … can we try and be consistent?<br>
&lt;bramus> … there is no way to figure out what we are talking about when landing on an issue … can we link all from the top of the spec, of start an ED?<br>
&lt;bramus> astearns: do we have a resolution to add this to a draft? dont remember …<br>
&lt;bramus> kizu: i think it was done at TPAC<br>
&lt;bramus> … we changed it<br>
&lt;bramus> astearns: and I made you editor IIRC<br>
&lt;bramus> kizu: not sure if I will able to describe this well. Kent and Tab suggested they could do it<br>
&lt;bramus> … happy to review the text<br>
&lt;bramus> TabAtkins: sure<br>
&lt;bramus> astearns: so lets try to get this landed and demos together and see where we go from there<br>
&lt;bramus> … thanks kent for joining<br>
&lt;fantasai> s/against css-text-5/against css-text-5/tagged against css-fonts-5 but spec is resolved to be in css-text-5/<br>
</details>


-- 
GitHub Notification of comment by css-meeting-bot
Please view or discuss this issue at https://github.com/w3c/csswg-drafts/issues/12886#issuecomment-3814480919 using your GitHub account


-- 
Sent via github-notify-ml as configured in https://github.com/w3c/github-notify-ml-config

Received on Wednesday, 28 January 2026 23:31:54 UTC