Changing UA defaut for text-decoration-skip?

26 views
Skip to first unread message

Dominik Röttsches

unread,
Sep 26, 2017, 4:30:40 AM9/26/17
to blink-api-ow...@chromium.org, Emil A Eklund

Dear API Owners,


in Chrome 57 we shipped support for text-decoration-skip: ink; which allows skipping descenders in underlines (illustration, see also image below). We initially suggested making it the default in the intent process but dropped this part for shipping because we had assumed a spec change was required. So we went back to the spec process to gather additional feedback.


Koji and I would like to come back to enabling this underline rendering mode as a default. We would like to ask for your feedback and ideally permission to go ahead with this change. eae@ previously expressed his support.


We believe it brings better typography to the web. Safari has been doing this since October 2014 in Safari 8, and their spec contributors initially proposed to make it a default. FF considers implementing skip: ink; so is generally positive towards it, but they did not get around to implementing it yet. It generally seems to be a desired feature. Interoperability risks can be considered low, see below.


To put this in place, we would like to like to extend the body rule in the UA default style sheet with:


body {

…

   text-decoration-skip: ink objects;

}


What has changed from our initial attempt at changing the default:  We initially thought, this would require a spec change of the initial value. However, the feedback by fantasai@ in the CSS working group spec process has been that UA's may choose to define such default behavior and such a change would not require changing the CSS initial value in the specification.


The spec discussion has converged on deferring text-decoration-skip: from the CR process of CSS Text Decorations level 3 to the text-decoration-skip property to CSS Text Decorations level 4 to add additional values and possibly longhand properties. Out of these, Blink supports objects and ink.


Similarly, Safari supports only the -webkit-* prefixed version, and also only supports objects and ink. It parses 'none' as a value, but behaves like 'objects' when this setting is set. FF


Safari enabled default descender skipping in underlines and uses the currently not-specced value 'auto' as the default for the -webkit-text-decoration-skip property, but it is interoperable with Blink in that ‘none’ disables the skipping.


Koji and I believe that we would gain a visual improvement and better typography on the web with default descender skipping in underlines and would like to hear your feedback, or ideally your approval on this proposed change.


Thanks,


Koji & Dominik


Dimitri Glazkov

unread,
Sep 26, 2017, 11:59:33 AM9/26/17
to Dominik Röttsches, blink-api-ow...@chromium.org, Emil A Eklund
Thanks for providing the details and context. This seems like a pretty straightforward intent to ship to me. 


FWIW, I am not sure I agree with fantasai that this is not something that should be in the spec. I view the role of the specifications as ensuring interoperable implementations, and leaving UA default unspec'd seems to go against that.

:DG<

--
You received this message because you are subscribed to the Google Groups "blink-api-owners-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-api-owners-d...@chromium.org.
To post to this group, send email to blink-api-ow...@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-api-owners-discuss/CAN6muBtNv4hbbQeVw6D1-ZYMs3JQxfH-RRaLV05tuFNJs9dnjw%40mail.gmail.com.

Domenic Denicola

unread,
Sep 26, 2017, 12:07:38 PM9/26/17
to Dimitri Glazkov, Dominik Röttsches, blink-api-ow...@chromium.org, Emil A Eklund

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-api-owners-discuss/CAOtFfx4SYW25-oHACTG0sv2%3D-ZAMKQVUi3-aqnmoPD%3D0StDp%2BQ%40mail.gmail.com.

Dimitri Glazkov

unread,
Sep 26, 2017, 12:13:16 PM9/26/17
to Domenic Denicola, Dominik Röttsches, blink-api-ow...@chromium.org, Emil A Eklund
Ah great! So, it sounds like the spec change needs to happen in the HTML Standard then, right?

:DG<

Domenic Denicola

unread,
Sep 26, 2017, 12:15:51 PM9/26/17
to Dimitri Glazkov, Dominik Röttsches, blink-api-ow...@chromium.org, Emil A Eklund
Ideally, yes, but historically this area hasn't been super-well-coordinated. So, I think submitting a spec/web platform tests PR would be a good thing to do, and set a good precedent, but it would be going above and beyond what previous changes in this area have required.

To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-api-owners-discuss/CAOtFfx6pRnbUbUuSvhG_v_3N0kg6ue_thFf3pCZr_dWHNsFYPw%40mail.gmail.com.

Chris Harrelson

unread,
Sep 26, 2017, 12:40:48 PM9/26/17
to Domenic Denicola, Dimitri Glazkov, Dominik Röttsches, blink-api-owners-discuss, Emil A Eklund
I agree with the earlier comments - seems like a good change, but should be spec'ed. We should also reach out to the other browsers to drive consensus and make changes to all of them to improve interop on this.

On Tue, Sep 26, 2017 at 9:15 AM, 'Domenic Denicola' via blink-api-owners-discuss <blink-api-ow...@chromium.org> wrote:
Ideally, yes, but historically this area hasn't been super-well-coordinated. So, I think submitting a spec/web platform tests PR would be a good thing to do, and set a good precedent, but it would be going above and beyond what previous changes in this area have required.
On Tue, Sep 26, 2017 at 12:13 PM Dimitri Glazkov <dgla...@chromium.org> wrote:
Ah great! So, it sounds like the spec change needs to happen in the HTML Standard then, right?

:DG<

To unsubscribe from this group and stop receiving emails from it, send an email to blink-api-owners-discuss+unsub...@chromium.org.
To post to this group, send email to blink-api-owners-discuss@chromium.org.

--
You received this message because you are subscribed to the Google Groups "blink-api-owners-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-api-owners-discuss+unsub...@chromium.org.
To post to this group, send email to blink-api-owners-discuss@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-api-owners-discuss/CAOtFfx4SYW25-oHACTG0sv2%3D-ZAMKQVUi3-aqnmoPD%3D0StDp%2BQ%40mail.gmail.com.

--
You received this message because you are subscribed to the Google Groups "blink-api-owners-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-api-owners-discuss+unsub...@chromium.org.
To post to this group, send email to blink-api-owners-discuss@chromium.org.

--
You received this message because you are subscribed to the Google Groups "blink-api-owners-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-api-owners-discuss+unsub...@chromium.org.
To post to this group, send email to blink-api-owners-discuss@chromium.org.

--
You received this message because you are subscribed to the Google Groups "blink-api-owners-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-api-owners-discuss+unsub...@chromium.org.
To post to this group, send email to blink-api-owners-discuss@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-api-owners-discuss/CAM0wra8dmc3qbygkC%3Dviv7AYcnuKP6VsDFU3V9nq8%2BBJH7xxWA%40mail.gmail.com.

Rick Byers

unread,
Sep 26, 2017, 9:41:20 PM9/26/17
to Chris Harrelson, Domenic Denicola, Dimitri Glazkov, Dominik Röttsches, blink-api-owners-discuss, Emil A Eklund
Is the resulting change at all observable to the web platform?  I.e. is there any risk of it breaking something (eg. by changing text metrics)?  If not, then I think this gets blurrier (arguably not a platform change at all).  Though I suppose UA styleSheets are always observable via getComputedStyle...

Regardless I'd suggest erring on the side of an intent to ship where we can hash out the details (spec or not, compat risk zero or non-zero), but I too expect totally smooth sailing.

On Tue, Sep 26, 2017 at 12:40 PM, Chris Harrelson <chri...@chromium.org> wrote:
I agree with the earlier comments - seems like a good change, but should be spec'ed. We should also reach out to the other browsers to drive consensus and make changes to all of them to improve interop on this.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-api-owners-discuss+unsubscr...@chromium.org.

--
You received this message because you are subscribed to the Google Groups "blink-api-owners-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-api-owners-discuss+unsubscr...@chromium.org.

To post to this group, send email to blink-api-owners-discuss@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-api-owners-discuss/CAOtFfx4SYW25-oHACTG0sv2%3D-ZAMKQVUi3-aqnmoPD%3D0StDp%2BQ%40mail.gmail.com.

--
You received this message because you are subscribed to the Google Groups "blink-api-owners-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-api-owners-discuss+unsubscr...@chromium.org.

To post to this group, send email to blink-api-owners-discuss@chromium.org.

--
You received this message because you are subscribed to the Google Groups "blink-api-owners-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-api-owners-discuss+unsubscr...@chromium.org.

--
You received this message because you are subscribed to the Google Groups "blink-api-owners-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-api-owners-discuss+unsubscr...@chromium.org.

To post to this group, send email to blink-api-owners-discuss@chromium.org.

--
You received this message because you are subscribed to the Google Groups "blink-api-owners-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to blink-api-owners-discuss+unsub...@chromium.org.
To post to this group, send email to blink-api-owners-discuss@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-api-owners-discuss/CAOMQ%2Bw_UqowNTxQ%3Dc6b%2BKX1MoXkH-rkh3WrT_jKfc0u71R_H1A%40mail.gmail.com.

Emil A Eklund

unread,
Sep 26, 2017, 9:52:33 PM9/26/17
to Rick Byers, Chris Harrelson, Domenic Denicola, Dimitri Glazkov, Dominik Röttsches, blink-api-owners-discuss
On Wed, Sep 27, 2017 at 10:40 AM, Rick Byers <rby...@chromium.org> wrote:
Is the resulting change at all observable to the web platform?  I.e. is there any risk of it breaking something (eg. by changing text metrics)?  If not, then I think this gets blurrier (arguably not a platform change at all).  Though I suppose UA styleSheets are always observable via getComputedStyle...

No, skip-ink applies after text layout and only impacts the painting of text decorations. The default font, fallback fonts, and minimum font size (all of which are up to user or ua discretion) have a much bigger effect on rendering.
 

Regardless I'd suggest erring on the side of an intent to ship where we can hash out the details (spec or not, compat risk zero or non-zero), but I too expect totally smooth sailing.

Given that apple has been shipping this for years I don't expect much in terms of compat risks either.
Reply all
Reply to author
Forward
0 new messages