Deprecating SVG fonts?

686 views
Skip to first unread message

Philip Rogers

unread,
Oct 1, 2013, 7:06:25 PM10/1/13
to blink-dev
blink-dev,

I would like to gauge support for deprecating SVG fonts once we have DirectWrite on Windows (Emil is driving this in[1]). SVG fonts are currently used as a workaround for our poor Windows font rendering.

Why deprecate SVG fonts?
Due to lack of support, SVG fonts are being dropped from SVG2[2] and neither Firefox nor IE support them[3]. Some interesting work is going into adding SVG to OpenType by Mozilla[4]. I added a UseCounter for SVGFontElement and found they are used on less than 0.02% of pages. SVG fonts are poorly implemented in Blink today and users would be best served by not using them.

My current plan is to have one release between adding DirectWrite support and removing SVG fonts to give authors time to migrate their pages (http://crbug.com/242735 is tracking this work). WDYT?

Philip

35D.gif

Dirk Schulze

unread,
Oct 2, 2013, 1:17:24 AM10/2/13
to Philip Rogers, blink-dev

On Oct 2, 2013, at 1:06 AM, Philip Rogers <p...@chromium.org> wrote:

> blink-dev,
>
> I would like to gauge support for deprecating SVG fonts once we have DirectWrite on Windows (Emil is driving this in[1]). SVG fonts are currently used as a workaround for our poor Windows font rendering.

Why does the request rely on support for DirectWrite? When exactly are SVG Fonts used on Windows and how?

Greetings,
Dirk

>
> Why deprecate SVG fonts?
> Due to lack of support, SVG fonts are being dropped from SVG2[2] and neither Firefox nor IE support them[3]. Some interesting work is going into adding SVG to OpenType by Mozilla[4]. I added a UseCounter for SVGFontElement and found they are used on less than 0.02% of pages. SVG fonts are poorly implemented in Blink today and users would be best served by not using them.
>
> My current plan is to have one release between adding DirectWrite support and removing SVG fonts to give authors time to migrate their pages (http://crbug.com/242735 is tracking this work). WDYT?
>
> Philip
>
> [1] crbug.com/25541 <35D.gif>

Stephen Chenney

unread,
Oct 2, 2013, 11:02:05 AM10/2/13
to Dirk Schulze, Philip Rogers, blink-dev
On Wed, Oct 2, 2013 at 1:17 AM, Dirk Schulze <dsch...@chromium.org> wrote:

On Oct 2, 2013, at 1:06 AM, Philip Rogers <p...@chromium.org> wrote:

> blink-dev,
>
> I would like to gauge support for deprecating SVG fonts once we have DirectWrite on Windows (Emil is driving this in[1]). SVG fonts are currently used as a workaround for our poor Windows font rendering.

Why does the request rely on support for DirectWrite? When exactly are SVG Fonts used on Windows and how?


People have been using SVGFonts to get high quality font rendering on Windows, because the current Chrome version uses GDI for standard fonts with resulting poor quality. We're moving to DirectWrite, which in theory should remove the need to use SVGFonts in most cases.

Stephen.

PhistucK

unread,
Oct 2, 2013, 11:36:25 AM10/2/13
to Stephen Chenney, Dirk Schulze, Philip Rogers, blink-dev
Is DirectWrite supported on Windows XP?
If it is not, would there actually be no way of having high quality font rendering there?


PhistucK


To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+...@chromium.org.

Philip Rogers

unread,
Oct 2, 2013, 2:50:21 PM10/2/13
to PhistucK, Stephen Chenney, Dirk Schulze, blink-dev
PhistucK,

You're right that XP does not have directwrite support. The workaround for nice fonts on Windows using SVG fonts is not used often (less than 0.02%) so I'm not too worried but it's certainly a concern. XP users are upgrading so I hope this is only a minor issue by the time we actually deprecate.

Philip

Dirk Schulze

unread,
Oct 3, 2013, 1:53:00 AM10/3/13
to Philip Rogers, PhistucK, Stephen Chenney, blink-dev

On Oct 2, 2013, at 8:50 PM, Philip Rogers <p...@chromium.org> wrote:

> PhistucK,
>
> You're right that XP does not have directwrite support. The workaround for nice fonts on Windows using SVG fonts is not used often (less than 0.02%) so I'm not too worried but it's certainly a concern. XP users are upgrading so I hope this is only a minor issue by the time we actually deprecate.
>

How do you get to the conclusion that the 0.02% (which is indeed not much) is just from Windows usage?

Greetings,
Dirk

Philip Rogers

unread,
Oct 3, 2013, 1:56:18 AM10/3/13
to Dirk Schulze, PhistucK, Stephen Chenney, blink-dev
Dirk,

The 0.02% figure is not just Windows usage, it's total usage across all platforms.

Elliott Sprehn

unread,
Oct 3, 2013, 1:59:21 AM10/3/13
to Philip Rogers, Dirk Schulze, PhistucK, Stephen Chenney, blink-dev
If we removed it would pages break? I assume we'd fall back to normal fonts. That number is so low for a feature with this much complexity that no one else has plans to support. Removing it sounds great to me.

Dirk Schulze

unread,
Oct 3, 2013, 3:32:13 AM10/3/13
to Elliott Sprehn, Philip Rogers, PhistucK, Stephen Chenney, blink-dev

On Oct 3, 2013, at 7:59 AM, Elliott Sprehn <esp...@chromium.org> wrote:

> If we removed it would pages break? I assume we'd fall back to normal fonts. That number is so low for a feature with this much complexity that no one else has plans to support. Removing it sounds great to me.

Right, text would fall back to default fonts. Main problem is maybe art work created with Illustrator or other programs which maybe is not hosted on web pages. These graphics sometimes use SVG Fonts embedded in the SVG file and definitely would be harmed by fall back fonts. I have no numbers available how often that happens but think it is just a very small percentage. Illustrator allows SVG Fonts but does not select SVG Fonts by default.

The lack of support by other main browsers and design flaws that no one is willing to fix is indeed a problem for SVG Fonts. As a contributor to Blink's SVG Font code I tend to agree that it might not be worth the maintaining cost.

SVG Fonts were the only way to have web fonts on iOS in the first years. So there might still be legacy content using it (which could be of interest for the Android platform). This should be taken into consideration as well.

Greetings,
Dirk

Rik Cabanier

unread,
Oct 3, 2013, 7:06:49 PM10/3/13
to Elliott Sprehn, Philip Rogers, Dirk Schulze, PhistucK, Stephen Chenney, blink-dev
+1

It doesn't help the web platform to have features that are not implemented across browsers.
OpenType is the accepted way to do fonts on the web.

Erik Dahlstrom

unread,
Oct 4, 2013, 9:59:12 AM10/4/13
to blin...@chromium.org
On Wed, 02 Oct 2013 01:06:25 +0200, Philip Rogers <p...@chromium.org> wrote:

> blink-dev,
>
> I would like to gauge support for deprecating SVG fonts once we have
> DirectWrite on Windows (Emil is driving this in[1]). SVG fonts are
> currently used as a workaround for our poor Windows font rendering.
>
> Why deprecate SVG fonts?
> Due to lack of support, SVG fonts are being dropped from SVG2[2] and
> neither Firefox nor IE support them[3].

The issues are also mentioned in the SVG2 draft,
https://svgwg.org/svg2-draft/fonts.html.

> Some interesting work is going into
> adding SVG to OpenType by Mozilla[4]. I added a UseCounter for
> SVGFontElement and found they are used on less than 0.02% of pages.

It probably won't make a dramatic difference, but note that it's not
necessary for SVGFontFaceElement to be a child of an
SVGFontElement to be usable in the same way as an SVGFontElement.

> SVG fonts are poorly implemented in Blink today and users would be best
> served by not using them.

I'll miss having a simple, reliable* cross-platform font format, easily
scriptable and human-readable, independent from the OS's font engine and
supported by the browser. SVG fonts were unique in that sense.

That said, I agree with your conclusion (particularly given the
performance implications), and won't object to the removal.
/ed

* reliable in the cross-platform rendering sense, and (obviously) not in
the cross-browser sense (due to IE / Firefox). Normal platform fonts (and
webfonts actually) do look different enough across platforms due to what
text rendering engine is used that the only really dependable way is using
pure path data (converting all glyphs -> svg paths), at least if you want
smooth scaling animations and such (and no, 'text-rendering:
geometricPrecision' didn't always help).

--
Erik Dahlstrom, Core Technology Developer, Opera Software
Co-Chair, W3C SVG Working Group
Personal blog: http://my.opera.com/macdev_ed

Philip Rogers

unread,
Oct 4, 2013, 4:33:11 PM10/4/13
to Erik Dahlstrom, blink-dev
Last night I discovered my original measurement of SVGFontElement contained a bug and didn't properly measure SVG fonts referenced via CSS. I've filed a bug and uploaded a new patch for properly measuring SVG font usage: http://crbug.com/304219.

I apologize for this embarrassing mistake. Fortunately, we have plenty of time to get solid numbers before revisiting this feature. I will update this thread once we have correct UseCounter numbers from the stable channel.

Philip

Kouhei Ueno

unread,
Feb 25, 2014, 10:35:16 PM2/25/14
to Philip Rogers, Erik Dahlstrom, blink-dev
Any updates on this? I really want this feature to be removed for code sanity...
The latest use counter shows 0.015%.



To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+...@chromium.org.



--
Kouhei Ueno

Erik Dahlström

unread,
Feb 26, 2014, 3:35:16 AM2/26/14
to Philip Rogers, Kouhei Ueno, blink-dev
I think we should wait with the removal until non-svg webfonts render
properly on Windows, http://crbug.com/137692 (which is blocked by
http://crbug.com/25541).

SVGFontInCSS:
http://www.chromestatus.com/metrics/feature/timeline/popularity/156
(0.2041%)

/ed
Erik Dahlstrom, Web Technology Developer, Opera Software

Emil A Eklund

unread,
Feb 26, 2014, 11:20:05 AM2/26/14
to Erik Dahlström, Philip Rogers, Kouhei Ueno, blink-dev
On Wed, Feb 26, 2014 at 12:35 AM, Erik Dahlström <e...@opera.com> wrote:
> I think we should wait with the removal until non-svg webfonts render
> properly on Windows, http://crbug.com/137692 (which is blocked by
> http://crbug.com/25541).

That is indeed the plan.
Reply all
Reply to author
Forward
0 new messages