1761586 - Inspiration font "rev=1" is rejected by sanitizer at https://fonts.google.com/ (but the "rev=2" version on the font-specific page is fine)

Status: ASSIGNED

(Web Compatibility :: Site Reports, defect, P3)

Description

[Daniel Holbert [:dholbert]]

Updated STR from Jan 2023:

1. Visit https://fonts.google.com/?query=inspiration
2. Look inside the preview box

(I've edited this comment to replace some older STR from 2022, which were dependent on the fact that this font was featured on the front page of https://fonts.google.com/. This font no longer seems to be featured there, but it does show up if you search for it like this.)

ACTUAL RESULTS:
The characters are missing (and used to be shown as blank rectangles in 2022).

Web Console shows a bunch of errors associated with this font:

downloadable font: Layout: Bad lookahead count 71 in chain class rule subtable (font-family: "Inspiration script=latin rev=1" style:normal weight:400 stretch:100 src index:0) source: https://fonts.gstatic.com/l/font?kit=x3dkckPPZa6L4wIg5cZOEvoDhTjNKS1yegVZ6QnJSKQfRrxhOgPge58T&skey=426336bee49760d9&v=v1
downloadable font: Layout: Failed to parse chain class rule 0 in chain class set (font-family: "Inspiration script=latin rev=1" style:normal weight:400 stretch:100 src index:0) source: https://fonts.gstatic.com/l/font?kit=x3dkckPPZa6L4wIg5cZOEvoDhTjNKS1yegVZ6QnJSKQfRrxhOgPge58T&skey=426336bee49760d9&v=v1
downloadable font: Layout: Failed to parse chain class set table 1 in chain context format 2 (font-family: "Inspiration script=latin rev=1" style:normal weight:400 stretch:100 src index:0) source: https://fonts.gstatic.com/l/font?kit=x3dkckPPZa6L4wIg5cZOEvoDhTjNKS1yegVZ6QnJSKQfRrxhOgPge58T&skey=426336bee49760d9&v=v1
downloadable font: Layout: Failed to parse chaining context format 2 subtable (font-family: "Inspiration script=latin rev=1" style:normal weight:400 stretch:100 src index:0) source: https://fonts.gstatic.com/l/font?kit=x3dkckPPZa6L4wIg5cZOEvoDhTjNKS1yegVZ6QnJSKQfRrxhOgPge58T&skey=426336bee49760d9&v=v1
downloadable font: Layout: Failed to parse lookup subtable 5 (font-family: "Inspiration script=latin rev=1" style:normal weight:400 stretch:100 src index:0) source: https://fonts.gstatic.com/l/font?kit=x3dkckPPZa6L4wIg5cZOEvoDhTjNKS1yegVZ6QnJSKQfRrxhOgPge58T&skey=426336bee49760d9&v=v1
downloadable font: Layout: Failed to parse subtable 0 (font-family: "Inspiration script=latin rev=1" style:normal weight:400 stretch:100 src index:0) source: https://fonts.gstatic.com/l/font?kit=x3dkckPPZa6L4wIg5cZOEvoDhTjNKS1yegVZ6QnJSKQfRrxhOgPge58T&skey=426336bee49760d9&v=v1
downloadable font: Layout: Failed to parse lookup 2 (font-family: "Inspiration script=latin rev=1" style:normal weight:400 stretch:100 src index:0) source: https://fonts.gstatic.com/l/font?kit=x3dkckPPZa6L4wIg5cZOEvoDhTjNKS1yegVZ6QnJSKQfRrxhOgPge58T&skey=426336bee49760d9&v=v1
downloadable font: GSUB: Failed to parse lookup list table (font-family: "Inspiration script=latin rev=1" style:normal weight:400 stretch:100 src index:0) source: https://fonts.gstatic.com/l/font?kit=x3dkckPPZa6L4wIg5cZOEvoDhTjNKS1yegVZ6QnJSKQfRrxhOgPge58T&skey=426336bee49760d9&v=v1
downloadable font: GSUB: Failed to parse table (font-family: "Inspiration script=latin rev=1" style:normal weight:400 stretch:100 src index:0) source: https://fonts.gstatic.com/l/font?kit=x3dkckPPZa6L4wIg5cZOEvoDhTjNKS1yegVZ6QnJSKQfRrxhOgPge58T&skey=426336bee49760d9&v=v1
downloadable font: rejected by sanitizer (font-family: "Inspiration script=latin rev=1" style:normal weight:400 stretch:100 src index:0) source: https://fonts.gstatic.com/l/font?kit=x3dkckPPZa6L4wIg5cZOEvoDhTjNKS1yegVZ6QnJSKQfRrxhOgPge58T&skey=426336bee49760d9&v=v1

EXPECTED RESULTS:
Font should load (or if it's corrupt/broken somehow, we should perhaps ask Google to get it fixed).

Notably, Firefox release 98.0.2 (from Mozilla) gives EXPECTED RESULTS with no issues reported in web console. But mozregression-spawned Nightlies give ACTUAL RESULTS as far back as they can successfully load Google Fonts; e.g. Nightly 2020-08-03 (v81.0a1) gives ACTUAL RESULTS. So there's some Nightly-specific feature involved here.

Also interestingly: in an affected Nightly, if I click the box to go to the font-specific page ( or just directly visit https://fonts.google.com/specimen/Inspiration ), then the font renders just fine, with no boxes and no issues shown in web console. It's only on the front page at https://fonts.google.com/ where I'm seeing the problem.

jfkthame, do you know what might be going on here / is this a known issue? FWIW I'm encountering this on my work thinkpad running Ubuntu 21.10, and I also see it in a vanilla Ubuntu 21.04 VM running in gnome-boxes, so I don't think there's a specific system-config option that I've somehow activated to trigger this.

Comment 1

[Daniel Holbert [:dholbert]]

Also interesting:

If I start at https://fonts.google.com/, I see the issue; then if I click through to the font-specific page, it loads properly; then if I navigate back, it still loads properly (on the previously-affectedfront page). Then if I reload, it fails again.

Comment 2

[Daniel Holbert [:dholbert]]

Attachment: screencast of bug (also showing that it goes away on back-navigation from font-specific page)

Comment 3

[Daniel Holbert [:dholbert]]

(I'm seeing the same issue in Nightly on MacOS 12.3, too, FWIW.)

Comment 4

[Daniel Holbert [:dholbert]]

(And in Firefox Nightly on Windows 10 and Android.)

Comment 5

[Jonathan Kew [:jfkthame]]

The reason this is Nightly-only is that gfx.downloadable_fonts.otl_validation is true on Nightly, but false on release channels.

OTS is reporting that one of the OpenType layout tables (GSUB) is broken, and therefore (when validation is enabled) the font is rejected. So yes, this is something that should be fixed by the Google Fonts folk.

As for why it fails on the initial index page, but then works after visiting the font-specific page, it looks like there are two different versions of the font. On the main index page, the Inspiration sample uses font-family: "Inspiration script=latin rev=1", which refers to the rule

@font-face {
  font-family: 'Inspiration script=latin rev=1';
  font-style: normal;
  font-weight: 400;
  font-display: block;
  src: url(https://fonts.gstatic.com/l/font?kit=x3dkckPPZa6L4wIg5cZOEvoDnSzlBTElO0sTzQ7MRKgSTblmPh3-YYUWROi5lQ&skey=426336bee49760d9&v=v1) format('woff2');
}

and the resulting resource is invalid. But the font-specific page uses font-family: "Inspiration script=latin rev=2", which loads a different resource:

@font-face {
  font-family: 'Inspiration script=latin rev=2';
  font-style: normal;
  font-weight: 400;
  font-display: block;
  src: url(https://fonts.gstatic.com/s/inspiration/v1/x3dkckPPZa6L4wIg5cZOEsoBlyvHBwsz.woff2) format('woff2');
  unicode-range: U+0000-00FF, U+0131, U+0152-0153, U+02BB-02BC, U+02C6, U+02DA, U+02DC, U+2000-206F, U+2074, U+20AC, U+2122, U+2191, U+2193, U+2212, U+2215, U+FEFF, U+FFFD;
}

and this works fine. Then when returning to the index page, it also uses the rev=2 version of Inspiration (i.e. it appears to get updated CSS applied).

Given the rev=1 vs rev=2 tags here, I'm guessing there may have been an update to the font at some point (to fix this issue, maybe), but the original index page is still using the old (broken) version.

cc'ing Dave Crossland here: Dave, this looks like a Google Fonts bug, delivering a font resource with a broken GSUB table (though it's masked for browsers that don't validate the OTL tables). Can you look into getting it fixed?

Comment 6

[BugBot [:suhaib / :marco/ :calixte]]

Redirect a needinfo that is pending on an inactive user to the triage owner.
:lsalzman, since the bug has recent activity, could you have a look please?

For more information, please visit auto_nag documentation.

Comment 7

[Daniel Holbert [:dholbert]]

This is still an issue FWIW, though it manifests slightly differently now:

• Inspiration isn't featured as the 5th font on the front page anymore, so it's not as front-and-center of a problem.
• If I search Google Fonts for Inspiration (with gfx.downloadable_fonts.otl_validation at its default true value in Nightly), then I see a preview box briefly appear and then Google Fonts shows an error page:

(^_^)b
Something went wrong. Please try again soon.

(with the same errors from comment 0 -- downloadable font: Layout: Bad lookahead count 71 etc -- appearing in my JS console)

However, the direct link to the font's Google Fonts page loads just fine, because it's using a different version of the Inspiration font.

Looking at our font devtools, it looks like the broken search-results page is using Inspiration rev=1 with this URL:
https://fonts.gstatic.com/l/font?kit=x3dkckPPZa6L4wIg5cZOEvoDhT7NUmJQfwJb6wfHTqIGWaV2&skey=426336bee49760d9&v=v3

...vs. the working-just-fine font page is using Inspiration rev=2 with this URL:
https://fonts.gstatic.com/s/inspiration/v3/x3dkckPPZa6L4wIg5cZOEsoBlys.woff2

(rev=1 / rev=2 are literally in the "font-family" styling on these elements, too. On the broken search-results page, the preview element literally has font-family: "Inspiration rev=1" in its style attribute.)

Comment 8

[Daniel Holbert [:dholbert]]

Oh, I guess the rev=1 (broken) vs. rev=2 (working) thing isn't new; jfkthame mentioned it in comment 5 as well.

I guess at this point we should recategorize this as a Web Compat bug.

Comment 9

[Calin Tanase]

Attachment: Inspiration font.png

That's still an issue. The characters from preview are no longer replaced by rectangles, the preview is blank instead.

Tested on:
Operating system: Windows 10
Browser/Version: Firefox Nightly 111.0a1 (2023-01-31) / Firefox Release 109.0.1 / Chrome 109.0.5414.75

Note: Not reproducible on Firefox Release and Chrome

EDIT: Tried to perform regression but it seems to go way back

Comment 10

[Daniel Holbert [:dholbert]]

(In reply to Calin Tanase from comment #9)

That's still an issue. The characters from preview are no longer replaced by rectangles, the preview is blank instead.

Indeed, still an issue. (Also for what it's worth, I'm updating the STR in comment 0, since my original STR assumed that this font was on the front page of Google Fonts -- and that's no longer the case.)

Note: Not reproducible on Firefox Release [...]
EDIT: Tried to perform regression but it seems to go way back

Yup, this is explained by first sentence of comment 5. (We do extra validation only-in-Nightly, which causes this font to be rejected.)