New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.
Starred by 268 users

Issue metadata

Status: Available
Merged: issue 152430
Owner: ----
EstimatedDays: ----
NextAction: 2017-09-30
OS: All
Pri: 3
Type: Feature

Blocked on: View detail
issue 591099
issue 726125

issue 463348

Participants' hotlists:

Sign in to add a comment

Issue 6606: Missing MathML implementation

Reported by, Jan 17 2009

Issue description

Chrome Version       :
URL                  :
Other browsers tested:
    Firefox 3: OK
         IE 7: OK (MathPlayer)

What steps will reproduce the problem?
1. Open a page with MathML content

What is the expected result?
You should see the correct MathML rendering.

What happens instead?
You see just a few symbol without correct rendering.
23.5 KB View Download
32.1 KB View Download
21.3 KB View Download

Comment 1 by, Jan 21 2009

Labels: -Area-Misc Area-WebKit Webkit-Specific Report-to-Webkit
Status: Untriaged
Safari 3.2: not ok

Comment 2 by, Feb 24 2009

Labels: -Report-to-Webkit Reported-to-Webkit
Status: Upstream
Bug filed upstream :

Comment 3 by, Mar 19 2009

Comment 4 by, Mar 24 2009

 Issue 9230  has been merged into this issue.

Comment 5 by, Apr 5 2009

This is a bug for almost 4 years in webkit. Can't you, chromium guys, implement this?
Because I don't think they're really working on this @ webkit. And I think it's a
really needed feature in Chromium.

Reading the thread @, we could do something like this: mathml to svg.
And SVG could be displayed correctly.

Thanks for reading!

Comment 6 by, Jul 20 2009

Labels: -Type-Bug Type-Feature Mstone-X

Comment 7 by, Nov 13 2009

I'm using Chromium to use Google Wave preview (it feels faster and more stable than 
with firefox). But I started using Google wave for Collaborate on Math stuff. There is 
nice robots there for rendering LaTeX as mathML. Too bad the MathML rendering is so 
poor on Chromium....

Comment 8 by, Aug 6 2010

Status: ExternalDependency

Comment 9 by, Sep 4 2010

This is now enabled in WebKit, should be turned on in Chromium builds as well.

Comment 10 by, Oct 20 2010

Status: Untriaged
Considering this no longer is dependent on an external dependency, seeing WebKit enabled it by default, I'm removing the label and moving this back to untriaged.

The following thread on chromium-dev is relevant to this issue:

Comment 11 by Deleted ...@, Dec 21 2010

Opera 11 is OK.

Comment 12 by, Dec 21 2010

Well, to be honest, Opera's MathML implementation is quite broken and cannot handle diacritics properly.  It is unsuitable for real world applications (tried to use it for my class notes, for which IE/MathPlayer & Firefox worked fine)

Comment 13 by, May 13 2011

Status: Available

Comment 14 by, Jun 1 2011

 Issue 84491  has been merged into this issue.

Comment 15 by, Jul 26 2011

Labels: Review-Security
Given that Safari 5.1 now supports MathML, I believe we should look into implementing this. Allegedly one of the reasons it hasn't been already is that it has not had a security audit. Marked it for review.

Comment 16 by, Jul 26 2011


Comment 17 by, Jul 26 2011

Labels: -Review-Security
@rschoen - If someone wants to take ownership of landing MathML then they can file a feature metabug and work through any changes needed to land it (including what I would anticipate to be a large number of security crashes). However, unless someone antes up for the work it's not going to happen on its own.

Comment 18 Deleted

Comment 19 Deleted

Comment 20 by, Jan 4 2012

Blocking: 22554

Comment 21 by, Jan 16 2012

So this bug seems to be just about enabling MathML parsing from comment 9 onward, but for displaying MathML you might – depending on the platform targeted – also need to think about font support, i.e. about whether you will ship with the STIX/XITS fonts on e.g. Windows (BTW which are now included in OS X 10.7 because of Safari's MathML support I would guess).

Comment 22 by, Feb 27 2012

 Issue 115809  has been merged into this issue.

Comment 23 by, Apr 21 2012

It's 2012 not 1999.

Chrome 20.0.1105.0 (Official Build 132547) dev fails the Mozilla MathML test (

Comment 24 by, Apr 27 2012

I cannot believe this.

Google accepts 2 projects for the Google Summer of Code 2012 for improving MathML support in Firefox (
but still does'nt implement MathML at all in their own Browser Chrome.

Comment 25 Deleted

Comment 26 by, Apr 28 2012

I explained on comment #17 what's preventing MathML from being enabled. Simply put, someone needs to take ownership of the feature and get it to a state where it's demonstrated stable and safe enough to ship.

Comment 27 by, Apr 28 2012

Concerning comment #17, does the remark about potentially "a large number of security crashes" still have any validity whatsoever? Does Apple do that much less testing of the code they use than Google? Not to mention that the MathML code is out in open  for a year now.

Comment 28 by, Apr 28 2012

Are you volunteering and do you have the necessary expertise to do the work? If not, I don't see that your comments are contributing anything useful at this stage. We have an obligation to our users to meet a minimum bar before enabling a feature. No amount of complaining on this bug is a substitute for meeting that obligation.

Comment 29 by, Apr 28 2012

FYI, Dave Barton is working on Webkit's MathML implementation and is concentrating his effort on fixing the most important bugs that could prevent passing security review. People who are really interested in seeing this bug fixed should follow his work on Webkit's bugzilla and possibly help him.

about comment 24: the two MathML projects mentioned were just proposals from Mozilla. Finally one student was accepted (for a different MathML project).

Comment 30 by, Oct 2 2012

 Issue 61672  has been merged into this issue.

Comment 31 by, Oct 25 2012

Blocking: -chromium:22554

Comment 32 by, Oct 25 2012

Mergedinto: 152430
Status: Duplicate
Marking as duplicate of newer  Issue 152430 

Comment 33 by, Mar 11 2013

Project Member
Labels: -Area-WebKit Cr-Content

Comment 34 by, Apr 6 2013

Project Member
Labels: -Cr-Content Cr-Blink

Comment 35 by, Dec 12 2016

Components: -Blink Blink>Layout
Labels: -Reported-to-Webkit -Pri-2 -Mstone-X mstone-X Reported-To-WebKit Pri-3
Status: Available (was: Duplicate)
Reopen.   Issue 152430  isn't a feature request.

Comment 36 by, Dec 12 2016

Google, please make an effort to fix this.

Comment 37 Deleted

Comment 38 by, Dec 12 2016

The previously heavily criticized Webkit implementation has improved quite a lot since 2013:

Thanks goes out to Frédéric Wang and the other people working on it or arguing for it, I hope Blink will be their next target (and once Blink has MathML, has Microsoft will have to follow suit now that they made it an official stance that Edge has to be on-par with Chrome + they probably need it for better EPUB support).

Comment 39 by, Dec 12 2016

Blocking: 463348

Comment 40 by, Dec 12 2016

Please add this! This would revolutionize my daily mathematical life on the Internet!

Comment 41 by, Dec 12 2016


Comment 42 by, Dec 12 2016

I join Chrome users in urging Google to advance MathML technology in order to promote universal access to scientific knowledge. Thank you.

Comment 43 by, Dec 12 2016

Please "star" this issue to vote for it, rather than leaving a comment. The star button is located at the top of the page, next to the "Issue" text.

Comment 44 by, Dec 13 2016

As mentioned, the current WebKit MathML implementation is vastly different from the one which was removed from Blink circa 2013. For an overview of the changes, see

Hopefully, this refactoring addresses the technical objections to the previous MathML implementation.

On the user-experience side, it cannot be emphasized enough that the proposed "alternative" to a native MathML implementation (namely MathJax) provides a distinctly-inferior user experience.

Comment 45 by, Jan 23 2017

Ojan/Justin, thoughts on this bug after #44? Predictability program has set an OKR to gain traction on the top 50 starred bugs this quarter: either by closing them, stating what milestone the change will ship, or setting a NextAction date so that we know when to check back in.

Comment 46 by, Jan 24 2017

eae/ikilpatrick/dknox are the best people to express opinions on this now.

Comment 47 by, Feb 14 2017

eae@, ikilpatrick@, dknox@, what are your thoughts here? Please note that this is the #8 top starred issue in Blink.

Comment 48 by, Feb 14 2017

It is something that we'd like to support eventually but only if it can be done in a maintainable and secure way.

Comment 49 by, Feb 15 2017

#48 My view is that one can very close with plain CSS -- no javascript -- and it can only get better as CSS is developed further.  Implementation of CSS flexible boxes added quite a lot of value that, I think, was not available in 2011-2012.  In my talk at TUG 2014 I showed examples, e.g.,
of math rendering using only CSS (actually, a different XML markup, but Frederic Wang has written corresponding CSS for MathML). In the write-up of the talk for Tugboat I made comments about my wishes for the development of CSS.  Those ideas are NOT specific to the needs of math.
For more:

Comment 50 by, Feb 15 2017

#49 how accessible is that XML? We rely on MathML for accessible (for users with disabilities using assistive tech) math, so students using AT are currently restricted to supporting browsers or whoever can run MathJax (which sometimes hangs/crashes pages with many complex equations (esp on phones) and does not accurately display some types of very complex equations, resulting in us rendering images of math (which don't scale) with a hidden MathML fallback as a sort of alt text).
I like the idea of being able to turn LaTeX into MathML/whatever and styled with CSS, so long as the whatever can read sensibly to screen readers and scales nicely with magnification. I don't think I can hope for speech support :(
The more browsers support math, the more choice our end-users have in which browser they can use to consume our content.

Comment 51 by, Feb 15 2017

The suggestion to drop MathML has also surfaced on the Firefox dev forum. See this post which lists some good reasons to keep it/re-enable it.!msg/

Comment 52 by, Feb 15 2017

#51 Misleading.  That thread about firefox has not been touched since May 2013.  Certainly in #49 I did NOT suggest dropping MathML.

Comment 53 by, Feb 15 2017

#50  "how accessible is that XML?"

AFAIK nothing has been done to tie it to access-related software.  The CSS provided with the example is entirely visual.  The idea with LaTeX profiles,, is that if a community with specific access concerns -- or other specialized concerns -- has its own ideal XML document type, it should be possible to provide reliable automatic translation from a LaTeX profile to that XML document type.

Comment 54 by, Feb 15 2017

Personally, I think supporting a standard makes sense from an access point
of view, because then screen readers only have a single standard to
implement, and don't have to work around workarounds that people have put
in place.

Comment 55 by, Mar 15 2017

While rendering MathML via CSS might work, it isn't the same as native
support. It also requires that one learns another library or language, or
use sometimes complicated hack to get it to do what you want. Supporting a
standard simplified the process and lowers the barrier to entry, whilst
simultaneously making sure that mathematical equations can be rendered
correctly in multiple places - without having to convert from one format to

On 15 Feb 2017 3:18 am, "gel… via monorail" <monorail+v2.4120037860@> wrote:

Comment 56 by, Mar 15 2017

#55, I don't see why you can't just use CSS to fully-support the standard.

Comment 57 Deleted

Comment 58 by, Mar 15 2017


Comment 59 by, Mar 15 2017

#57 ah I see, you need the editor. If it was just about meaning, CSS could
still do that, but yeah you need native support for a MathML editor

Comment 60 by, May 24 2017

Labels: -Webkit-Specific -mstone-X -Reported-To-WebKit
eae@ We're trying to understand the plan for all top-starred bugs.  When you say "done in a maintainable way", are you implying that this is perhaps blocked on LayoutNG (or maybe custom layout)?

I'm just hoping that we can mark this bug blocked on some other work and set a Next-Action date for when we should check back in on that work.

Comment 61 by, May 24 2017

Blockedon: 591099 726125
We (layout team and igalia) have a tentative plan to build a MathML implementation on top of LayoutNG and Custom Layout. I've marked this as blocked on CSS Custom Layout and LayoutNG.

Comment 62 by, May 31 2017

NextAction: 2017-09-30
Great, thanks!  I know LayoutNG is a big project, so setting next-action to the end of Q3 to trigger check-in for Q4.

Comment 63 by, Sep 30 2017

The NextAction date has arrived: 2017-09-30

Comment 64 by, Mar 19 2018

Regarding comment #61: Can MathML development begin now with the current state of LayoutNG and Custom Layout? Or do we need to wait these to be completely finished before MathML development can start?

Comment 65 by, May 8 2018


Comment 66 Deleted

Comment 67 by, Oct 28

I have been following this issue for years and I have to admit that the issue has been rendered irrelevant for me for the past few years by the availability of the MathJax JavaScript solution. Is this issue actually still being actively worked on?

Comment 68 by, Nov 5

I have a web components bases solution, so a pure client side rendering. 
In case anyone finds it useful:

Comment 69 by bugdroid, Feb 4

Project Member
The following revision refers to this bug:

commit 279615e68a0ea8c7a26e7de4463ddd482ccb30b2
Author: Frédéric Wang <>
Date: Mon Feb 04 13:42:27 2019

Import MathML WPT tests without running them

MathML is currently not supported by Chromium but implementation is in
progress in a downstream branch that is regularly rebased against
Chromium ToT. This downstream branch relies on MathML WPT tests. However
it is a bit tedious to manually run the import script each time new
tests are added ; or to deal with conflicts between test import commits
at each rebase. In order to facilitate the work of downstream
maintainers, this CL enables automatic import and update of MathML WPT
tests into Chromium ToT while excluding them from the list of executed
tests for now. That way one only has to perform a standard rebase
against Chromium ToT in order to synchronize MathML WPT tests.

Bug: 6606
Change-Id: I91fe90190d1934b5c111f89dde2f3a6fcd8bbfd7
Commit-Queue: Frédéric Wang <>
Reviewed-by: Philip Jägenstedt <>
Cr-Commit-Position: refs/heads/master@{#628733}

Sign in to add a comment