-
Notifications
You must be signed in to change notification settings - Fork 61
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Recommended technique for Math formulas in EPUB #845
Comments
Additional offscreen design proposal: https://rawgit.com/mhakkinen/extdesc/master/Extended-desc.md.html |
MathJax's Peter K. 's thoughts on MathML Media Query: |
An old issue...but perhaps a good time to mention the more recent DAISY proposals: |
What about this? It's based on 2a and the
|
Tested, it works in both Calibre and on my Kobo. ePubCheck complains that |
Guess what I'm suggesting is option 2a but add an Screen readers can be triggered by an |
sorry I haven't finished my coffee. Option 2a with the |
Hello,
Attached is a test epub with the various options we are testing for screen reader users, low vision users, persons who are dyslexic, and for mainstream folks. We want to come up with a best practice.
I think we have one more test to add before we start our extensive testing.
I am liking the test 020, but we have some issues with some systems. We are trying to track that down.
Best
George
From: VeryGoodErotica <notifications@github.com>
Sent: Friday, March 27, 2020 8:30 AM
To: w3c/publ-epub-revision <publ-epub-revision@noreply.github.com>
Cc: Subscribed <subscribed@noreply.github.com>
Subject: Re: [w3c/publ-epub-revision] Recommended technique for Math formulas in EPUB (#845)
sorry I haven't finished my coffee.
Option 2a with the epub:type="math" and an aria-label="MathSpeak string" in the object node.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub <#845 (comment)> , or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABW4OSCPTG2Q5VDFXMIN7PDRJSZ7VANCNFSM4COMKJDA> .
|
Dear George,
Sorry, I don’t see any attachment ?
Luc
De : George <notifications@github.com>
Répondre à : w3c/publ-epub-revision <reply@reply.github.com>
Date : vendredi 27 mars 2020 à 16:51
À : w3c/publ-epub-revision <publ-epub-revision@noreply.github.com>
Cc : Subscribed <subscribed@noreply.github.com>
Objet : Re: [w3c/publ-epub-revision] Recommended technique for Math formulas in EPUB (#845)
Hello,
Attached is a test epub with the various options we are testing for screen reader users, low vision users, persons who are dyslexic, and for mainstream folks. We want to come up with a best practice.
I think we have one more test to add before we start our extensive testing.
I am liking the test 020, but we have some issues with some systems. We are trying to track that down.
Best
George
From: VeryGoodErotica <notifications@github.com>
Sent: Friday, March 27, 2020 8:30 AM
To: w3c/publ-epub-revision <publ-epub-revision@noreply.github.com>
Cc: Subscribed <subscribed@noreply.github.com>
Subject: Re: [w3c/publ-epub-revision] Recommended technique for Math formulas in EPUB (#845)
sorry I haven't finished my coffee.
Option 2a with the epub:type="math" and an aria-label="MathSpeak string" in the object node.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub <#845 (comment)> , or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABW4OSCPTG2Q5VDFXMIN7PDRJSZ7VANCNFSM4COMKJDA> .
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub<https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fw3c%2Fpubl-epub-revision%2Fissues%2F845%23issuecomment-605074262&data=02%7C01%7Claudrain%40hachette-livre.fr%7C480bf68887e447d9f24008d7d266aea4%7Cf881a2c50a89483181b1c7846c49594d%7C0%7C0%7C637209210779534350&sdata=kENf2pfk0twyQrT5XpTI9veeKuCDHd%2Bti0dFFSkMkgc%3D&reserved=0>, or unsubscribe<https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAECB46XCLVV73M7QCH5BPGTRJTDPDANCNFSM4COMKJDA&data=02%7C01%7Claudrain%40hachette-livre.fr%7C480bf68887e447d9f24008d7d266aea4%7Cf881a2c50a89483181b1c7846c49594d%7C0%7C0%7C637209210779534350&sdata=Z5jVBu3gA2vm7bd3bAYuuvIexOYs7aeP3e4X2GbW%2B8A%3D&reserved=0>.
|
Hello,
If the mail server stripped out the attached EPUB, we have the math test book at:
http://www.epubtest.org/test-books
The advanced test book is in process of being revised, but this is a start. We hope to have the revised version available soon.
Best
George
From: Luc Audrain <notifications@github.com>
Sent: Friday, March 27, 2020 11:16 AM
To: w3c/publ-epub-revision <publ-epub-revision@noreply.github.com>
Cc: George <kerscher@montana.com>; Comment <comment@noreply.github.com>
Subject: Re: [w3c/publ-epub-revision] Recommended technique for Math formulas in EPUB (#845)
Dear George,
Sorry, I don’t see any attachment ?
Luc
De : George <notifications@github.com>
Répondre à : w3c/publ-epub-revision <reply@reply.github.com>
Date : vendredi 27 mars 2020 à 16:51
À : w3c/publ-epub-revision <publ-epub-revision@noreply.github.com>
Cc : Subscribed <subscribed@noreply.github.com>
Objet : Re: [w3c/publ-epub-revision] Recommended technique for Math formulas in EPUB (#845)
Hello,
Attached is a test epub with the various options we are testing for screen reader users, low vision users, persons who are dyslexic, and for mainstream folks. We want to come up with a best practice.
I think we have one more test to add before we start our extensive testing.
I am liking the test 020, but we have some issues with some systems. We are trying to track that down.
Best
George
From: VeryGoodErotica <notifications@github.com>
Sent: Friday, March 27, 2020 8:30 AM
To: w3c/publ-epub-revision <publ-epub-revision@noreply.github.com>
Cc: Subscribed <subscribed@noreply.github.com>
Subject: Re: [w3c/publ-epub-revision] Recommended technique for Math formulas in EPUB (#845)
sorry I haven't finished my coffee.
Option 2a with the epub:type="math" and an aria-label="MathSpeak string" in the object node.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub <#845 (comment)> , or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABW4OSCPTG2Q5VDFXMIN7PDRJSZ7VANCNFSM4COMKJDA> .
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub<https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fw3c%2Fpubl-epub-revision%2Fissues%2F845%23issuecomment-605074262&data=02%7C01%7Claudrain%40hachette-livre.fr%7C480bf68887e447d9f24008d7d266aea4%7Cf881a2c50a89483181b1c7846c49594d%7C0%7C0%7C637209210779534350&sdata=kENf2pfk0twyQrT5XpTI9veeKuCDHd%2Bti0dFFSkMkgc%3D&reserved=0>, or unsubscribe<https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAECB46XCLVV73M7QCH5BPGTRJTDPDANCNFSM4COMKJDA&data=02%7C01%7Claudrain%40hachette-livre.fr%7C480bf68887e447d9f24008d7d266aea4%7Cf881a2c50a89483181b1c7846c49594d%7C0%7C0%7C637209210779534350&sdata=Z5jVBu3gA2vm7bd3bAYuuvIexOYs7aeP3e4X2GbW%2B8A%3D&reserved=0>.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub <#845 (comment)> , or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABW4OSGEX44KDACPCFLNZJLRJTNMHANCNFSM4COMKJDA> .
|
Thank you very much for those working on this. The solution I came up is currently working in the ePub viewers I personally have access to, but I definitely would prefer a method with proper accessibility testing behind it. |
Until we have a technique that reasonably works everywhere, I don't think there's anything we can do with this issue. I'm going to put a deferred label on it as I don't see it being resolved in our current charter timeline. We can return to it whenever there's more clarity (assuming mathml support doesn't overtake the need for this). |
There used to be a discussion regarding MathML.
I think most reading applications does not support it.
…--
Ori Idan CEO Helicon Books
http://www.heliconbooks.com
On Mon, Mar 8, 2021 at 10:31 PM Matt Garrish ***@***.***> wrote:
Until we have a technique that reasonably works everywhere, I don't think
there's anything we can do with this issue. I'm going to put a deferred
label on it as I don't see it being resolved in our current charter
timeline. We can return to it whenever there's more clarity (assuming
mathml support doesn't overtake the need for this).
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#845 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAB43QGU4XY66QEI7DMCAC3TCUXYZANCNFSM4COMKJDA>
.
|
We developed Math techniques and have done extensive testing on it. Which will improve further with Chrome's support for MathML. And explanation of the techniques is at: So, we can consider closing this issue. |
@avneeshsingh It appears that your report is, to a large extent, focused on accessible maths, which is undeniably important, but also not nearly the whole problem. And... Do I understand correctly that the proposed solution for reasonably supported maths in EPUB books is... PNG images with no possibility of progressive enhancement? (OK, PNG images with I can understand that as a faithful representation of the current state of the reader world, but it’s still just so sad. |
It is. But there is some light in the tunnel. W3C is discussing setting up a new MathML Working Group whose main goal is to consolidate a profile of MathML that can, and will be, available in browsers. If that happens, then the PNG fallback become less prominent. Fingers crossed :-) |
@iherman wrote:
I’m not even particularly against a PNG fallback; I mean, having several representations of the same formula isn’t pretty and does add some bloat, but that’s what tooling is for. However, the discussion above leads me to believe that having a reasonably compatible, no-scripting-required MathML to SVG to PNG fallback isn’t possible with current readers, not with Preferably such a fallback would allow for yet another alternative representation targeted at screen readers (although, TBH, a significant portion of maths that I care about best exists in a two-dimensional form—commutative diagrams, deduction trees, even similar algebraic expressions with related terms lined up—and I’ve never seen coherent guidelines as to how deal with these in an accessible way). Ideally it would even allow an additional, separate representation for copying (AsciiMath or TeX code, probably). But so far even the non-accessible fallback described above seems impossible; I’ve subscribed to this thread to see myself proven wrong on this count and haven’t succeeded so far. :( |
Closing this issue, but feel free to continue to discuss. The issue here is generally poor support for MathML and we can't solve that in the techniques at this time. |
Is there any status update on this? There’s a discussion Best Practices for Authoring MathML in EPUB from DAISY but I can’t find a definite guideline or recommendation? |
@jenstroeger In addition to the best practices document that you mentioned above, DAISY knowledge base also has a section for MathML. |
Thank you @avneeshsingh — so what’s the recommendation today to target current and likely dated reading systems? The equation in my particular case is a simple inlined E=mc2 which I could just note down/mark up as <span role="math">E = mc<sup>2</sup></span> to ensure that current reading systems can reliably handle it? |
It would be interesting to know that more than 50% of the reading systems that we tested support MathML.
This change was visible after Chrome and Edge started supporting MathML natively. Because many reading systems are built on these browsers, MathML support started to grow.
So, I think that it would be OK if you keep the equation in MathML 4.
|
Do you know how those “more than 50% tested” correlate to the devices in use out in the wild? What if 80% (just making up a number) of the devices in use make up the other 50%? 🤔 |
In coming months, we will be posting accessibility testing results of popular reading systems on https://epubtest.org/results
It will provide more accurate information.
|
Hi,
We will know more, but I suspect that of the Reading Systems used in Higher Education that we may get close to 100%.
Best
George
From: Avneesh Singh ***@***.***>
Sent: Saturday, May 18, 2024 5:00 AM
To: w3c/epub-specs ***@***.***>
Cc: George ***@***.***>; Comment ***@***.***>
Subject: Re: [w3c/epub-specs] Recommended technique for Math formulas in EPUB (#845)
In coming months, we will be posting accessibility testing results of popular reading systems on https://epubtest.org/results
It will provide more accurate information.
—
Reply to this email directly, view it on GitHub <#845 (comment)> , or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABW4OSBUXZG56UZJ2CIOIBDZC4YCDAVCNFSM4COMKJDKU5DIOJSWCZC7NNSXTN2JONZXKZKDN5WW2ZLOOQ5TEMJRHA3TOOBXGUYQ> .
You are receiving this because you commented.Message ID: ***@***.***>
|
Hi @jenstroeger, we have discussed this possible approach (that looked promising) in both the DAISY TIES group and W3C. Using |
Hi,
I just tested with NVDA and the span was not read. It is because of the role of math.
Best
George
From: Gregorio Pellegirno ***@***.***>
Sent: Sunday, May 19, 2024 3:25 AM
To: w3c/epub-specs ***@***.***>
Cc: George ***@***.***>; Comment ***@***.***>
Subject: Re: [w3c/epub-specs] Recommended technique for Math formulas in EPUB (#845)
The equation in my particular case is a simple inlined E=mc2 which I could just note down/mark up as
<span role="math">E = mc<sup>2</sup></span>
Hi @jenstroeger <https://github.com/jenstroeger> , we have discussed this possible approach (that looked promising) in both the DAISY TIES group and W3C. Using role="math" to mark formulas in plain text does not seem like a good solution. The direction we give now, even for inline formulas, is to transform them to MathML.
—
Reply to this email directly, view it on GitHub <#845 (comment)> , or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABW4OSBMF53WZAUATLHELZDZDBVWDAVCNFSM4COMKJDKU5DIOJSWCZC7NNSXTN2JONZXKZKDN5WW2ZLOOQ5TEMJRHEYTMNZQGAZQ> .
You are receiving this because you commented.Message ID: ***@***.***>
|
Recent discussions (EPUB a11y call on August 24 + follow-up thread on the mailing list) have revived the longtime question of how best to represent math formulas in EPUB. This issue intends to track the various discussions and solution candidates.
Background
The issue of math formulas in EPUB consists of at least two distinct (but related) challenges:
Historically, EPUB has promoted MathML as the preferred way to represent math. However, MathML is poorly supported by numerous Reading Systems. Many of them do not even parse the MathML to render the fallback image defined in MathML's
altimg
attribute. Many content publishers –even the ones that can produce MathML– end up commenting out the MathML and just use images of formulas instead.It must also be noted that MathML is considered by some experts as a failing appraoch to represent math on the web (see notably @pkra's thoughts on the topic).
As for accessibility, the situation is mixed. For some very simple formulas, simple alternative text is probably enough. For more complex content, MathML is reasonably well supported by major screen readers (sometimes with 3d party libraries like MathPlayer), despite being limited by the inherent lack of semantics in presentation MathML. Some specialized JS-based renderers (like MathJax's accessibility extensions) may also be relevant alternatives.
All in all, there is no ideal solution. What is needed is one (or several?) best practices that can be recommended to content publishers in the short-to-mid term.
Candidate techniques
The constraints are:
Option 1 -data-mathml
atttributeThe formula is represented with an
img
element, with alternative text in itsalt
attribute. Additionally, the equivalent MathML markup is available (escaped) in adata-math
attribute that can be used by a javascript renderer.Example:
Pros:
Cons:
data-math
attribute is a non-standard used of the HTML spec if it is intended for use by the RS (it's probably OK if the javascript is provided by the Author).Option 2a -
object
element with MathML fallbackThe formula is represented as an image in the
object
element, for which an alternative text can be defined with thearia-label
attribute. Theobject
can have MathML child content, which could then be used as a fallback by the RS or AT that prefers it.Example:
Pros:
Cons:
object
's fallback content can be shown only "if the user has indicated a preference" (for instance if "that content uses a format that the user finds more accessible").Option 2b -object
element with image fallbackThe formula is represented as an external MathML document embedded with an
object
element, for which an alternative text can be defined with thearia-label
attribute. Theobject
can have an image as its child content, which could then be used as a fallback by the RS.Example:
Pros:
Cons:
Option 3 - Media Query
A new Media Query "detecting" MathML support. We could then assign
display:
none' to the math element and a non-none display to an image fallback sibling for legacy reading systems and have that new MQ assign a non-none display value to the math element anddisplay: none
to the image fallback sibling if MathML is implemented.See @therealglazou's proposal.
Example:
Pros:
Cons:
Option 4 -aria-details
The formula is rendered with an image, and the MathML is referenced from an
aria-details
attribute on the image.Example:
Option 5 - custom classes in package metadata
Special CSS classes for both MathML and non-MathML formulas are declared in Package metadata. Reading systems that support MathML would override the display property to render the MathML instead of the image.
Example:
Content Doc:
CSS:
Package:
A MathML-aware RS would override the CSS based on the values defined in Metadata above:
Other references
The text was updated successfully, but these errors were encountered: