@Beta16 The feature isn't documented on https://www.mediawiki.org/wiki/Extension:Math/advancedSettings, could you update the doc?

In T120938#2157576, @Dereckson wrote:

@Beta16 The feature isn't documented on https://www.mediawiki.org/wiki/Extension:Math/advancedSettings, could you update the doc?

I'm not a developer of the extension Math.
Maybe @Physikerwelt or @mobrovac can update the documentation.

What exactly is this feature?

I see references in T113708 but not in the source code of the extension.

$ cd extensions/Math
$ ag speechRule
(nothing)
$ ag speakText
(nothing)

Would that be linked to http://www.gh-mathspeak.com/examples/grammar-rules/ and Web speech API?

@mobrovac ping ?

Speech help hints are generated by the Mathoid service, not the extension itself (it just relays whatever Mathoid generates). It is currently disabled for performance reasons, but if there is enough support to have it, we can probably enable it.

@Beta16 I agree that this is very handy for accessibility but we will need evidence that community support exists for this before considering enabling it. Can you provide us with a page/document that shows some kind of community discussion?

/cc @Physikerwelt @Jdforrester-WMF , thoughts from your end?

@mobrovac I think enabeling speech support should follow T78046

by the way... if someone (TM) would review https://gerrit.wikimedia.org/r/282175 it would no longer block the issue with the special page

@Dereckson , @Danny_B , why does this specific accessibility improvement need a community support discussion? Surely we have broad community support for any reasonable improvement that helps our blind readers and editors. Is this not a reasonable improvement? cost/benefit..? I assume all possible renderings include some sort of alt text ; so is the concern that a spoken version isnt strictly necessary for accessibility?

@mobrovac , what are the performance reasons you mention? I'm assume you mean server-side performance, and not increased network traffic or client-side browser processing time, or any of myriad other types of performance reasons.
Is this performance reason serious? how bad does the performance degrade? Will users somehow experience this performance problem (e.g. slower render), or is something only the ops team will 'see' and fix/manage by adding more processors where/when/as required?

@jayvdb I didn't add the tag. I just sorted the task on board.

@Zorkow long time no see...are you still actively developig the speech rule engine? Do you think it would be advisable to reconsider using its output in production (i.e, alt-texts)?

@Zorkow long time no see...

I wonder why that is...

are you still actively developig the speech rule engine?

Yes, people seem to use it and some fund me to continue developing it.
v3.2 should be out in beta soon.
We've got 5 European languages now and Hindi is in preparation.

Do you think it would be advisable to reconsider using its output in production (i.e, alt-texts)?

I am aware of a number of publishers using it.
But can you give me some context? What do you mean by "reconsider"?
Where do you want to use it? In a research project, in a commercial
system or in a publishing workflow?

Best,
Volker

TASK DETAIL
https://phabricator.wikimedia.org/T120938

EMAIL PREFERENCES
https://phabricator.wikimedia.org/settings/panel/emailpreferences/

To: Physikerwelt
Cc: Zorkow, jayvdb, Jdforrester-WMF, mobrovac, Dereckson, Matanya, JEumerus, Aklapper, StudiesWorld, Physikerwelt, Beta16, Mooeena, Akuckartz, CptViraj, Dibya, Viztor, 94rain, DannyS712, Tks4Fish, Maosef, Jayprakash12345, Orienteerix, Kizule, Flycatchr, EHCliffe, Zppix, DatGuy, Devwaker, Niklitov, Debenben, Urbanecm, Ananthsubray, Superzerocool, Volker_E, Tulsi_Bhagat, Izno, Wong128hk, Luke081515, SimmeD, Snowolf, fredw, Dcljr, Pkra, Dinoguy1000, Shizhao, scfc, Rxy, Jay8g, Krenair

Ping!
Was that a serious inquiry or a waste of time?

sorry for the late reply.

In T120938#6803897, @Zorkow wrote:

@Zorkow long time no see...

I wonder why that is...

are you still actively developig the speech rule engine?

Yes, people seem to use it and some fund me to continue developing it.
v3.2 should be out in beta soon.
We've got 5 European languages now and Hindi is in preparation.

Do you think it would be advisable to reconsider using its output in production (i.e, alt-texts)?

I am aware of a number of publishers using it.
But can you give me some context? What do you mean by "reconsider"?
Where do you want to use it? In a research project, in a commercial
system or in a publishing workflow?

I was somehow confused by this question. In MediaWiki -> Wikipedia starting with English. The timeline is however, more longtime. The question why I was asking ist that mathoid has currently pinned SRE was pinned to version 2.4 because updating causes failing tests. Now, I know that it is worthwhile to invest in updating the dependency.
Thank you.

Best,
Volker

I was somehow confused by this question. In MediaWiki -> Wikipedia starting with English.
The timeline is however, more longtime. The question why I was asking ist that mathoid has currently
pinned SRE was pinned to version 2.4 because updating causes failing tests. Now, I know that it is
worthwhile to invest in updating the dependency.

My apologies, but I was under the impression that you had sent me an
email inquiry on SRE. I was not aware that I was roped into an entire
thread I had to catch up on. In that case it is always helpful to give
people some context to get them up to speed.

I think it is great to see that Wikipedia starts to take some interest
in accessibility even if only in English.
Unfortunately, I am not familiar with mathoid, but I assume you are
talking about your system here: https://www.npmjs.com/package/mathoid
SRE >=3.0 does introduce a number of breaking changes that can cause
problems downstream.
Have you analysed what is causing your tests to fail?

Github offers a feature that allows you to track system releases
(https://github.com/zorkow/speech-rule-engine/releases)
There should be sufficient information in the release notes to let you
get to the root of your problem. But if you have more questions it is
best to follow up on our issue tracker.

Best,
Volker