Apply MathCAT settings with libmathcat.SetPreference when changed by codeofdusk · Pull Request #19613 · nvaccess/nvda

codeofdusk · 2026-02-13T07:17:23Z

Link to issue number:

Closes #19597.

Summary of the issue:

After #19373, NVDA math preferences were not honoured by MathCAT.

Description of how this pull request fixes the issue:

Per #19373 (comment), use libmathcat.SetPreference to apply settings changes when they are saved to the NVDA configuration.

Testing strategy:

Confirmed that I could reproduce #19597 before this PR and not after. Verified that preferences persist across NVDA restarts.

Known issues with pull request:

None known

Code Review Checklist:

Documentation:
- Change log entry
- User Documentation
- Developer / Technical Documentation
- Context sensitive help for GUI changes
Testing:
- Unit tests
- System (end to end) tests
- Manual testing
UX of all users considered:
- Speech
- Braille
- Low Vision
- Different web browsers
- Localization in other languages / culture than English
API is compatible with existing add-ons.
Security precautions taken.

Copilot

Pull request overview

This PR fixes NVDA MathCAT settings not being honored at runtime by applying NVDA’s stored MathCAT preferences through libmathcat.SetPreference when MathCAT initializes and when preferences are saved.

Changes:

Added applyUserPreferences to push stored MathCAT preferences into the MathCAT runtime via libmathcat.SetPreference.
Updated MathCATUserPreferences.save() to apply runtime preferences before writing mathcat.yaml.
Updated MathCAT initialization to apply all user preferences (not just braille code) after setting the rules dir / TTS mode.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
`source/mathPres/MathCAT/preferences.py`	Introduces `applyUserPreferences` and uses it from `MathCATUserPreferences.save()` to update MathCAT runtime preferences.
`source/mathPres/MathCAT/MathCAT.py`	Applies stored MathCAT preferences during MathCAT startup via `applyUserPreferences()`.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

source/mathPres/MathCAT/preferences.py

seanbudd

Thanks @codeofdusk

source/mathPres/MathCAT/preferences.py

…nings

seanbudd · 2026-02-13T07:44:49Z

@GeorgeKerscher - can you confirm testing of this build when it completes? You should be able to find it under an "artifacts" menu under the checks tab for the pull request

GeorgeKerscher · 2026-02-13T17:33:34Z

I could not find a link to an alpha to test. Can you provide a direct link and I will test.

codeofdusk · 2026-02-13T19:19:24Z

@GeorgeKerscher honestly it's a little hard to find if you aren't familiar with the UI, but here's a test build. It's unsigned, so you won't be able to use it on secure screens and it'll show a warning when you run it. In the installer, just click "continue running" or create (and run) a portable copy.

GeorgeKerscher · 2026-02-13T19:52:34Z

I installed pr19613-4581,c4437fbd (2026.1.0.4581)

I installed as a portable copy. I found that there are still issues:

Beps are now being spoken, so that is fixed.
The pause factor is not being saved or applied. Even if I manually save settings, the next time I launch, the pause factor returns to 100.
The relative speech rate is being saved, but not applied. The speech rate does not change from the Math reading default.
The pause factor is also not being applied.

codeofdusk · 2026-02-13T20:25:44Z

@GeorgeKerscher turns out that I can also reproduce the pause factor bug under 890d2d6, and it's unrelated (the factor wasn't being scaled properly when saved to config).

As for relative speech rate, which synthesizer are you using? If you aren't using espeak-ng, can you test under that synth?

GeorgeKerscher · 2026-02-13T20:59:09Z

I was using Windows Core voices. When I switched to E-Speak, NVDA did change the voice speed. However, when I went outside the MathML, it did not return to the original speed setting; it stayed at the speech rate set in the MathCAT settings.

codeofdusk · 2026-02-13T21:00:47Z

However, when I went outside the MathML, it did not return to the original speed setting; it stayed at the speech rate set in the MathCAT settings.

Yes, that's been the case with all versions of MathCAT (including the add-on).

GeorgeKerscher · 2026-02-13T23:47:58Z

I think that is incorrect. Once outside the MathML expression, the speed returns to the setting in NVDA. I just tested with:
(NVDA) Version: 2025.3.2 (2025.3.2.53588)
and
MathCAT: speech and braille from MathML; Status: Enabled; Installed version: 0.7.5; Channel: Stable; Author: Neil Soiffer soiffer@alum.mit.edu; Publication date: 11/17/2025; Install date: 12/3/2025; Minimum NVDA version: 2025.1; Last tested NVDA version: 2025.3 10 of 18

I am using one core voices, David. Perhaps that is a behavior with E-Speak?

seanbudd

Approved pending testing confirmation with George

source/gui/settingsDialogs.py

GeorgeKerscher · 2026-02-16T15:52:53Z

Please provide a link to an alpha you want me to test. Probably after the merge.

seanbudd · 2026-02-17T00:57:58Z

@GeorgeKerscher - weren't you still having issues with speech speed with the current PR build?

codeofdusk · 2026-02-17T01:19:05Z

@seanbudd I can reproduce that issue as of #18323 itself. It's unrelated to my changes.

seanbudd · 2026-02-17T01:23:43Z

Thanks for confirming - @GeorgeKerscher can you open a separate issue for the speech problems?

GeorgeKerscher · 2026-02-17T02:21:58Z

I just installed NVDA 2026.1 Beta 3 and my preliminary testing resolves the speed, pause, and bep problems initially reported. Two thumbs up!

seanbudd · 2026-02-17T02:29:51Z

Hi @GeorgeKerscher - this PR is not in beta3, can you test alphas or beta4 and confirm?

GeorgeKerscher · 2026-02-17T21:07:05Z

A few things:

I am no using NVDA 2026.1 beta 3 as installed. I find four bugs in beta 3:
Zoomed out all the way is not spoken
MathML cannot be copied into Word.
Relative speech rate not honored
Pause factor is not honored or saved

If you give me a link to another alpha to test, I will be happy to test. Also, Should I create a new issue with these items on beta 3?

seanbudd · 2026-02-17T23:15:05Z

As mentioned in my previous comment, beta3 doesn't contain any of the fixes from this PR. Please test with alpha or beta4 and open a new issue with anything remaining

SaschaCowley · 2026-02-18T00:00:38Z

@GeorgeKerscher to clarify, beta4 will not be out until next week, but these changes are in NVDA alpha 54640,2f8b7d4b

GeorgeKerscher · 2026-02-18T02:13:26Z

I am using alpha-54640,2f8b7d4b (2026.2.0.54640)

I am finding:
MathML copied from a web page cannot be copied into Word.
Relative speech rate not honored
Pause factor is not honored

I will submit a new issue. It may be a day or so.

Follow-up to #18323, #19373 and #19613 ### Summary of the issue: Since the introduction of MathCAT into NVDA in #18323, NVDA has persisted MathCAT settings to a YAML file . This is at odds with current project configuration standards, which use either configobj ini files, or JSON. This also duplicates settings, as they're also saved to `nvda.ini`. Originally, these were stored in `%AppData%\MathCAT\prefs.yaml`, but as of #19373, they are saved to `mathcat.yaml` in NVDA's config directory. In neither case was `shouldWriteToDisk` checked before saving the preferences. Additionaly, [NV Access's Corporate & Government page](https://www.nvaccess.org/corporate-government/) states "To prevent NVDA users from modifying their configuration or add-ons directly, user write access to [`%APPDATA%\nvda`] must also be restricted". If implemented, this would have stopped users from changing their MathCAT settings in memory, as they had to be saved to YAML to be loaded by MathCAT. In #19613, the logic was changed to programmatically apply these settings with `libmathcat.SetPreference`, which should solve the problem of users without write access to their NVDA config directory being able to change their MathCAT preferences. However, the code that wrote the settings to the YAML file was not removed, nor did it contain any error handling. ### Description of user facing changes: None, but may avoid some errors in the logs. ### Description of developer facing changes: Pyyaml is no-longer a dependency. `mathPres.MathCAT.preferences.MathCATUserPreferences.save` has been renamed to `apply`, and no longer saves the settings to YAML. Settings persistance is handled by NVDA's configuration system. ### Description of development approach: Removed the code that wrote MathCAT preferences to YAML, and any code that only it depended on. Renamed `save` to `apply` using VS Code's refactoring tool. Searched the repo for `yaml` and after confirming that we don't rely on `pyyaml` anywhere else, ran `uv remove pyyaml`. ### Testing strategy: Tested by loading the [Wikipedia page on the quadratic formula](https://en.wikipedia.org/wiki/Quadratic_formula) in firefox. Located some math and read it. Changed settings one at a time and tested that they were applied correctly. Restarted NVDA and ensured that the new settings were still present. ### Known issues with pull request: None known

Actually apply MathCAT settings with libmathcat.setPreference

b272770

codeofdusk requested a review from a team as a code owner February 13, 2026 07:17

codeofdusk requested review from Copilot and seanbudd February 13, 2026 07:17

Copilot started reviewing on behalf of codeofdusk February 13, 2026 07:17 View session

codeofdusk mentioned this pull request Feb 13, 2026

MathCAT settings not being applied #19597

Closed

Copilot AI reviewed Feb 13, 2026

View reviewed changes

source/mathPres/MathCAT/preferences.py Show resolved Hide resolved

source/mathPres/MathCAT/preferences.py Show resolved Hide resolved

seanbudd approved these changes Feb 13, 2026

View reviewed changes

seanbudd added this to the 2026.1 milestone Feb 13, 2026

seanbudd reviewed Feb 13, 2026

View reviewed changes

source/mathPres/MathCAT/preferences.py Show resolved Hide resolved

codeofdusk and others added 2 commits February 12, 2026 23:39

On second thought, log parsing/SetPreference errors as errors not war…

c66f4f7

…nings

Pre-commit auto-fix

694df4c

codeofdusk and others added 2 commits February 13, 2026 12:21

Fix pause factor

0a4bb25

Pre-commit auto-fix

25a2416

seanbudd marked this pull request as draft February 16, 2026 03:11

seanbudd approved these changes Feb 16, 2026

View reviewed changes

seanbudd reviewed Feb 16, 2026

View reviewed changes

source/gui/settingsDialogs.py Outdated Show resolved Hide resolved

seanbudd added the conceptApproved Similar 'triaged' for issues, PR accepted in theory, implementation needs review. label Feb 16, 2026

Review action

f77703c

codeofdusk marked this pull request as ready for review February 16, 2026 05:23

seanbudd merged commit 81c7e50 into nvaccess:beta Feb 17, 2026
34 of 39 checks passed

SaschaCowley mentioned this pull request Feb 18, 2026

Don't save mathCAT settings to yaml #19637

Merged

5 tasks

RyanMcCleary mentioned this pull request Feb 21, 2026

MathCAT relative rate and pause not respected for OneCore #19644

Open

Uh oh!

Conversation

codeofdusk commented Feb 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Link to issue number:

Summary of the issue:

Description of how this pull request fixes the issue:

Testing strategy:

Known issues with pull request:

Code Review Checklist:

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

seanbudd left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

seanbudd commented Feb 13, 2026

Uh oh!

GeorgeKerscher commented Feb 13, 2026

Uh oh!

codeofdusk commented Feb 13, 2026

Uh oh!

GeorgeKerscher commented Feb 13, 2026

Uh oh!

codeofdusk commented Feb 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

GeorgeKerscher commented Feb 13, 2026

Uh oh!

codeofdusk commented Feb 13, 2026

Uh oh!

GeorgeKerscher commented Feb 13, 2026

Uh oh!

seanbudd left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

GeorgeKerscher commented Feb 16, 2026

Uh oh!

seanbudd commented Feb 17, 2026

Uh oh!

codeofdusk commented Feb 17, 2026

Uh oh!

seanbudd commented Feb 17, 2026

Uh oh!

Uh oh!

GeorgeKerscher commented Feb 17, 2026

Uh oh!

seanbudd commented Feb 17, 2026

Uh oh!

GeorgeKerscher commented Feb 17, 2026

Uh oh!

seanbudd commented Feb 17, 2026

Uh oh!

SaschaCowley commented Feb 18, 2026

Uh oh!

GeorgeKerscher commented Feb 18, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

codeofdusk commented Feb 13, 2026 •

edited

Loading

codeofdusk commented Feb 13, 2026 •

edited

Loading