# User Details

User Since
Oct 24 2014, 1:27 PM (207 w, 3 d)
Availability
Available
IRC Nick
physikerwelt
LDAP User
Physikerwelt
MediaWiki User

# Sep 6 2018

Physikerwelt lowered the priority of T202266: MathHooks: table creation yields warnings on quibble from High to Normal.

# Sep 5 2018

Physikerwelt updated the task description for T123459: Lower log level to debug for syntax errors in math tags.
Physikerwelt added a comment to T123459: Lower log level to debug for syntax errors in math tags.

@Krinkle thank you for this report. However, I can't do anything in the foreseeable future, since the math CI tests are broken T202223.

Physikerwelt renamed T123459: Lower log level to debug for syntax errors in math tags from Lower log level to debug for syntax erros in math tags to Lower log level to debug for syntax errors in math tags.
Physikerwelt renamed T123459: Lower log level to debug for syntax errors in math tags from [Regression] Math: "Rendering failed. Printing error message." to Lower log level to debug for syntax erros in math tags.

# Sep 4 2018

It would be great if someone could merge https://gerrit.wikimedia.org/r/#/c/mediawiki/extensions/Math/+/455482/ so that the failing tests do not block other developments. Thank you in advance.

# Sep 3 2018

Adding fallback configuration does not solve this problem. We need to find a way to fix / workaround this issue asap.

# Aug 29 2018

Physikerwelt added a comment to T202266: MathHooks: table creation yields warnings on quibble.

@Pablo-WMDE no it's not. I'm a bit frustrated that I blocks me from doing anything. I don't understand how a change in Wikibase can affect the Math extension configuration. I suspected that the problem is the use of globals, but it did not help as well. Is there a way one can debug or reproduce the new unit testing environment?

# Aug 27 2018

I am not sure what to do. Creating all tables is certainly a workaround. However, that would be different from the production setup and create tables that might never be used. I suspected that it could be related to the use of the globals to access the config. However, that did not help either.

# Aug 20 2018

Physikerwelt changed the status of T192634: Prepare removal of LaTeXML rendering mode from Open to Stalled.
Physikerwelt changed the status of T192634: Prepare removal of LaTeXML rendering mode, a subtask of T121100: Move research and experimental code from Math to MathSearch extension, from Open to Stalled.
Physikerwelt added a comment to T192634: Prepare removal of LaTeXML rendering mode.

I unassigned @Andreg-p since there is no progress :-(

Physikerwelt placed T192634: Prepare removal of LaTeXML rendering mode up for grabs.

# Aug 19 2018

@SalixAlba sure. However, according to my experience it's easier to get code review if the changes are not too big. So I will do that in a follow up if this one works as expected.

Just a quick update from the math extension development. I added a tracking category for pages that use the deprecated mhchem syntax. It is waiting for code review at https://gerrit.wikimedia.org/r/#/c/mediawiki/extensions/Math/+/442124/ and currently the testing framework is broken T202223 .

# Aug 8 2018

I am not sure if that's the way to go. Neither the Math extension depends on Wikibase nor the other way round. For the CI math depends on wikibase. This is good for WMF applications. However, for private wikis that use Math but not wikibase this is not optimal since unintended dependencies on from math to wikibase would not be discovered. The same holds for the VisualEditor.

@WMDE-leszek thank you. Yes. My local wikibase copy was not up to date when I created the change (cf. Patch set 1).

# Aug 7 2018

@WMDE-leszek thanks for fixing. However, I wonder how this could be merged without failing tests. Right after the change was merged, all subsequent unit tests failed. Why did I not see the error message when I tested the change on Jul 24th?

is that the same issue as T200493?

# Jul 24 2018

if the expression is not valid according to the grammar rules in texvcjs, it issues a warning. this might not resolve all problems but it will make it easier to understand the current rendering...

In https://gerrit.wikimedia.org/r/#/c/mediawiki/extensions/Math/+/442124/ I added the functionality to add a new tracking category for pages that use <ce/> tags with the old syntax.

# Jul 23 2018

@zeljkofilipin is the status table up to date? For instance, for Wikidata I see

06:37:01 /usr/bin/find: cannot delete '/workspace/src/tests/selenium/log': Permission denied

in https://integration.wikimedia.org/ci/job/Wikibase-npm-browser-run-selenium-daily-node-6-docker/lastBuild/console which I think is the last build.

# Jul 2 2018

@Jdlrobson I am currently traveling until Wednesday.... From my understanding the problem is not wikidata but wikibase. Thus the styles need to be added somewhere using a wikibase hook.

# Jun 25 2018

@Jdlrobson yes. The (only) other way I am aware of is using properties of type mathematical formula from wikidata.

Physikerwelt renamed T147319: No MathML rendering in Firefox - SVG or PNG is used from No MathML rendering - SVG or PNG is used to No MathML rendering in Firefox - SVG or PNG is used.
Physikerwelt added a comment to T147319: No MathML rendering in Firefox - SVG or PNG is used.

Looking at these bugzilla tickets gives the impression that not very much has happened recently. I have no idea what would happen if we changed the CSS to display MathML for all Firefox users. People who install the plugin to show MathML are probably not a representative sample. Thus, I have the feeling that it would be good to have someone from Mozilla involved if did this. Otherwise, we might be ending up in a situation, with bug reports regarding the Firefox MathML rendering that nobody will be able to handle.

@Jdlrobson I agree that global should be avoided. Any other solution is highly appreciated, I am just unaware of one. However, whenever math is displayed the associated CSS and js classes should be loaded. I think a good solution is that the math extension follows the standard procedure for all tag extensions that require custom CSS and js for their content. Is there a good example of such an extension?

@Pkra @fredw I agree that this should be fixed within mathjax, so I am closing this ticket here.

Physikerwelt closed T73787: Test MathML mode on mobile devices as Resolved.

I could not reproduce theses problems with the most up to date Firefox.

Physikerwelt closed T73787: Test MathML mode on mobile devices, a subtask of T67341: Release Math 2.0, as Resolved.

Since I don't see progress here and do not have any contacts to Mozilla to talk about this goal I'm closing this ticket. In case anyone from Mozilla finds this ticket via a web search engine here a short summary of our request:

# Jun 24 2018

@Jdlrobson The patch you are proposing would prevent math from being properly displayed if its used in a lua module. See https://en.wikipedia.org/wiki/Module:ShowMath for an example. I would not consider this as a bug. I currently don't understand why it's only sometimes (for instance on mobile) loaded and sometimes not (for instance on desktop).

# Jun 21 2018

Thank you @mobrovac I think the next step is to add a warning to texvc if any of the commands above are used in the same way as we did it for invalid mhchem2 markup

Physikerwelt added a comment to T192634: Prepare removal of LaTeXML rendering mode.

@Andreg-p can you come up with a time plan for this change?

# Jun 20 2018

@Debenben I had a quick look at the bot code. According to my experience using regexp to replace macros in LaTeX leads to quite complex expressions soon and requires special treatment of many cases even in relatively small data-sets. I think a simpler approach is to customize texvcjs (or another latex paerser) and do the replacements in the parsed expression. However, if you only want to run the bot once and take care of the problems manually this approach might be good enough for the task at hand. That being said, I am not feeling comfortable to become a maintainer of this regexp based bot.

# Jun 17 2018

...
Math apparently now defaults to using WMF's public rest service for math rendering, which might not be the best default (?)
..

@brion do you have a suggestion, where and how to document how this decision was made?

@Debenben can we put the source code of the bot to a public repository. Otherwise, it might hard to decide if the bot flag should be granted. Or did I just miss the link to the bot code?

# Jun 15 2018

@Jdlrobson from my understanding what happens during hook registration code is that this code is only called if the parser sees a math,ce, or chem tag in the template expanded wikitext. I could imagine that during template expansion math tags might appear under certain circumstances. But this is really a wild guess.

Another potential source of caching problems is the setting dependent cache key https://github.com/wikimedia/mediawiki-extensions-Math/blob/df1689c7177d5b7ef946ae2ff525594bd377da3f/Math.hooks.php#L133-L136 depending on the user setting for math rendering different versions of the page are served. I am happy about any suggestions, how the code could be improved.

@Jdlrobson I am maintaining the code in the Math extension. It there anything with the code within the math extension which is unclear?

# Jun 14 2018

Cool. As long as the not does not perform braking changes such as macro expansion for \$ and % and \or I don't see any problems. However, as soon as we do breaking changes we also have to change the tag name or add an attribute. Otherwise old revisions would not render correctly. Once we agreed to a new syntax we could maybe migrate to a new tag name such as formula.

# Jun 11 2018

Gerrit Code Review <gerrit@wikimedia.org> committed rWDQGcd8c97075860: Create change (authored by Physikerwelt).
Create change

# Jun 10 2018

@Debenben thank you. I don't really understand the purpose of your TeX Header. How is a different from the ctan texvc latex packge http://ctan.math.utah.edu/ctan/tex-archive/macros/latex/contrib/texvc/texvc.pdf ?

# Jun 7 2018

Physikerwelt updated subscribers of T195861: Create a committee to improve the math support in Wikimedia projects.

@Pkra I started with the same impression. But over the last years, the processes inside WMF are much clearer to me. While one might at first glance get the (wrong) impressions that "they" rejected things because "the staff" has more rights, time and power I can now understand the rationality behind these decisions. If there is something we cannot implement as volunteers, we need to request for resources (about one year in advance). To send a request that is likely to be approved we need a proposal as explained in the description of the ticket.

One thing that needs to change is the "sanitization" (probably should be called pollution or infection). For example...

@Pkra I am quite certain that there is nobody at WMF that has a strategy for the future of math. Thus we need to take care of it by ourselves in the wiki way. Now I am making a broad claim: Neither WMF nor WMDE will stop us from improving the math support if it is reasonable.

# Jun 6 2018

For the regression test, we could reuse mathpipe (description). However, it would be good to update to the latest selenium version to get a realistic (how it looks in a browser) rendering.

# Jun 5 2018

See https://stackoverflow.com/questions/48238751/search-wikipedia-for-articles-containing-a-special-tag we have the list of all WMF formulae but we don't know their origin. More later.

# May 31 2018

Physikerwelt updated the task description for T195847: Clean up artifacts from LaTeX based math rendering.
Physikerwelt triaged T196055: Remove table math from the database as Lowest priority.

# May 30 2018

Physikerwelt updated subscribers of T195861: Create a committee to improve the math support in Wikimedia projects.

@Debenben thank you for adding more points. Please also feel free to invite others.

# May 29 2018

Physikerwelt updated the task description for T195847: Clean up artifacts from LaTeX based math rendering.
Physikerwelt triaged T195871: Remove dead texvc code from the math extension as Normal priority.
Physikerwelt triaged T195869: Make PHP code of the Math extension follow the SOLID Principles as Low priority.
Physikerwelt added a comment to T195847: Clean up artifacts from LaTeX based math rendering.

@MoritzMuehlenhoff yes. I just started with this list as overview while brainstorming.

Physikerwelt triaged T120440: at sign creates parsing error in math code where it doesn't in TeX as Normal priority.
Physikerwelt closed T63090: Math related Debian packages needed to test coverage of extension math, a subtask of T63051: MathJax equations are centered, as Resolved.
Physikerwelt closed T74240: Remove PNG only rendering mode as Resolved.

This is done. Let's keep track of the cleanup process in T195847.

Physikerwelt closed T74240: Remove PNG only rendering mode, a subtask of T30258: texvc should support commutative diagrams via XY-packet, as Resolved.
Physikerwelt closed T74240: Remove PNG only rendering mode, a subtask of T22902: \xrightleftharpoons macro for [itex]-environment, as Resolved.
Physikerwelt closed T74240: Remove PNG only rendering mode, a subtask of T63090: Math related Debian packages needed to test coverage of extension math, as Resolved.
Physikerwelt closed T74240: Remove PNG only rendering mode, a subtask of T120440: at sign creates parsing error in math code where it doesn't in TeX, as Resolved.
Physikerwelt closed T74240: Remove PNG only rendering mode, a subtask of T165751: Feature: Let Math accept certain Unicode input (eg. ∫ \frac{∂f}{∂τ}dτ ), as Resolved.
Physikerwelt closed T74240: Remove PNG only rendering mode, a subtask of T166367: Consider removing the options from the Math extension, as Resolved.
Physikerwelt added a comment to T195765: Export original TeX string of math values in RDF.

;-) I think everything falls into this category. See Table 4 for a list of examples that are being considered as examplary information needs of users.

If you look at the wikitext, you see that this was not written as a math tag, but using a template {{Intorient|symbol=varoint}} , which links to an SVG file on commons. Thus the addon of @fredw is not aware that it's a math symbol.

Physikerwelt added a comment to T188879: Remove texvc.

@Debenben Now, since we have stopped using texvc in production, we are technically in a position that we can change anything we can image. However, the things you mention require a lot of work thus we should make sure, to plan wisely before we start with the implementation and to make sure the implementation suites the community needs. I would, therefore, suggest that we form a committee that discusses the changes we want to make to the <math/> tag before we start with the implementation.

Physikerwelt added a comment to T195765: Export original TeX string of math values in RDF.

@Lucas_Werkmeister_WMDE with original TeX string you mean the userInputTex. It has been shown, that this representation is not suitable for Math Information Retrieval Tasks. Also, note that you need to install texvc in order to render the userInputTex. Reading the request a query page, I wonder what the underlying information need of the user was. Does it fall into the "what is the name of the formula ..." category?

Physikerwelt added a comment to T192634: Prepare removal of LaTeXML rendering mode.

Short update: HTTP has been implemented and works on drmf-beta.wmflabs.org

# May 20 2018

I did discuss with @zeljkofilipin to disable quit selenium tests for now.
However, the tests seem to be fixed. Now I'm unsure how to proceed.

# May 19 2018

Physikerwelt triaged T195038: Handle the storing of large transcluded blobs as Low priority.

# May 18 2018

Physikerwelt added a comment to T74240: Remove PNG only rendering mode.

@mobrovac simple but not trivial. We could introduce another class (e.g., MathPNG) based on the MathML class that overwrites the output method only. This class would always deliver png rather than mathml + svg fallback.

Physikerwelt added a comment to T74240: Remove PNG only rendering mode.

I have discussed this issue again with @mobrovac today. We will keep the PNG mode for now, but serve the PNG images from mathoid.

# May 15 2018

@Debenben Please refer to the respective tickets that explain why WMF decided to use texvc to restrict the input grammar.
I am discussing with @Pkra how to improve the accessibility and visual appearance offline. I would like to focus on the objectives and measurable outputs rather than a technology discussion on a marketing level.