# User Details

User Since
Oct 24 2014, 1:27 PM (278 w, 6 d)
Availability
Available
IRC Nick
physikerwelt
LDAP User
Physikerwelt
MediaWiki User

# Tue, Feb 25

Physikerwelt added a comment to T246011: Replace istanbul with nyc .

I hope if we switch to nyc we can get rid of extra code branches to improve code coverage test such as https://coveralls.io/builds/28940235/source?filename=lib/index.js#L44

Physikerwelt closed T245344: Reenable tests for github.com/wikimedia/texvcjs as Resolved.

@MarcoAurelio thank you. Works great.

# Fri, Feb 21

Physikerwelt placed T184879: Fix \ce grammar with macros above/below texts up for grabs.

# Wed, Feb 19

Physikerwelt edited projects for T159735: Chemistry formulae ignore significant spaces, opposite to LateX source, added: Texvcjs; removed Mathoid, Math.

It seems to be a pure texvcjs bug. If someone provides a test case, we can probably fix it in the grammar. What I understand is that the input

• \ce{2.5} should be sanitized to \ce{2.5} and
• \ce{2 . 5} should be sanitized to \ce{2 . 5}

The dollar sign was problematic. There is the begin{math} end{math} workaround.

Physikerwelt closed T188879: Remove texvc, a subtask of T137788: Add the middle command to math symbols, as Resolved.
Physikerwelt closed T188879: Remove texvc, a subtask of T184880: texvcjs - Find solution for other commands within \ce, as Resolved.
Physikerwelt closed T188879: Remove texvc, a subtask of T185027: Bad spacing around \xrightarrow, as Resolved.
Physikerwelt closed T188879: Remove texvc, a subtask of T185552: Math rendering behavior change for \limits, as Resolved.
Physikerwelt closed T188879: Remove texvc as Resolved.

Its javascript successor texvcjs has now replaced texvc.
There is a debate to also remove texvcjs or to convert it to a non-modifying LaTeX sanity checker T245631

I think it would be good to focus on:

@DavidEppstein to that one must show that sane texvcjsoutput implies sane input. In general this does not seem obvious to me.

OK. I think we understand the LaTeX commands allowed in MediaWiki quite well. And to rewrite another sanity checker that does not modify the output in any way would require quite a bit of time, probably cause new problems and would need to get a quite intensive security review, which might also need additional time. Thus I am proposing to fix what we have. In 2015 at NIST we got rid of the additional curlies

# Tue, Feb 18

The list of functions where the spacing gets wrong with additional braces is everything that is defined like a unary or binary operator because those get a spacing of \medmuskip. If you place them in braces they become a subformula with spacing rules of an ordinary math expression. By default you have block layout in LaTeX where the size of all spaces between letters and words depends on how much needs to fit into a particular line.
If you search for \stackrel in enwiki https://en.wikipedia.org/w/index.php?sort=relevance&search=insource%3A%2F%5C%5Cstackrel%2F&title=Special:Search&profile=advanced&fulltext=1&advancedSearch-current=%7B%7D&ns0=1 you will see that people just insert spaces manually. Workarounds like inserting the spaces manually take the average editor half a second. If they see that other editors also insert those spaces they just copy what the others do and think that the behavior is intentional.

wrong spacing is hard to fix,

There is a very easy fix: Not modify the source code at all. With this you can be sure that you do not create errors.
If you want to modify the source code the only other genuine fix is parsing everything like in LaTeX or Mathjax in order to see what characters and macros insert what spacing at the position where they are placed.

While reading this, I was thinking about a private constructor. Not sure if that makes sense, but it would be an alternative to a code style rule.

There is no \newcommand in MediaWiki and wrong spacing is hard to fix, if the correct spacing is unknown. A sperate issue that lists the commands would be helpful. \sqrt is from the class fun_ar1 https://github.com/wikimedia/texvcjs/blob/master/lib/texutil.js#L647-L697 and stackrel from fun_ar2 \xrightarrow from fun_ar1opt this is quite a large list of functions. I wonder that it has never been discovered that the spacing is wrong due to additional brackets. Please note that the spacing is marginally different in LaTeX, MathJax, and MathML.

Didn’t get you. What do you mean by shorter/longer form?

# Mon, Feb 17

Is there a follow up change to the coding convention check. Otherwise people like me might reintroduce the shorter alternative in the future without being aware of the longer form.

Physikerwelt changed the status of T209563: Default wgMathMathMLUrl points to non-working mathoid.testme.wmflabs.org from Open to Stalled. Can someone^TM from the core platform team look at the op PR and decide how to proceed with them. The underlying issue was resolved I think. # Sun, Feb 16 I lost you. Are you saying it should fail or it should not? yes, it should fail and if it renders it should render different. The example was meant to illustrate that there is not one small problem but several small problems which can potentiate, most notably: • The align and aligned environment are different. One difference is that aligned can take optional arguments like t and b for top and bottom alignment in square brackets and align not, therefore align should not be replaced with aligned. We can change that. This seems straight forward. • A minimal example like "a \sqrt[b] c" should stay like it is and not become "{\sqrt[{b}]{c}}" because it makes it hard to read. I think this is very annoying. I tried to change it long ago, but I was not successful in getting this plain tex inspired extra curlies removed. Maybe I will try again to get this change merged. On the other hand, it is just hard to read, not a real error. Or? • If you combine it with the additional problem of detecting where the optional argument begins and where it ends you somehow end up with "\sqrt {[}..." in the more complicated example. Not exactly. This should be possible with pegjs rules. I have not investigated why in this particular case the optional argument is destroyed. If you really want to make sure that you can detect things like those optional arguments right, you have replicate the macro expansion like LaTeX or MathJax does it when it parses the expression (e.g. hardcode into texvc what kind of expressions and environments can take what kind of arguments) and I do not see the point of this since texvcjs does not render it. It also does not provide helpful error messages to the user therefore it should not reject the input. MathJax already has the ability to reject input and provide helpful error messages, I do not see the point of replicating that either. To operate independently of MathJaxs (or whatever future rendering engine) problems and possible security exploits texvc was introduced in 2003. This security feature can not be dropped without broader discussion beyond the Math functionality horizon. In conclusion, texvcjs might be replaced by another security and whitelisting guard, but it won't be dropped :-( # Sat, Feb 15 Physikerwelt added a comment to T245343: texvcjs should not replace\omicron$with$\mathrm{o}$. I realized that \Omicron (the uppercase variant) does not render on https://www.tuhh.de/MathJax/test/sample-dynamic.html Physikerwelt updated subscribers of T245343: texvcjs should not replace$\omicron$with$\mathrm{o}$. @Debenben Can you have a look at the replacement list https://github.com/ag-gipp/texvc-latex/blob/master/texvc.pdf and check if there are further commands, where the replacements generate false results. Just by looking at the PDF, one can, for instance, see that also \Omicron is not italic-ish. I did never realize that because I was not aware of how \omicron is supposed to look. I do relay on your input here;-) @Physikerwelt Thank you very much for creating the pull request, it is definitely a very good fist step in the right direction. I guess nobody really cares about the name of the category, we could simply use "Pages that use a deprecated format of the math tags" in analogy to the chem tracking category. # Wed, Feb 12 Physikerwelt added a comment to T208408: Check whether math project requires NFS or not. We use project and dumps intensively. However, scratch we use for temporary files only. # Sun, Feb 9 @Debenben the mathoid side looks good for now. For the math extension change, I would be happy to receive a suggestion for a tracking category name. I hope I get the mathoid change merged during the upcoming week so I can work on the math extension change next weekend. I implemented the change in texvc https://github.com/wikimedia/texvcjs/pull/34 So first I documented the deprecations in the texvc package documentation. https://github.com/ag-gipp/texvc-latex/commit/72835b8e4636ab0ad821edaf2b72dda627d1e5d0 Oh no. I just realized that I did only implement the tracking categories for the mhchem-deprecation pages. But not the deprecated texvc syntax. I will start this today and see how far I can get. I think it boils down the question if we want to extend the set of whitelisted macros and allow \emph as part of the Wikimedia LaTeX syntax. I personally think we don't really need it, but we can also make if poll on this. Physikerwelt changed the status of T182041: Display math generates div inside of paragraph (HTML5 violation), a subtask of T21719: HTML5 features (tracking), from Open to Stalled. Physikerwelt changed the status of T182041: Display math generates div inside of paragraph (HTML5 violation) from Open to Stalled. I am not sure what to do here. But I think it is not "a good first bug" until there are more specific instructions on how to transition from the current to the desired behavior. Physikerwelt added a comment to T239359: Organize the open tasks for Project Math. @Aklapper sorry for the long silence! @Physikerwelt: Ah, thanks! I have some thoughts but before sharing them, please help me first to understand a bit better: Could you explain why not all tasks in the "Mathoid" column on the current Math workboard also have the Mathoid project tag on them? Not really. Note that the tasks are not made or curated by a group of professionals that agreed on standards. There is no systematic assignment of tags or columns. So these are two technically different ways to transport the same message. This task is related to mathoid;-) Same question for the texvc column vs the Texvcjs tag (though those might be different things?), and the VisualEditor column vs the VisualEditor tag. I have the same unsatisfying answer here. Also, I'm wondering if that Design column is really needed: You could always filter for design tasks, no matter which column they are in, by clicking "Open Tasks" in the upper right corner, choose "Advanced Filter...", and query for Design tasks (or even add a shortcut link to the side bar which does that, for convenience). Maybe that would be good enough. My problem is just that I have difficulties with the columns n>4 (which someone else created). While I see that they serve a purpose they are a bit problematic for the other columns. Physikerwelt added a comment to T229939: Use wikidata entity selector in OOUI input field. I am not sure how to move forward here. @Esanders, do you have a hypothesis? What might be the problem here? I think the approach to use$output->addModules and specify the dependencies there is not fundamentally incorrect as outlined here
https://phabricator.wikimedia.org/T229939#5699422 but maybe the loader does not load all dependencies recursively? Or are other sources for the problem more likely?

Physikerwelt updated the task description for T239356: Enable link feature for all wikis.
Physikerwelt added a comment to T208758: Display popups for annotated [itex] tags.

@alexhollender the two issues on the link are independent of the goodwill of the page preview team. We can just fix these drawbacks of the current implementation. For the popup itself, we need a commitment of the page preview team to that feature. Therefore I suggest we fix the hard things first and decide on an implementation plan for the popup feature. Or do you have other thoughts? I think it is important to join forces and move together in the same direction.

# Fri, Feb 7

Yes. Almost certainly. Especially in the beginning of the weekend it is sometimes frustrating if you start to realize that wikibase became incompatible to the math extension 🤔

Physikerwelt added a comment to T208758: Display popups for annotated [itex] tags.

@Physikerwelt following up from our email, here is some design feedback:
Imagine you are reading this section and you would like to see definitions for the equations. Some equations you can hover and the preview will appear, others will not show a preview because they have not been linked. The math equations that are linked should have some kind of visual treatment that indicates to people that you can hover over it and a preview will appear. I understand that you do not want to turn all math equations blue. Is there some other treatment you could explore here?
Interactivity
This goes along with the point above — when someone is hovering over a linked math equation there should be some visual indication (aside from the cursor changing) that the link is being hovered. As you can see with normal blue links when they are hovered an underline appears. This gives valuable feedback to the person using the website about what is happening, or what will happen if they click.

# Jan 18 2020

I think this is not ubn. Why not disable the test?

Physikerwelt lowered the priority of T243122: Math master is currently broken due to a Wikibase change from Unbreak Now! to Medium.

# Jan 16 2020

wait... I can just write a travis job and then I will be notified if something broke post merge. theoretically not optimal but in practice almost as good for this little extension.

Physikerwelt lowered the priority of T117659: mwext-testextension-zend should load extension mathsearch after math from High to Medium.
Physikerwelt updated subscribers of T117659: mwext-testextension-zend should load extension mathsearch after math.

@hashar @Jdforrester-WMF is there anything I can contribute to get the php unit tests running for new patches uploaded to gerrit? Was there any progress to allow self-serviced test setups a la travis?

Physikerwelt committed rEMASe153793cd5f6: Fix typos in extension.json (authored by Physikerwelt).
Fix typos in extension.json

Seem to be a typo in Ib6c3044603ead1a58b63a4468c525820331ca565

The fact that MathSearch is currently disabled in Jenkins is not optimal. See ,e.g., T242957

Physikerwelt raised the priority of T117659: mwext-testextension-zend should load extension mathsearch after math from Medium to High.

# Jan 14 2020

The changes could be merged by basically all WMDE and WMF employees. I have not looked into the special requirements to backport changes and there might be good reasons not to merge certain changes.

I have no permissions to do that.

@Theklan for me it works when I am logged in as well. Can you try to identify why this problem occurs? For example change your settings disable gadgets, try another wiki or another account and try to understand when this problem occurs and when it does not. Thank you.

# Jan 13 2020

For me it worked also when I was logged in. Maybe this can be determined in a separate ticket.

works for me in a private firefox window (not logged in). maybe it is a caching issue?

It is available in all languages. Only the guide https://en.wikipedia.org/wiki/Help:Displaying_a_formula#Semantics_and_Links might not yet be translated.

# Jan 12 2020

Physikerwelt committed rEMAS5ecf210ae79e: Use API for translation if available (authored by Physikerwelt).
Use API for translation if available

# Jan 1 2020

Physikerwelt committed rEMASc0d65873df76: Tolerate errors in MathID generation (authored by Physikerwelt).
Tolerate errors in MathID generation

# Dec 29 2019

Physikerwelt committed rEMASc5e78bc976a1: Write cache for math rendering on Wikibase sites (authored by Physikerwelt).
Write cache for math rendering on Wikibase sites

@StefanEis the failure in "Comparing the generated SVG with the reference failed." is not severe. I also happens on Wikipedia https://en.wikipedia.org/wiki/Special:MathStatus

Physikerwelt changed the subtype of T241531: Fatal exception of type "Wikimedia\Rdbms\DBQueryError" Exception caught inside exception handler with Extension:Math from "Task" to "Bug Report".

In it's default configuration Extension:Math does not use a database. There might be an issue in the interaction between Math and VisualEditor or Math and Wikibase. Can you provide more information on the error. Otherwise it will be hard to help here.

Physikerwelt committed rEMAS4b46339d184f: Make max length of Contentmath configurable (authored by Physikerwelt).
Make max length of Contentmath configurable

# Dec 28 2019

Physikerwelt updated the task description for T229939: Use wikidata entity selector in OOUI input field.
Physikerwelt added a comment to T229939: Use wikidata entity selector in OOUI input field.

I am seeing similar problems on test.wikidata.org

Physikerwelt closed T104548: Set up a stable math demo instance as Resolved.

I checked this today. And it, unfortunately, does not work as expected.


curl -X POST "https://en.wikipedia.org/api/rest_v1/media/math/check/tex" -H  "accept: application/json" -H  "Content-Type: multipart/form-data" -F "q=%"

does not show any warnings. I need to look at this again:-(

Physikerwelt moved T239356: Enable link feature for all wikis from Incoming to Next-up on the Math board.
Physikerwelt updated the task description for T239356: Enable link feature for all wikis.

# Dec 26 2019

Physikerwelt updated the task description for T241438: Investigate if objectcache tries to write to database during installation.
Physikerwelt updated the task description for T241438: Investigate if objectcache tries to write to database during installation.
Physikerwelt renamed T241438: Investigate if objectcache tries to write to database during installation from Investigate if Object cache tries to write to database during installation to Investigate if objectcache tries to write to database during installation.
Physikerwelt renamed T241438: Investigate if objectcache tries to write to database during installation from Investigate requirements to set up mediawiki for languages that need extra packages to Investigate if Object cache tries to write to database during installation.
Physikerwelt updated the task description for T241438: Investigate if objectcache tries to write to database during installation.
Physikerwelt updated the task description for T241438: Investigate if objectcache tries to write to database during installation.

All these languages seem to be main (dash free) variants of languages with variants.

	public static \$languagesWithVariants = [
'en',
'crh',
'gan',
'iu',
'kk',
'ku',
'shi',
'sr',
'tg',
'uz',
'zh',
];`

# Dec 25 2019

Physikerwelt updated the task description for T241438: Investigate if objectcache tries to write to database during installation.