# Start a new discussion

## Not signed in

Want to take part in these discussions? Sign in if you have an account, or apply for one below

## Site Tag Cloud

Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.

• CommentRowNumber1.
• CommentAuthorPaoloPerrone
• CommentTimeOct 22nd 2019

• CommentRowNumber2.
• CommentAuthorDmitri Pavlov
• CommentTimeMay 15th 2020

Added the general definition of a Radon measure on a Hausdorff topological space.

• CommentRowNumber3.
• CommentAuthorDmitri Pavlov
• CommentTimeMay 15th 2020

\hbox is not working. How am I supposed to insert plain text in formulas?!

• CommentRowNumber4.
• CommentAuthorUrs
• CommentTimeMay 15th 2020

With \text{...}

• CommentRowNumber5.
• CommentAuthorDmitri Pavlov
• CommentTimeMay 15th 2020
• (edited May 15th 2020)
Re #4: \text{...} does NOT work:

> Invalid LaTeX block: \mu(B)=\inf\{\mu(V)\mid \text{$V\supset B$ and $V$ is open}\}.
• CommentRowNumber6.
• CommentAuthorUrs
• CommentTimeMay 16th 2020

There is no maths allowed in \text-environments just plain text.

  $\inf\{\mu(V)\mid V\supset B \;\text{and}\; V \;\text{is open}\}$

• CommentRowNumber7.
• CommentAuthorDavidRoberts
• CommentTimeMay 16th 2020
• (edited May 16th 2020)

Why not something like $a \text{ and } b$?

$a \text{ and } b$

EDIT: ok, so spaces aren’t recognised inside \text, then, unlike LaTeX.

EDIT2: trying $a \text{\ and\ } b$$a \text{\ and\ } b$

• CommentRowNumber8.
• CommentAuthorUrs
• CommentTimeMay 16th 2020

I suppose if sombody knowledgeable in the relevant software would spare an hour, they could easily improve on this Instiki behaviour. Anyone interested in lending a hand should contact Richard Williamson for info on how to get started.

• CommentRowNumber9.
• CommentAuthorDmitri Pavlov
• CommentTimeMay 16th 2020
I can probably easily write the necessary patch if somebody can point me to the precise file and method
where this type of parsing happens.
• CommentRowNumber10.
• CommentAuthorDmitri Pavlov
• CommentTimeMay 16th 2020
Re #7: This type of bizarre pseudo-(La)TeX syntax is what drives me mad.

I seriously suggest we consider switching from “iTeX” to MathJax.

The main obstacle, as far as I understand, is iTeX's handling of multiletter identifiers like “colim”.
This can be easily handled programmatically.
• CommentRowNumber11.
• CommentAuthorUrs
• CommentTimeMay 16th 2020

if somebody can point me to the precise file and method

Richard Williamson can do that. Could you contact him by email?

• CommentRowNumber12.
• CommentAuthorDmitri Pavlov
• CommentTimeMay 16th 2020

Re #11: I wrote him an email and added you as a CC.

• CommentRowNumber13.
• CommentAuthorDmitri Pavlov
• CommentTimeMay 17th 2020

• CommentRowNumber14.
• CommentAuthorDavidRoberts
• CommentTimeMay 17th 2020

Added some spacing for the problematic definitions.

• CommentRowNumber15.
• CommentAuthorDmitri Pavlov
• CommentTimeMay 17th 2020

Honestly have no clue why \begin{definition} … \end{definition} does not work in the section on pushforwards of Radon measures.

• CommentRowNumber16.
• CommentAuthorDmitri Pavlov
• CommentTimeMay 17th 2020

Appears to be suddenly working now! Strange.

• CommentRowNumber17.
• CommentAuthorDmitri Pavlov
• CommentTimeMay 17th 2020

• CommentRowNumber18.
• CommentAuthorRichard Williamson
• CommentTimeMay 19th 2020
• (edited May 19th 2020)

Hi Dmitri,

Thanks for the email, and apologies for the slow reply. To fix this properly, you would need to edit itex2MML, it is nothing to do with Instiki directly. I don’t think we have ever tweaked itex2MML, so probably we are using the same source as here. If you make a patch and compile the binary for Linux, I am happy to push that binary to the nLab platform.

If you prefer, you could try to edit the nLab renderer instead. I have done that in a very limited way here in the tex_post_parser method, but it is a hack really. If you would like to get into editing itex2MML, I suggest that you fork it and add it to the nLab repository and also add support llbracket, esh, and so on in itex2MML rather than doing what I have in the renderer there.

I seriously suggest we consider switching from “iTeX” to MathJax.

We have certainly considered this in the past. The problem, as I have mentioned numerous times on the nForum, is that MathJax is fundamentally client side. It can be run server side, but one loses crucial client side information like the font the browser is using, the font size, etc. If we rely on client side rendering, there will be a performance hit, and also (even though I use a browser myself (qutebrowser) where MathJax is used) the user experience is not as good, with the shifts in the page rendering as MathJax kicks in; it is much better I think for us render to MathML server-side and allow the browser to render it when it is capable of it. If you know any alternative to itex2MML, or are willing to write your own, I am happy to consider it.

Or if you wish to experiment with using MathJax client side without using itex2MML, we can certainly try that on small pages if others agree. I don’t think we can do that on large pages, but, as I have again mentioned before, we could consider writing a bit of Javascript to split the page up in some way so that much of it loads invisibly, rather than using MathJax to load the entire page at once as currently. If you wish to try that, please go ahead: you can just take the HTML source for a very large page like this one and play about with adding some Javascript to it to achieve this,

1. An alternative is to take the approach used for rendering Tikz diagrams, and render all mathematics actually in LaTeX, convert the pdfs to svgs, and use those. This has the advantage of ensuring that LaTeX is the same as normal. But it has a lot of disadvantages as well.

• CommentRowNumber20.
• CommentAuthorDavidRoberts
• CommentTimeMay 19th 2020

Yes, let’s not use SVGs.

Chromium is slowly getting MathML support back, so that once that becomes live, Chrome and Edge (and others) will be doing proper MathML, rather than the MathJax replacement.