Guess how's back: user242026

@barbarabeeton There are many thought provoking points in this article but I find it fraught with the "Evil English" mindset that seems to dominate so much. Especially in linguistics people talk about "decolonising research" and whatnot, as if white language researchers (and of course this only applies to white language researchers) are uncapable of stepping out of their cultural realm into others just because they speak English.

@HenriMenke back again. Shady's back, tell a friend!

@barbarabeeton I repeatedly get from my son while speaking English "well, that's what I call it" and I politely remind him that's not how language (any language) works. You can't just call it whatever, nobody will understand you. And you might get away with it in Iceland, with your friends who speak your "dialect", but it won't work among native speakers.

And this "On the contrary, communication ends because [the foreign researchers] cannot explain to the American, in simple language, the advanced topics they were discussing. Yet, the American takes over the conversation.", might be problematic concerning the American demanding to be in charge in spite of not understanding the work but if you can't explain your research in simple language you've failed. No matter what your native language is.

templates .....

\usepackage{color}
\@ifpackageloaded{color}{
    \PassOptionsToPackage{usenames,dvipsnames}{color}
}{%
    \usepackage[usenames,dvipsnames]{color}
}

Why test if a package is loaded, if the template explicitly loads it earlier?

Or does this already anticipate random changes made by user?

@samcarter_looks_forward_TUG'21 that seems a) a thinko or b) some byproduct of an incorrect merge (could happen...)

@Rmano the template is doing something similar for hyperref, so it looks deliberate

@samcarter_looks_forward_TUG'21 who knows... ;-)

siunitx v3.0.0 on its way to CTAN

Now I wait for the fun complaints

@JosephWright Expect lots of them by the end of summer as that's when we'll most likely deploy it :-)

@yo' You'll like the performance improvements

@JosephWright are they significant?

btw, it's probably not the most popular package out there. We mostly struggle with mhchem in large documents I think

@yo' A factor of two, on average

@yo' :)

@yo' I've tended to go with chemformula in recent years

@JosephWright next Joseph's project: sichemx

@yo' texdev.net/2021/05/17/siunitx-v3

@yo' Clemens has done a pretty good job in chemmacros to be honest

@yo' That's made use a of a lot of expl3 regex stuff in recent releases: I'm not surprised it might be tricky (needs an up-to-date l3kernel)

@JosephWright the problem is really the compilation time. Have there been any recent improvements? We're at LaTeX2e <2020-02-02> patch level 5, L3 programming layer <2020-07-17>

@yo' Biggest things I know about are where I've removed stuff: s column, requiring complex numbers/products to use dedicated interfaces, some of the font changes

@JosephWright “also re-us bits”

@yo' In mhchem? Not that I know of, although Bruno has been pretty active over the last month or so, including adding new regex features and thinking hard about performance (along with @Skillmon)

@PhelypeOleinik Fixed

@PhelypeOleinik Thanks

@JosephWright I haven't seen people use s much tbh

@yo' Ah, compared to that, some, yes, but not enough to make an impact on mhchem

@JosephWright I lost track of all the chem packages

@JosephWright ok. That means we can expect things to get worse over time :-)

@yo' Quite: it was a 'cute' idea but honestly it's not a great idea - better to use collcell and pass to \unit (\si) explicitly

@yo' One of the reasons for the v3 re-write of siunitx is to act as a demo of how to get good performance from expl3 code - I've used the data structures where they make sense, but used more 'classical' approaches when that works better

@yo' I've also documented all of that in the code part - for units, I use a prop-based method, but for numbers I've gone for a carefully thought-out internal structure of my own

@yo' I guess I know a lot about that :)

@JosephWright sounds too complicated for my L3 skills :D

Next, I have to write a chapter about siunitx for FMi ....

@yo' Certainly it's at the 'advanced' end

@yo' Part of my plan is that siunitx is supposed to be the 'model' expl3-as-a-user package: it's big, it uses most of expl3, it's written by a core team member

@yo' And now it's got the tests, API, modularity, etc. that is part of how expl3 code is 'supposed' to be written

@JosephWright that's not unexpected from one of the maintainers of the system, right? :)

@JosephWright cool

@yo' :P)

@JosephWright I haven't yet dared to touch performance of l3regex... :)

@Skillmon :)

@JosephWright yeah! I use it quite a lot, so let's see... (never used s though ;-))

@Plergux -- I've had some direct experience with this "explaining clearly", taking part in an ISO working group. ISO standards (almost?) always have an English version, and the text must be crafted carefully. While use of copious synonyms is recommended for "stylish" English, repetition of a single well-understood term is preferred for the text of a standard. Recently I've received some phone calls with the caller reading a script. Not successful.

@barbarabeeton :(

@JosephWright the problem in general is not the usage of the expl3 data structures. While some tend to be on the slower end, they are all reasonably fast and what's more important: Stable. It is a different issue to use the slower functions for everything (e.g., using \regex_replace_all:nnN all over the code cough, cough)

@barbarabeeton The phone cuts off a lot of the signal and we have to fill in the gaps, so if the phonology isn't what we're used to it's going to be harder. Also we actually use visual cues in speech perception the McGurk effect so in person speech will always be easier.

@barbarabeeton But i agree with @Plergux on some bits of the commentary. One problem with these sorts of discussions is that journalists, despite having been told over and over again by linguists that "good language" is a social not linguistic judgement, it continues to creep into things. Of course these ideas are deeply embedded in most if not all cultures, since language is such a potent tool for group separation.

Hey guys, there's this duck-maintained cool package which was updated to version 3. :)

@PauloCereda ooh!

@Plergux ooh

gizmodo.com/…

@Plergux ^^

By default, the unordered lists should showed as black dots or bullets. But I don't understand why it's appearing as square boxes. Is it affected by the other packages?

@PauloCereda The next generation of package maintainer duckoftheday.co.uk/2021/05/17-may-2021.html

@samcarter_looks_forward_TUG'21 ooh :)

\begin{itemize}
  \item The individual entries are indicated with a black dot, a so-called bullet.
  \item The text in the entries may be of any length.
\end{itemize}

@raf some fonts have squares for the textbullet slot

@PauloCereda but... but... but... the burnt grill crap is the best bit! O.o

@Plergux Science ruins everything

@PauloCereda truth! :p

@samcarter_looks_forward_TUG'21 foofiee!

@Plergux :D

@samcarter_looks_forward_TUG'21 oh I see. I am actually using the package LatexBangla for writing in a Bengali font: ctan.org/pkg/latexbangla?lang=en

Is it possible to get back the default black bullet?

@raf Making a minimal working example instead of just a few code fragments helps in such cases

@raf yes, it is possible

oh, okay, I am showing a MWE.

@MarcelKrüger how do one activate the math tagging? I tried with \luamml_flag_structelem: but it seems to do nothing.

:58026827 \documentclass{article}
% For a bilingual document
\usepackage[banglamainfont=Kalpurush,
 banglattfont=Siyam Rupali]{latexbangla}
%activate polyglossia
\setdefaultlanguage[numerals=Bengali, changecounternumbering=true]{bengali}
%number all levels
\setcounter{secnumdepth}{5}
  \setotherlanguage{english}
\usepackage{csquotes}

\begin{document}
\section{Intro}
\begin{itemize}
  \item The individual entries are indicated with a black dot, a so-called bullet.
  \item The text in the entries may be of any length.

% !TeX TS-program = xelatex
\documentclass{article}
% For a bilingual document
\usepackage[banglamainfont=Kalpurush,
 banglattfont=Siyam Rupali]{latexbangla}
%activate polyglossia
\setdefaultlanguage[numerals=Bengali, changecounternumbering=true]{bengali}
%number all levels
\setcounter{secnumdepth}{5}
  \setotherlanguage{english}
\usepackage{csquotes}

\usepackage{enumitem}
\setlist[itemize]{label*={\fontfamily{cmr}\selectfont\textbullet}}


\begin{document}
\section{Intro}
\begin{itemize}
  \item The individual entries are indicated with a black dot, a so-called bullet.

@raf You could switch back to the default font for the textbullets. Or use some other roundish symbol of your choice

Thank you so much.

@raf You're welcome!

Btw, it showed the squares that means the black dots are set for something else?

@barbarabeeton Q: What do you call a bee that lives in America? A: USB

:)

@samcarter_looks_forward_TUG'21 The latexbangla package loads enumerate so there may be unintended consequences from loading enumitem too? It might be easier to just redefine \textbullet.

@PauloCereda Bzzt.

@AlanMunn ooh

@PauloCereda ROFL

@UlrikeFischer \luamml_flag_structelem: should work. Normally it should be used just before the end of the math equation to be tagged, but it should also work globally. Do you have an example where it didn't work?

@AlanMunn no, I don't think there are any unintended consequences of using enumitem. But I am feeling curious to know where latexbangla originally defined \textbullet ?

@raf I don't think it defines it, it's just the bullet glyph from the Kalpurush font.

@MarcelKrüger I tried this (the [ ] errors with \IMPOSSIBLE)

\RequirePackage{pdfmanagement-testphase}
\DeclareDocumentMetadata{pdfversion=2.0,uncompress}


\documentclass{article}
\usepackage{luamml}
\usepackage{tagpdf}
\tagpdfsetup{
 activate-all,%paratagging,
 interwordspace=true,
}

\usepackage{amsmath}

\begin{document}
\ExplSyntaxOn
\luamml_flag_structelem:
\ExplSyntaxOff
\tracingmathml=2
\tagstructbegin{tag=Document}

\begin{align}
  ax^2+bx+c=0 =\sqrt{4}
\end{align}

% \[ a=b\] %errors

\tagstructend
\end{document}

@AlanMunn ohh, I see. Now I get it. Thank you.

@AlanMunn Oh, thanks for the hint. In this case maybe \renewcommand{\labelitemi}{\fontfamily{cmr}\selectfont\textbullet} is safer

@raf ^^^ see @samcarter_looks_forward_TUG'21's comment.

@UlrikeFischer The align code hasn't been adapted and inserts it's own flags currently. The \[ line works iff you load tagpdf before luamml.

@MarcelKrüger ah, better.

@MarcelKrüger are the structure and the stream method exclusive or can one have both for one math expression?

@AlanMunn yes, I checked

@UlrikeFischer \luamml_flag_structelem: does still allow you to get a stream with \luamml_get_last_mathml_stream:e.

@Skillmon That depends of course. What I was getting at is that in expl3 a prop makes perfect sense for units (open-ended) but for a 'rigid' data structure like a (parsed) number, a dedicated set of functions can be better

@MarcelKrüger I tried with equation and it more or less works. I only have two math-structures, one from your code and one from my \tag_struct_begin:n{tag=math,AF=my/mathml#1}. How could I tell your code to use additional keys like the AF key?

@UlrikeFischer It's best to put the AF on a Formula tag surrounding the math tag.

@MarcelKrüger yes, I just got the same idea ;-)

The existance of such a Formula tag is required in PDF/UA and IMO is a good idea anyway.

@MarcelKrüger I get two math-structures, one with an empty mtext, which seems to be from the equation number, but it is not as kid in the structure.

@MarcelKrüger actually in the stream it is an artifact.

@UlrikeFischer Yes, that's a combination of two issues actually: 1. Equation numbers are not currently handled specially, so they just act like independent formulas. (Not sure yet how to represent them in MathML, as far as I can tell MathML only has equation numbers in tables) 2. The structure element code does not support handling \hboxes yet (or \vboxes, but they are complicated anyway)

@MarcelKrüger hm, is an equation number part of the equation?

@UlrikeFischer Especially boxes are quite interesting. I think the best approach to handle them is to use \everyhbox to automatically add something like \tag_mc_begin:n{tag=mtext,stash} to every hbox which starts in math mode, but I'm not sure yet how reliable that would be.

@UlrikeFischer I assumed they are, but I never really thought about it.

But since the equation numbers in an mtable are part of the equation, I think it makes sense to also include individual equation numbers (if only for consistency).

@MarcelKrüger well I think that with your code we have a good tool to create all sort of examples, which we can throw into the math discussion group and then they can fight it out ;-). Can one write the mathml stream also to some file and not only into the log?

@Plergux -- I rather enjoyed the photos of the visitor to the volcano roasting hot dogs on the fresh lava.

@barbarabeeton yeah, your hot dogs don't get much manlier than that :p

@LukeTheWolf The problem is that titlesec is incompatible with beamer

@samcarter_looks_forward_TUG'21 How can I position something both down and to the right of a node in TikZ? e.g. \node [below=.25cm AND right =.5cm of X]?

@AlanMunn \node[below right=.25cm and .5cm of X].

@TeXnician Thanks! I had the and in the wrong place.

@AlanMunn alternatively you could use use the calc library and calculate the coordinate based on X

@samcarter_looks_forward_TUG'21 Thanks. For what I'm doing (simple block diagrams) this works perfectly.

Good evening!

Random question: is it just me, or did the fonts used on the site change?

@samcarter_looks_forward_TUG'21 Thanks a lot! I didn't noticed it

@chsk there is featured link on every page to:

Ah!

@DavidCarlisle Leave it to me to ignore such things by default. ;) Thank you!

@samcarter_looks_forward_TUG'21 I see they decided against ubuntu font

@LukeTheWolf First rule of beamer: if something strange happens, blame a package and not the class :)

@samcarter_looks_forward_TUG'21 and if that fails, blame @UlrikeFischer

@DavidCarlisle I wonder why did not use your comic sans suggestion :)

@samcarter_looks_forward_TUG'21 they just don't have my flair for typographic design and elegance

@UlrikeFischer broke beamer apparently (@samcarter_looks_forward_TUG'21 needed someone to blame)

@DavidCarlisle you should sent them a sample of your typewriter package - they'll immediately hire you as designer

@samcarter_looks_forward_TUG'21 with background panels as subtly provided by colortbl documentation?

@DavidCarlisle yes :)

(would also make a good Halloween design for ctan :) )

@samcarter_looks_forward_TUG'21 now that's just offensive suggesting that I'd produce such a style:-)

@LukeTheWolf You're welcome! Have fun with beamer :)

@samcarter_looks_forward_TUG'21 This is about as likely as @DavidCarlisle reading documentation.

@samcarter_looks_forward_TUG'21 TY! I think I will use it for my automata lessons (ﾉ◕ヮ◕)ﾉ

@DavidCarlisle You don't want to improve the current Halloween design? You would become famous :)

@samcarter_looks_forward_TUG'21 29 days late.

@samcarter_looks_forward_TUG'21 -- I think "infamous" is the appropriate description. (He's already on his way.)

@barbarabeeton :)

@UlrikeFischer -- I didn't thank you yet for the greeting, but it's delightful. (I'm sure I've seen it before.) Nice to see Riley's smile.

@UlrikeFischer Currently not, but the same rule applies as for most other interface decisions in luamml: If you tell me a sensible interface for that, I'll implement it :)

@MarcelKrüger well interfaces are not so important here, I was more pondering about debug and test options. neither looking in the pdf directly and using adobe to click through the tags is really fast. It would be good to have a more or less well formatted xml directly.

@barbarabeeton ;-)

@HenriMenke My thoughts exactly.