6:07 AM
@JosephWright great, now I have to learn the new siunitx syntax :(

6:51 AM
@Skillmon For most users, it should not be a big thing - the existing syntax 'just works'

@UlrikeFischer so slow

@DavidCarlisle typing on the smartphone is hard ;-)
@JosephWright is detect-weight gone or was that only in my local copy?

@UlrikeFischer It's there as a deprecated key - you want reset-text-series and perhaps text-series-to-math as a 'proper' replacement
First siunitx v3 issue: github.com/josephwright/siunitx/issues/416

7:29 AM
@UlrikeFischer Working for you?
\listfiles
\documentclass{article}
\usepackage{siunitx}
\sisetup{detect-weight}
\begin{document}

\qty{10}{\metre}
{\bfseries \qty{10}{\metre}}

\end{document}
 *File List*
article.cls    2021/02/12 v1.4n Standard LaTeX document class
size10.clo    2021/02/12 v1.4n Standard LaTeX file (size option)
siunitx.sty    2021-05-17 v3.0.0 A comprehensive (SI) units package
translations.sty    2021/01/17 v1.10a internationalization of LaTeX2e packages
(CN)
etoolbox.sty    2020/10/05 v2.5k e-TeX tools for LaTeX (JAW)
pdftexcmds.sty    2020-06-27 v0.33 Utility functions of pdfTeX for LuaTeX (HO)
infwarerr.sty    2019/12/03 v1.5 Providing info/warning/error messages (HO)
@UlrikeFischer shows the expected bold for me

7:46 AM
@JosephWright no, not bold. My version is probably a bit older, let me update.
@JosephWright with the newest version it works again.

8:22 AM
@JosephWright small thing (but maybe it's me):
θ91° [romano:~/tmp] % texdoc -l siunitx
1 /home/romano/texlive2021/texmf-dist/doc/latex/siunitx/siunitx-code.pdf
= Code documentation
2 /home/romano/texlive2021/texmf-dist/doc/latex/siunitx/siunitx.pdf
= User manual
Enter number of file to view, RET to view 1, anything else to skip:
Now texdoc siunitx gives by default the code, not the manual (maybe this is to please @DavidCarlisle?)
I do not know if it's related to the siunitx upgrade, but I had to remove the .aux files because I had this:
! pdfTeX error (font expansion): auto expansion is only possible with scalable
fonts.
<argument> ...shipout:D \box_use:N \l_shipout_box
Anyway, the circuitikz manual compiles ok with the new siunitx release: nice work @JosephWright!

@Rmano for me its the other way around.

Yep, I think I have a texdoc acting strange here. No idea why though, I just added my local TEXMF to TEXDOC...

8:50 AM
@Rmano I'll report to the TL list
@Skillmon Same here: order is 'correct'

@JosephWright wait, it can be a local thing.

@Rmano That one sounds like the kernel upgrade

@JosephWright yes, probably --- cured itself by rereunning
@JosephWright Found; the order is correct if I unset TEXDOCS. But with TEXDOCS=/home/romano/texmf/doc//: the order is reversed, even if there is nothing about siunitx in my home texmf

@Rmano Cool
I think I've tested all the obvious things for siunitx: all the old manual (which was the test file, really)

9:23 AM
All this talk about texdoc made me remember a reoccurring problem of mine. Until TL2020, texdoc symbols would give me the comprehensive list of symbols and having typed this very often, I'm pretty used to it.
But with TL2021 I now get
texdoc -l symbols
1 /usr/local/texlive/2021/texmf-dist/doc/fonts/newcomputermodern/newcm-unimath-symbols.pdf
= Math coverage and symbol comparison
2 /usr/local/texlive/2021/texmf-dist/doc/latex/comprehensive/symbols-a4.pdf
= The document (for A4 paper)
...
Is there any local configuration I could change to get symbols-a4.pdf if I type texdoc symbols?

@samcarter_looks_forward_TUG'21 Same here, with or without TEXDOCS defined.

9:41 AM
... maybe I could work around this problem with a bash function superuser.com/a/479816/383391 ....

@samcarter_looks_forward_TUG'21 moan on texlive list and they'll adjust the default priorities

@DavidCarlisle Good idea!
I did something totally crazy: I read the documentation and it turns out I can simply set an alias with alias symbols = symbols-a4.pdf in my texdoc.cnf file

9:58 AM
@samcarter_looks_forward_TUG'21 you did what?

@Skillmon yes, crazy, as I said. I hope this won't have any lasting ill effects ...

my dsl is broken, and I tried to report the disturbance. They want a email to report back in the form and reject everyone I give them as invalid ;-(

@UlrikeFischer :( companies from the telecommunications sector always seem to be the worst with communication stuff...

10:14 AM
First siunitx v3 bug reports are arriving: I expect there will be a few v3.0.x releases

10:33 AM
@JosephWright about that: is the \ensuremath{\text{what}} the correct way to define the unit? (For example, consider ppm --- parts per millions).

10:46 AM
@samcarter_looks_forward_TUG'21 but my first suggestion would save others from having to read the doc

10:57 AM
@DavidCarlisle I opened an issue on github

@samcarter_looks_forward_TUG'21 can you point it to me? Maybe I will add mine with TEXDOCS

@Rmano here you go github.com/TeX-Live/texdoc
(shockingly the github repo was mentioned in the doc) [ok, this reading the doc stuff really has lasting ill effects, should stop it asap]

11:47 AM
@Rmano I'd just use \DeclareSIUnit{\ppm}{ppm}

12:05 PM
@JosephWright yes, that's what I have in my docs...
@samcarter_looks_forward_TUG'21 github.com/TeX-Live/texdoc/issues/67 let's see... maybe it's in the manual. Checked but not found (even worse that reading the docs and find things ;-P)

12:29 PM
@Rmano I'm curious what the answer will be. That's really funny behaviour :)

1 hour later…
1:42 PM
@PhelypeOleinik feature request for lipsum, perhaps \lipsum[] should check the value of the option and complain if [] is just given. Currently we only get an interesting \__int_eval_end: error.

@daleif Debatable if feature request or bug fix ;-)
@daleif Thanks, I'll take a look

Stupid question: Does anyone think the code in tex.stackexchange.com/a/597578/117050 deserves to be in a package?

@daleif What do you think \lipsum[] should do? Nothing, same as \lipsum, an error, or something else?

@Skillmon wasn't there talk of having a standard pgf option to use l3 numerics just as there is to use lua ?

@Skillmon For sure it is useful...
I bookmarked it. A bit complex the usage of nested function, but if there is no other way...

1:57 PM
@DavidCarlisle none that I'm aware of, but it is pretty likely that I missed things in pgf-development :)

@Skillmon it's come up before somewhere not sure if it was in a gh issue or just chat here

@DavidCarlisle the issue is that you can't do the obvious \pgfmathdeclarefunction{foo}{1}{\fpeval{#1}} as that doesn't ensure the numbers being in a format l3fp understands... Hence the strange definition I use, so that it works in every context.
@Rmano well, the nested functions are just there to ease the code a bit, so it can be used generally without prior knowing the number of arguments. My original mock up here is less complex.

@Skillmon yes, can't look in detail now but still being a pgf library rather than a wrapping package might make sense even if it doesn't get into a core pgf distrib?

@Skillmon the ideal thing would be to have pure l3fp function definitions and doing everything with them... I know it's complex though. I often use a lot of declare function in my exams and avoiding rounding errors is a pain...

2:15 PM
@PhelypeOleinik if possible, I'd say either trow a proper error user might not understand that this comes from \lipsum[] or use 1 and throw a warning.

@Rmano Prod Bruno - he's on L3 code at the moment

@JosephWright Yes, I know, but I feel better not asking for difficult things when I have no idea on how difficult it could be ;-P . And for pgf, the nice option would be a "use l3fp everywhere"... again, probably too difficult for now.
But your macro as a stopgap seems really worth having.

@Rmano WE have code that can do it, it's just its not finished

@JosephWright nice! I will patiently wait ;-)

2:31 PM
@JosephWright thanks!

2:48 PM
@JosephWright imho, not Bruno's job. One could prod him to implement the missing functions to reach full pgfmath scope (poor Bruno), but implementing a pgf option to use l3fp instead of the fpu or the default implementation is more like Henry's job (and I duck somewhere so nobody sees me and gets the funny idea that I should do that)

@Skillmon I meant the idea of custom functions, not an l3fp-to- pgf bridge

@JosephWright does the team really have code to run l3fp in pgf?

@Skillmon I think we messed around a bit - it seems reasonably doable given the way pgf is set up

@JosephWright well, as long as it doesn't work, my hack can fill the gap. Now, should I implement it as a pgf library or as a standalone package?! (standalone package has the charme that I don't feel obliged to integrate it into the pgfkeys-interface for define function)

@Skillmon I think probably as a stand-alone - it's not really 'part' of pgf in any sense

2:51 PM
@JosephWright well, a pgf-library doesn't have to be part of pgf. See the tikzlings library, the beamer-overlay-thingy, etc.

@Skillmon Yes, but they shouldn't have dependencies beyond pgf or other libraries, whereas what you are talking about clearly does :)

3:06 PM
@JosephWright good argument :)

2 hours later…
4:41 PM

@samcarter_looks_forward_TUG'21 some people think 25 years is not an acceptable wait time for bugs reports

@DavidCarlisle Today's youth -- no patience :)

5:41 PM
@UlrikeFischer luamml's pdfstructelem branch now got support for writing into a file. See the test_structelem.tex example in the repo for details.

6:04 PM
@MarcelKrüger if one forget the command to set the file name, one get a low-level error ;-). Could we have newlines before a starting tag?

6:57 PM
@UlrikeFischer The missing filename case is fixed (The intended behavior is that you don't get a file if no filename is specified)
@UlrikeFischer The file output now has "pretty printing" enabled. It's overdoing it a bit, which is the reason why it was disables for log output and similar.

1 hour later…
8:07 PM
@MarcelKrüger very nice ;-)

8:34 PM
@UlrikeFischer very spacy:-)

@DavidCarlisle did you try?

8:57 PM
 <mrow>
<mi>
𝑎
</mi>
<msup>
<mi>
𝑥
</mi>
<mn>
2
</mn>
</msup>
<mo lspace="0.222em" rspace="0.222em" tex:class="bin">
+
</mo>
<mi>
𝑏
</mi>
@UlrikeFischer yes:-)

@DavidCarlisle ;-). What do you think one should do with the equation number?

@UlrikeFischer it's tricky you really want that to be a counter with document scope but mathml can only spec an individual math element it doesn't control the document. there is mlabeledtr but it was a bit of a compromise, we may try to do something better in mathml4, but it depends what we can get done

@DavidCarlisle do you think it really belongs to the math? It is not more like a caption and so "outside"?

@UlrikeFischer logically it's not part of the math (and it's set in text mode) it's a document level counter like section counters, for a single line equation you could reasonably handle it completely separate to the mathml, but align makes the placement tricky if it is being handled external to mathml

@DavidCarlisle yes multiline equation could be tricky. But couldn't the equation part have an id and the counter reference this?

9:11 PM
@UlrikeFischer possibly but how do you see the equation number position being specified? and is fleqn a property of the mathml or of the surrounding document? In the nag documents we only have single equation numbers (so multiline expressions like aligned rather than align and we handle the equation number purely in the surrounding html not in the math at all.

@DavidCarlisle well I don't know really much about the mathml details, but it sounds a bit odd to try to encode a number position. Even if it on the right, I would expect a link to jump to the left.

@UlrikeFischer linking is a different issue, like a section heading if you link to it you want to hit the whole expression, so top left anchor typically) wherever the number is tyoeset. but somethining needs to specify where the number is printed:-)

9:28 PM
@DavidCarlisle yes, but it doesn't need to be a full tag, it could be an attribute?

@UlrikeFischer well it could be but attributes don't render in general (and ones that do are tricky on the web platform which is why <mfenced> is dropped from mathml core so you do <mrow><mo>(</mo>... not <mfenced open="("...
@UlrikeFischer actually for tagged pdf it's probbaly simpler as the mathml tagging is basically a sematic annotation so it's probably clearer you should do the logical thing and tag the formula and equation number separately. In mathml-in-html it's trickier as you need to do something to make the number appear in a place coordinated with the formula rendering

@DavidCarlisle the coordination is what I find difficult to imagine with a full tag. I would find it easier to have something like <subequation id=2> and then lets some css decide if the number is left or right or inside a duck.

@UlrikeFischer yes that is one possibility but there are issues about calculating the space required if it is all handed off to css (and then not specified there) and of course not all mathml systems have css

9:47 PM
@DavidCarlisle yes, but numbers in the middle of an equation makes it probably more difficult for screen reader. Sounds as if one need more than one mathml to satisfy everyone ;-)

10:03 PM
@UlrikeFischer as I say for pdf use I think you will want it outside, the situation in web browsers is always more complex with multiple competing pressures,

@DavidCarlisle yes, we should probably do this for pdf, we only need a good "outside" place. Perhaps one should do something like <Formula><Lbl>[itex] with nested Formulas if needed.