TeX, LaTeX and Friends

6:00 AM

@DavidCarlisle What would this user do if eqnarray was removed? tex.stackexchange.com/q/647684/52406

7:02 AM

@mickep switch to conetext?

@DavidCarlisle Oh no, poor user.

I am curious on why someone "sadly has to use the outdated eqnarray instead of align for many reasons". But I will not interfer and ask.

3

@mickep reasons can be wrong but still numerous

every time I open the main site firefox closes and sends a crash report to mozilla:(

@DavidCarlisle shocked by the evil palindrom?

@UlrikeFischer possibly, but very consistent even after rebooting

7:20 AM

@DavidCarlisle I thought you used emacs.

TeXnician

Interesting new project for pandoc users who want MD→HTML conversion with LaTeX fragments and custom fonts etc.: github.com/ma-chengyuan/just-latex

@mickep They stopped including w3-mode by default, shame it was a nice browser and quicker than mosaic on many pages

@DavidCarlisle, @UlrikeFischer Still seems like the case changing code update is OK :)

@UlrikeFischer I see you are going to talk about spot colours ...

@JosephWright it broke one of my hyperref tests. $a=b$ is not longer converted to $A=B$ ;-). Btw do you have an idea why I get such diffs:

***** .\BUILD\TEST-CONFIG-3\refstarred-hyperref.luatex.tlg
    5:  ->\IfBooleanTF {#1}{\@refstar }{\T@ref }.
    6:  > \Ref=document command:
***** .\BUILD\TEST-CONFIG-3\REFSTARRED-HYPERREF.LUATEX.LOG
    5:  ->\IfBooleanTF {#1}{\@refstar }{\T@ref }.
    6:  <recently read> }
    7:  l. ...\ShowCommand\ref
    8:  > \Ref=document command:
*****

@JosephWright we'll see how many issues come in to textcase, for which I need that label

7:33 AM

@DavidCarlisle :)

@UlrikeFischer No ... something altered in \ShowCommand?

@JosephWright I thought I could make some use of the hours I spent to work around various colorspace/xcolor/pgf incompabilities and bugging @MarcelKrüger to add support in luaotfload.

@UlrikeFischer We should finalise l3color as the way forward: spot colours should then 'just work'

@JosephWright spot colors work in l3color. That is not really the problem. The problem is to use l3color with pgf. If pgf would simply use "official" (x)color interfaces it would be (rather) trivial. But it has in various places low-level code to extract/convert fill and stroke components.

@UlrikeFischer Ah, right, yes, but probably because color (unlike l3color) doesn't have interfaces for fill and stroke separately

@UlrikeFischer All sortable if we make the commitment, I suspect

@JosephWright and the other problems are objects like shadings and pattern: there are rgb and with some struggles cmyk. But that means you can only use grayscale.

7:40 AM

@UlrikeFischer Ah, yes: I'm meaning to look at those for l3draw

@DavidCarlisle oh no.

@JosephWright yes but we would imho also need the commitment of the pgf people. It is imho rather hard to do it all from the outside.

7:54 AM

I gave up with academia.com years ago but they still email from time to time, I like this one that just came in...

@DavidCarlisle Sigh

yo'

@DavidCarlisle very cool! In particular getting synced to 9s from 6s in a bit over a day :)

I'm sure everything I wrote in 97 is still true

@yo' lucky coincidence?

yo'

@DavidCarlisle don't take me wrong, but I don't believe in these :D

@DavidCarlisle Probably

@DavidCarlisle latex-project.org/publications/… right?

8:02 AM

@JosephWright they don't appear to tell where they found it, I was going to google for it, so no idea:-)

@DavidCarlisle ??

@DavidCarlisle Ah right

@DavidCarlisle I meant 'most of it is still right', mainly

@JosephWright that may have been an accurate description of the paper at the time:-)

@JosephWright seems to be in texlive I get

> \blub=document command:
#1:s
#2:m
->zzzz.
<recently read> }

l.55 \ShowCommand\blub

and in miktex which is still older

> \blub=document command:
  #1:s
  #2:m
->zzzz.

So I can probably blame @PhelypeOleinik ?

@UlrikeFischer Looks good to me ;)

8:27 AM

@DavidCarlisle It's interesting how much carried through: it's all spot-on up to the refactor just after I joined the team

@UlrikeFischer I do wonder when we'll get the first bug report that $a$ is no longer case changed ...

@UlrikeFischer I think it works if you load luaotfload-l3color without the l3opacity part (or maybe try to load all of it and don't use transparency). That should happen automatically once this gets into a better space.

@MarcelKrüger what do you mean by loading it without l3opacity? I don't load anything explicitly here. I mean if we want to update luaotfload then standard documents not using new code shouldn't break. (It doesn't work without the patched fontspec either but it works with the current luaotfload).

If it doesn't work without the patched fontspec then that should be fixed, is supposed to be fully backwards compatible (except for some error handling)

@MarcelKrüger ok let me check again ;-)

@MarcelKrüger I still don't get colors and I get warnings. And I think I use all the correct files. Can't you reproduce it?

@MarcelKrüger oh now it disappear after I disabled Franks tree. I need to check what it there.

8:48 AM

@UlrikeFischer ooh die Eiche

@UlrikeFischer I can reproduce (interestingly only for color, transparency seems to work) with current TeX Live even without my luaotfload changes...

@MarcelKrüger well in my case it was due to a local version of the patched fontspec-luatex.sty. So what can we do here? without l3color/pdfmanagement one needs the unpatched fontspec and with one needs a patched fontspec.

I take my issue back, I had forgotten that color names only work with xcolor.

@UlrikeFischer Am I right to assume that currently the unpatched fontspec always works when the old version worked and the patched fontspec is only needed when new behavior is wanted?

Then it should be easy to release the new luaotfload, then update l3color to automatically load something like luaotfload-l3color and then finally update fontspec.

@MarcelKrüger Sounds good

9:08 AM

@MarcelKrüger but l3color is in the kernel, so if loads l3color.lua it will probably break something if the pdfmanagement is not loaded too. So probably we should put the code first into pdfresources: if the pdfmanagement is loaded it can load the lua code and trigger the patch in fontspec.

I’m hereby founding the International Society Against Mismatched Text And Math Fonts (ISAMTAMF). Who is with me?

@Gaussler 𝙼𝘢𝘁𝕔𝒉𝕖𝕕 𝕗𝙤𝘯𝓉𝖘 𝙖𝗋𝘦 𝗶𝑚𝚙𝖔𝓇𝖙𝒂𝘯𝘵

@DavidCarlisle Indeed ;-). What I’m really, really tired of is people loading times for no reason and then don’t do anything about the math font, and the awkwardness of the whole setup never really seems to occur to them

9:24 AM

@DavidCarlisle oh

@PauloCereda 𝘣𝘳𝖾𝓪𝑘𝚏𝒂𝔰𝗍

@DavidCarlisle 𝙤𝒉 𝘯𝙤

@MarcelKrüger this here breaks (without the patched fontspec):

\directlua{require"luaotfload-l3color.lua"}
\documentclass{article}
\usepackage{fontspec}
\usepackage{xcolor}
\setsansfont{texgyretermes}[Color=red,Opacity=0.1]
\begin{document}
TTT \sffamily MMM
\end{document}

! Undefined control sequence.
<recently read> IMPOSSIBLE.

5

9:42 AM

@UlrikeFischer ooh

@MarcelKrüger I think you should update luaotfload. And I put the loading of the l3color.lua and the fontspec patch into the pdfresources firstaid for now.

@UlrikeFischer Great, but please move the entire luaotfload-l3color.lua there and don't just load it. I want to release without that file.

@UlrikeFischer That error should only be about the l3opacity parts (since they aren't fully activated without \DocumentMetadata. If you remove the l3opacity parts of the luaotfload-l3color.lua file then it should be safe to load it even without \DocumentMetadata and work with the new fontspec.

9:58 AM

@MarcelKrüger copied. I will rename it so that it is clear that is at a temporary location.

@MarcelKrüger yes I guessed that. We should perhaps split it later.

10:17 AM

@DavidCarlisle anorien.csc.warwick.ac.uk/mirrors/CTAN/macros/luatex/latex/… - see the quote

3

@JosephWright @MarcelKrüger did anyone check these packages? I took a look at the spacekern package from this author and directly managed to get an error in a tikz picture with \duck; in the code so I'm a bit vary.

@UlrikeFischer I've not looked

10:33 AM

\documentclass{ltxdoc}
\usepackage{hyperref}
\begin{document}
\end{document}

! LaTeX Error: Option clash for package hyperref.

See the LaTeX manual or LaTeX Companion for explanation.
Type  H <return>  for immediate help.
 ...

l.54 ]{hyperref}[2002/05/09]
                            %
?

What is this?

Happens on the latest IoT Docker image (81a2564ec8cb).

Skillmon

10:48 AM

@JosephWright what's with the monospaced es in this PDF?

@Skillmon ???

Skillmon

@JosephWright ^^^

Strange fat e.

@Skillmon Looks fine on my system.

@Skillmon ^ This is how it looks here.

@HenriMenke ltxdoc now loads hyperref (but with other options)

Skillmon

11:02 AM

@mickep zooming in fixes it, PDF.js was messing up

@UlrikeFischer We get the same issue in luaotfload... Shouldn't \usepackage{hyperref} (without any options) be ignored if hyperref is already loaded?

@MarcelKrüger I would have to adapt hyperref to the new key system. I can take a look later.

@MarcelKrüger @UlrikeFischer Could it do \PassOptionsToPackage or \hypersetup instead? The new behavior broke all the manuals for the different PGF packages.

11:28 AM

@UlrikeFischer I haven't changed \ShowCommand in a while. The last change was adding ltcmd support. I can take a look though, what test is that?

12:06 PM

@JosephWright :-)

@PhelypeOleinik I suspect more a change in NewDocumentCommand and friends. Something simple as that is enough:

\documentclass{report}

\NewDocumentCommand\blub{}{xx}
\ShowCommand\blub
\begin{document}

\end{document}

with the new latex it show additional lines in the log:

> \blub=expandable document command:
->xx.
<recently read> } %<--------- new
                           %<---------- new
l.53 \ShowCommand\blub %<----------new

@UlrikeFischer Oh, three lines are printed by the engine, with very little control at the macro level. Those lines should always appear...

@PhelypeOleinik well they didn't before, that's why my tests broke.

12:24 PM

@UlrikeFischer The tests were wrong before :)

@PhelypeOleinik :)

@PhelypeOleinik in miktex which has still 2021-11-15 I get this:

> \blub=expandable document command:
->xx.

(C:\Users\Nililand2020\AppData

@MarcelKrüger I looked at it. hyperref is actually not loaded yet. doc loads hypdoc only at begin document so if hyperref would use the new key processing (which is lots of work) one would loose all the hypdoc option. Imho the only way to avoid the error would be to change doc to pass the hypdoc options to all subsequent calls.

@HenriMenke there was already an issue about this, github.com/latex3/latex2e/issues/822. If you want some change here you should perhaps comment there.

@JosephWright is the documentation of l3keys2e actually up-to-date?

@UlrikeFischer Er, I think so

@JosephWright but it doesn't mention the new optional argument?

@UlrikeFischer That's not l3keys2e, that's the kernel code

12:38 PM

@JosephWright oh, one s less ;-)

a bit confusing.

@UlrikeFischer ah, got it

@UlrikeFischer github.com/latex3/latex2e/issues/739

@PhelypeOleinik oh my. You want to blame me.

@UlrikeFischer obviously ;-)

@UlrikeFischer On 2021-11-15 It used \iow_term:n, so the error context lines don't appear. Now it uses \tl_show:n, which stops as an error

1:23 PM

@egreg This sounds like a job for you. In particular, how can you measure the height of \delimsize| without accidentally opening a new math environment?

@Gaussler How in earth one would want a dashed line instead of a colon or a bar?

@egreg Haven't you always wished for a green - white - red one?

@egreg I could think of cases where such a notation could be desirable, in order to distinguish it from the standard set construction notation. Perhaps you are doing higher category theory and want the \infty-subcategory where some relation holds only up to homotopy or something?

@Gaussler I disagree. Anyway, look at the proposal.

@mickep green - white - red? This sounds like a good starting point for an xmas math package :)

1:36 PM

@samcarter I was thinking of the Italian flag, but if the Christmas entry is more motivating, why not? :)

@mickep depending on the season, it could be used for both :)

@samcarter Even more motivation.

ooh

2:29 PM

@egreg I tried updating your answer to allow automatic scaling, but the solution cannot detect `\displaystyle`. Can this be fixed somehow?
```
\documentclass{article}
\usepackage{mathtools}

\ExplSyntaxOn
\DeclarePairedDelimiterX\set [2] \lbrace \rbrace
{
\mspace{1.5mu}
#1
\tl_if_blank:nF { #2 } { \mathrel{\dashedsep{#1}{#2}{\delimsize}} #2 }
\mspace{1.5mu}
}
\ExplSyntaxOff

\makeatletter
\newcommand{\dashedsep}[3]{%
\sbox0{$\m@th\left.#1#3\vert#2\right.$}%
\vcenter to \dimexpr \ht0+\dp0 {\offinterlineskip

Not sure what went wrong with my code markers

@Gaussler Oh, well, it doesn't work with \left and \right. For that you need to measure the whole formula in advance.

@egreg That’s what I’m doing in the above example, more or less. The problem is that it cannot detect \displaystyle

Is there some \getstyle command?

@Gaussler Long detour with \mathchoice…

@egreg Omg

2:56 PM

@UlrikeFischer So Frank (or someone else) pushed the breaking release but not the release notes. So everybody's documentation is now broken and nobody knows why.

@Gaussler you used ``` which is not supported here. chat markdown is as much like q&a markdown as javascipt is like java....

@DavidCarlisle Javascript always did remind me so very much of an Indonesian island

@MarcelKrüger (@HenriMenke) we should probably make that happen, currently it isloading \AtBeginDocument so after the no-option load in the preamble

@Gaussler Java was supposed to

@DavidCarlisle A bit the same way as I pronounced your name “Carlizzle” for the first five years or so, before someone broke the truth to me ;-)

@Gaussler we did same between my siblings, just to be annoying

3:08 PM

@DavidCarlisle I can adapt hypdoc to test and give a warning if hyperref has already been loaded. Then it is up to the user to setup the right options. Would that be ok?

@HenriMenke Better description would be a change which had no reported issues in some weeks of latex-dev testing shows some after full release. workaround being just remove hyperref, but we can (probably) make the unchanged file not error, especially if you report an issue.

@UlrikeFischer but that is just the option clash error with different words. I think better might be to simply not load hyperref if already loaded

@DavidCarlisle yes I meant that, but along with a warning that some options from hypdoc are missing because of this.

@UlrikeFischer oh ok

@DavidCarlisle Some of my packages (in this case pgf-pie) move very slowly, so it is entirely possible that even if I used latex-dev to build the documentation I would miss the interval in which the breaking change is only in latex-dev. The other problem is that the error message is just a generic option clash with no further traceback and there are no new release notes on latex-project.org/news/latex2e-news.

@DavidCarlisle Well, you can always try to mimic Hyacinth Bucket and insist that this is the proper pronunciation of your name, just to keep up appearances and you know

3:17 PM

@HenriMenke yes I know, but it is hard to test every combination so we do what we can with dev testing. If bad interactions forseen we would probably have avoided them not documented them, only way to never have unexpected consequences is never change anything. we tried that 2000-2015, but that not a total win either.

@HenriMenke @DavidCarlisle I added the change and will build and push after I had some tea ...

@UlrikeFischer It is important with tea.

@Gaussler There is a whole series of educational videos on how to pronounce Carlisle youtube.com/watch?v=jSDTxbTrHqE&t=59s (the entertaining effect of the tv series was clearly only a side effect of this :) )

@samcarter ;-). Well, they are hopefully doing a better job than Game of Thrones is when trying to teach us how to pronounce “Tyrell”.

@mickep it definitely gives a lot of tranquili-tea

3:27 PM

I believe they have about four or five in-universe pronuncations

@Gaussler :)

@samcarter Otherwise, you can imagine how confused the Icelandic must be when they go to school and learn that the name they are used to calling their country means “island” in English (in Icelandic, there is an accent on the I, but not in other languages like Danish)

3:47 PM

@Gaussler the German word for "Iceland" is "Island" :)

@PauloCereda Oh, you are on fire today!

3:58 PM

@mickep ooh

4:13 PM

@UlrikeFischer ooh twitter.com/hyxpk/status/1536740258565353474

@DavidCarlisle v v v

\documentclass{ltxdoc}
\usepackage{csquotes}

\begin{document}
aa
\end{document}

! LaTeX Error: Command \enquote already defined.

@UlrikeFischer Ooops

@UlrikeFischer "only way to never have unexpected consequences is never change anything."

7

@DavidCarlisle ;-) I would suggest to surround the \providecommand by \AtBeginDocument ...

@UlrikeFischer if only csquotes had used \def like a proper package

4:26 PM

@DavidCarlisle blame @JosephWright

@UlrikeFischer sounds about right

@UlrikeFischer that doesn't seem fair

Nov 27, 2020 at 21:16, by Paulo Cereda

Apr 10 '18 at 13:16, by Paulo Cereda

6

@DavidCarlisle github.com/josephwright/csquotes

@UlrikeFischer still seems a poor excuse for not blaming the usual culprit

hypdoc on the way to ctan.

5:12 PM

2

yo'

5:23 PM

I wonder, is there an explanation of the output of \tracingparagraphs somewhere?

@yo' It's explained in the TeXbook between exercises 14.11 and 14.12.

5:39 PM

@PauloCereda shame, you can't eat cats

@DavidCarlisle oh

@DavidCarlisle will you make the fix? You already know how it works with a hotfix ;-)

Faheem Mitha

I ran into an issue which is probably a dupe of tex.stackexchange.com/q/122808/3406

Example code is gist.github.com/c883f322b7d04de946d46561f35cd9c2

@UlrikeFischer move all of these inside the hook?

\providecommand\url{\texttt}
\def\task#1#2{}
\providecommand\LuaTeX{Lua\TeX}
\providecommand\cls{\textsf}
\providecommand\pkg{\textsf}
\providecommand\enquote[1]{``#1''}
\DeclareRobustCommand\cs[1]{\texttt{\bslash\detokenize{#1}}}%
\AtBeginDocument{%
  \renewcommand\PrintMacroName[1]{\MacroFont\detokenize{#1}}%
  }

Faheem Mitha

Can anyone confirm this a dupe? Is it worth posting a new question? If it isn't clear, the issue is that pgfkeys strips a pair of braces for no reason, that I can see. And did anyone ever report this?

5:47 PM

@DavidCarlisle Frank seems to be on the work too so better ask him ...

@DavidCarlisle ooh fishing

@UlrikeFischer or do nothing

@DavidCarlisle and wait for the issue to solve itself?

is the enquote test reported anywhere?

I like this plan

5:49 PM

@PauloCereda usually works

@DavidCarlisle ooh

@DavidCarlisle no I didn't create an issue.

@UlrikeFischer I just caught up with email and saw discussion there

7:05 PM

@DavidCarlisle or someone intimately familiar with how \trivlist works: For this question, I suspect that \trivlist is started with a \vspace instead of \addvspace. but my laptop keeps hanging while trying to research the definitions in latex.ltx. If that's so, please post an answer; after an hour of rebootiing the laptop, I'm a bit stymied.

@barbarabeeton which question about lists?

I just discovered a new page specification in TeX called "textblock". Can anyone refer me to how it is defined? By what command?

@DavidCarlisle -- the complaint that there's such a large gap between a display and a following itemize, more than between an equation and a following paragraph: tex.stackexchange.com/q/615437

@DavidCarlisle -- sorry about that (leaving off the link); this misbehaving laptop has got me rattled.

Hi everyone, how to elongate figures beautifully something like this

@tush Is it related to tempdima?

7:14 PM

@barbarabeeton starting a paragraph with a math display never works well, the list is just a special case of that probably

@tush it is not a standard latex term

@CroCo Is it something in particular with those plots that you want to do that you cannot?

@DavidCarlisle -- But the display in the example is at the end of the previous paragraph, before the list.

@tush ?? \@tempdima is just a scratch length for local use, and has no general meaning in latex

I couldn't find it in the index of the TeX book. And not in source2e.dtx
Could you give me more info about it?

@mickep I'm trying to upload mine but it failed to upload it here.

7:17 PM

@barbarabeeton I'll look in a bit

Hmm.. I guess that `temp` stands for temporary..
Either way, could you give me a bit of info about what a textblock is?

I couldn't find it in the index of the TeX book. And not in source2e.dtx
Could you give me more info about it?

@DavidCarlisle -- Thanks. I'd really like to know the answer too.

@CroCo Do you want to upload an image to a question?

@tush Isn't the textblock just where the text goes (not headers/footers/margins)?

@tush no I have no idea what you mean, a "text block" is a general english term for a block of text but no special meaning in latex

7:20 PM

@mickep I discovered about it exactly in this context.

this is mine.

but I feel if I squeeze it, it will affect the quality.

I think one way is to change the initial values of the experiment and increase the running time. :>

@CroCo So, the aspect ratio could be better. You are wasting a lot of vertical space with the legend. If you remove that and change the aspect ratio it will look better.

@mickep I think the problem with green initial value.

@mickep I am writing my own cls file and until now I set up only the paper size with pdfpageheight and pdfpagewidth. But apparently this is not enough to fully specify the document boxes?

@CroCo hopefully they were not stretched by latex, but generated that size

7:25 PM

@DavidCarlisle this is my guess because I've checked their initial values and they are close to each other.

@CroCo Sorry, I simply do not understand what you are asking for. I hope someone else does.

@tush read the documentation of geometry.

@mickep the first picture I've posted taking the entire 4A paper but they look beautiful.

@tush Oh, I never looked closer at any cls files. I suggest you do as @UlrikeFischer tells you, and perhaps also look how they existing classes do to set this up.

@mickep I hovered over the article.cls file and size10.clo file but I don't understand what else is missing.

7:27 PM

Is it possible they are stretched?

This is from the pdf

@CroCo So it is the aspect ratio you want to change? You want your plot wider and not so tall?

My question is more art than latex. :)

Do we have artists here?

sorry for this tall order.

I gave up Matlab because their figures are horrible.

@UlrikeFischer Aha! pdfvorigin and pdfhorigin!

@CroCo One can read about these kind of questions in Tufte's books.

@CroCo my artistic skills are well known

7:38 PM

@tush I doubt that you need to set them.

@UlrikeFischer Actually when I set them to 0 I get this: i.stack.imgur.com/txz2o.png . I want the Lorem ipsum to begin at the left-most side of the page, as it begins with the up-most part of it.

@DavidCarlisle dudddde. I laughed so hard. what the hell was that

@CroCo unlikely the text is not distorted, but surely any plotting software (you have not said what you are using) lets you control the scale of x and y axes so you can make a long thin plot

@CroCo use that instead of those boring plots. Your readers will be happier.

7:48 PM

@DavidCarlisle "Never Let Them Know Your Next Move" I may surprise them with that