@UlrikeFischer @DavidCarlisle I'm wondering what the best solution is to fix this mparhack problem. As far as I can see the mparhack "fix" relies on messing with \hb@xt@. I think at this point I'm more inclined to remove the autoloading of mparhack, leave it as an open bug (as it is not really memoir related) and mention mparhack and the needed patch in the manual. I don't know how many users use twocolumn mode combined with marginpar.

@daleif I have the consent of the author to adapt the package and will upload a version with Frank 's fix. If you test files it would be nice if you could check if it works everywhere.

@UlrikeFischer cat it be fixed? it seems to be to rely on hooking into a kernel shortcut macro which does not seem nice

@daleif well in the long run, marginpar should be fixed not to required such a package. But mparhack works since 20 years, so it can't be a complete disaster to keep it a bit longer.

@daleif it's a hack (the clue is in the name) but I think we can safely make it work as before and not disturb memoir which would be a good thing for users (and politically:-) as @UlrikeFischer says longer term it would be good to have a fixed marginpar in the kernel (and that might be needed so you can do correct pdf tagging anyway) so you may want to hold off any memoir adjustments until then.

@DavidCarlisle not a problem. Luckily I only load mparhack if twocolumn class option is active. So we don't hit that many users.

BTW: note that \tag* issue with amsmath on github. Not quite why what the problem its, but I noticed it works just fine if one uses align instead of equatuion

@daleif thanks I'll try and look

Good morning everyone! Quick question relative to tex.stackexchange.com/a/595446/38080: is there a standard way to define a function in l3fp? As the original declare function in the question?

@Rmano Bruno has some code but at present it's not distributed

@JosephWright Ok, thanks

@JosephWright while at it, do I need to use parens around \x in fpeval? Like -(\x) to take into account the possibility that \x is negative, or it is taken care of?

@Rmano From memory, you don't need extra brackets here

@JosephWright Hmmm, plot (\x,{\fpeval{-\x^2}}) is doing strange things ---

It's neither (-\x)^2 nor -(\x)^2. Tricky

@Rmano I wouldn't rely on --1 to do something sensible, better be explizit.

@Rmano it would work if \x were a real fp variable, but as it is a simple command it breaks:

\ExplSyntaxOn
\fp_set:Nn\l_tmpa_fp{-1}
\def\x{-1}
\fpeval{-\l_tmpa_fp^2}\quad
\fpeval{-\x^2}
\quad
\fp_set:Nn\l_tmpa_fp{2}
\def\x{2}
\fpeval{-\l_tmpa_fp^2}\quad
\fpeval{-\x^2}

\ExplSyntaxOff

@UlrikeFischer Thanks, now I see it. Will use parens when in doubt.

After writing tex in my OS CMD line I get:" This is mikTeX, Version 3.141592653 (MiKTeX 21.3) ". How do I exit back to the CMD line?

@JosephWright @UlrikeFischer while at it, I discovered that:

@bad_coder at the * type \bye

\tikzset{declare function={f(\t)=\fpeval{0.01-0.0000186012 *(\t-8.852) * (\t-6)* (\t+0.0767179)* (\t+2)* (\t+5.90209) *(\t*\t-12.349*\t+72.7683)} ;} }
\draw[thick,blue,samples=500,domain=-3:8] plot (\x,{f(\x)}) node[left] {$f(x)$};

works... is it cheating?

Compter avec des lettres, en français

@DavidCarlisle I almost answered escape ZZ (automatically triggered by the "how do I exit") ;-)

@Rmano tex is so more polite than vim latex is a bit more abrupt, \stop rather than \bye

@DavidCarlisle I wouldn't get along with such a numbering, I would have constantly to count ;-)

@UlrikeFischer Ok you can test the other one that is set up: davidcarlisle.github.io/learnlatex.github.io/mr/#toc

@UlrikeFischer getting rid of the if page.lang == mr tests in various places and adding a new entry on the general translations file:

# optional list of localised numbers 0-20 (fr alphabetic just for testing)
navigational-elements_digits:
  fr: "*,a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t"
  mr: "०,१,२,३,४,५,६,७,८,९,१०,११,१२,१३,१४,१५,१६,१७"

I just wanted to say ३ looks like a 3 and it is a 3 ;-)

@UlrikeFischer there is some shared history:-)

@UlrikeFischer yeah, and ५ looks like a four and it's ... a 5 ;-)

@UlrikeFischer @yo' you are clearly struggling so I made it easier for you: davidcarlisle.github.io/learnlatex.github.io/de/#toc

@DavidCarlisle that makes it much clearer, that lessons are arranged by rows and not by columns!

@UlrikeFischer indeed!

Is the online style special in biblatex? See tex.stackexchange.com/q/595478/3929, why is changing the formatting for author or title not affected for an online entry?

@daleif title works fine for me, but author is a name and so need \DeclareNameFormat.

@UlrikeFischer if I change \DeclareFieldFormat[online]{author}{#1\addcomma} to \DeclareNameFormat[online]{author}{#1\addcomma} in the OP example I seem to get family= outputteed to the bibliography

@daleif well sure, name formats are bit more complicated than simple fields, as they need to define the format using the various name parts, instead of #1 you should use commands like \namepartfamily, but if the OP only wants to change a comma, it would probably better to adjust a delimiter or one of the bib macros.

@UlrikeFischer ahh, just shows that I normally don't use biblatex

@daleif You need to patch the driver macro itself.

\usepackage{xpatch}
\DeclareFieldFormat[online]{title}{\mkbibquote{#1}\addcomma}
\xpatchbibdriver{online}{\adddot}{\addcomma}{}{}

@daleif @UlrikeFischer The OP should use \mkbibquote instead of \enquote.

@AlanMunn why (just curious)

I assume it's because that is the default macro for that in biblatex

@daleif It can handle weird American English punctuation rules.

@daleif So probably not relevant in this particular example, but good practice to get into. There may be other consequences I'm not aware of.

@JosephWright I just killed my aws account so latexcgi.xyz is no more (well actually it can be resuscitated during the next 90 days) It would start to cost real money from next month.

@DavidCarlisle I'm not sure if it is new, but if one compiles a document with hyperref (and the default pdfpagemode UseOutlines) in latex online, then one get a rather agressive sidebar:

My template uses some font other than Computer Modern. How can I use the latter for some specific parts?

E.g. I want the \mathcal symbols from Computer Modern.

@UlrikeFischer yes I noticed that the other day, I'm not sure I can do much about it, other than perhaps see if there have been any pdf.js updates

I am using pdflatex by the way (forced to).

\DeclareMathAlphabet\mathcal{OMS}{cmsy}{m}{n} @ComFreek

@DavidCarlisle you could pass [pdfpagemode=UseNone] to hyperref ;-)

@DavidCarlisle Wonderful! Can I make it local such that only some usages of mathcal have this effect?

E.g. I am looking at

 \newcommand{\LR}[2]{\mathcal{LR}(#1,#2)}

How can I use computer modern here?

@ComFreek easiest is to define it as \cmmathcal and use \cmmathcal where you want that

ah

wasn't aware of that I can change "mathcal" in your code to a name that fits me!

Thanks a lot!

@UlrikeFischer ooh mparhack

@PauloCereda Petra found an error ;-(

@UlrikeFischer oh no

@UlrikeFischer we could blame... uh oh

@DavidCarlisle ^^ help

@UlrikeFischer :(

@JosephWright I mixed up the licence ...

@UlrikeFischer Oops

@UlrikeFischer MIT? ;) (@DavidCarlisle)

@JosephWright GPL, but I copied a readme.md from another package with the standard LaTeX license.

@JosephWright did you ever implemented support for a "style" key with l3keys?

@UlrikeFischer .meta:n

@UlrikeFischer In tagpdf, if I want to mess everything up implement something creative by reassigning a bunch of nodes to a marked content sequence after they were created, do I have to do anything else beside reassigning their mc attributes?

@UlrikeFischer I refer the honourable lady to my earlier reply

@UlrikeFischer Like I said, you'd use \keys_define:nn { siunitx } { mystyle .meta:n = {color=blue,add-decimal-zero,minimum-integer-digits = 4} }

@UlrikeFischer We split define and set, contrasting with pgfkeys, so you have to use \keys_define:nn

@PauloCereda dinner

@JosephWright yes I know that I can define a key like this, but could a user define it with \sisetup{mystyle/.style={...}} or with some other sensible interface similar to the pgfkeys one?

R should really be spelt Argh.

\documentclass{article}
\usepackage{siunitx}
\ExplSyntaxOn
\keys_define:nn { siunitx }
  { create-style . code:n = { \__siunitx_style_create:nn #1 } }
\cs_new_protected:Npn \__siunitx_style_create:nn #1#2
  { \keys_define:nn { siunitx } { #1 .meta:n = {#2} } }
\ExplSyntaxOff
\sisetup{create-style = {mystyle}{color=blue,add-decimal-zero,minimum-integer-digits = 4}}
\begin{document}

\num[mystyle]{1}

\end{document}

@UlrikeFischer ^^ one approach

@UlrikeFischer Currently there is nothign built-in but it's doable

@UlrikeFischer Point of course is that pgfkeys is OOP from the get-go, so there is no difference between defining keys and setting them - we went for distinct steps so you need some form of indirection

@JosephWright looks quite nice.

@JosephWright I didn't doubt this. I only wondered if there is already some setup to copy. Imho it wouldn't be so good if one package defines "create-style" and the next "storestyle" and the third "new-style" ;-)

@UlrikeFischer Honestly I'd go for some separate interface to \keys_define:nn - it would be pretty easy to arrange

@UlrikeFischer Probably one for the team list or LaTeX-L

@UlrikeFischer Easiest would be \DefineStyle (generic) or \DefineSIStyle (dedicated), the latter taking two arguments: \DefineSIStyle{mystyle}{...}, which would then pass to \keys_define:nn { siunitx } { #1 .meta:n = {#2} }

\documentclass{article}
\usepackage{siunitx}
\ExplSyntaxOn
\cs_new_protected:Npn \DeclareSIStyle #1#2
  { \keys_define:nn { siunitx } { #1 .meta:n = {#2} } }
\ExplSyntaxOff
\DeclareSIStyle{mystyle}{color=blue,add-decimal-zero,minimum-integer-digits = 4}
\begin{document}

\num[mystyle]{1}

\end{document}

@UlrikeFischer ^^^

@JosephWright does one need something to avoid that existing styles are overwritten?

@UlrikeFischer see Boris's post to texhax? Shocking suggestion.

@UlrikeFischer I called it Declare for a reason ;)

@UlrikeFischer pgf doesn't - we could have New... and do a check first

@DavidCarlisle oh no

@DavidCarlisle reading the documentation first is clearly the wrong order ;-)

@UlrikeFischer The thing for siunitx is of course that I really want to encourage picking one load of settings in the preamble and sticking to them :)

@JosephWright well I have some statistic with integers and percentages, and they really need different settings.

@UlrikeFischer Fair enough

@UlrikeFischer Well, you of course know you can just use \keys_define:nn :) And of course the new version does have full code-level documentation :)

@JosephWright ;-). But actually I'm more pondering in general about such styles, e.g. for form fields.

@UlrikeFischer Indeed

@UlrikeFischer Like I said, it's not hard in general; if we provided a generic document/design 'set stuff' interface, we could have a matching 'provide style' one

@UlrikeFischer @DavidCarlisle FYI: The LuaLaTeX MathML conversion code found a new home on GitHub.

@MarcelKrüger oh thanks, I'll try to give it a better test

@DavidCarlisle Thanks, that would be great.

@MarcelKrüger oh, good.