@Canageek thesis template meets units :-)

@AlanMunn -- Some interesting goings-on at Michigan State: nytimes.com/2021/05/11/science/…

@Canageek if \SI is being executed it can handle the argument and interpret it as it wants but if the caption is being written to the loit of figures, a robust command just writes itself unexpanded expecting its argumnet to do the same but you will have picked up the definition of \degree expacting \degree{Master of Science} or some such.

@Canageek You could try adding \MakeRobust{\degree} in the preamble.

I'm trying to find LuaTeX support for timing, which lead me to

Is there still no better way than trying to call PDFTeX macros from LuaTeX?

My mistake. It's not actually called the PDFTeX macros.

@FaheemMitha l3benchmark works with lualatex too and expl3 has suitable commands.

@UlrikeFischer Are those two alternatives?

I mean, two distinct alternatives.

It would be helpful if the l3benchmark documentation had an example. Searching the site now.

@FaheemMitha Usage is rather simple: Just use \benchmark:n{<Insert the TeX code you want to benchmark here>}

@MarcelKrüger but consider first what happens if this code is executed quite often ;-)

@MarcelKrüger There must be more to it than that.

What is n?

@FaheemMitha i's part of the command name, you have been here long enough to see commands with names \foo:nnn surely:-)

@FaheemMitha why must there be more than that?

@Canageek I think the change is that I removed some code that's redundant for siunitx itself, but your redefinition of \degree was 'relying' on it

@FaheemMitha Nope, Bruno's done the hard graft

@FaheemMitha vvv

\documentclass{article}


\usepackage{l3benchmark}
\ExplSyntaxOn
\newcommand\zz{\benchmark:n}
\ExplSyntaxOff

\begin{document}

\zz{
  \rlap{\begin{tabular}{cccccccccccccccccccccccccccc}
    &&&&&&&&&&&&&&&&&&&&&&&&&&&
  \end{tabular}}
}

\end{document}

@FaheemMitha terminal shows 4.36e-4 seconds (2.17e3 ops)

@DavidCarlisle We should probably just add this to the kernel (\zz is such a descriptive name)

@JosephWright I think so. More seriously a user callable name wouldn't hurt otherwise you end up like l3fo with a package just to give a 2e name.

@JosephWright we can't add \zz to the kernel, it would break half of David's answers here.

@DavidCarlisle We should probably deal with that too ....

@JosephWright yes, it would be nice if the commands were there by default.

@UlrikeFischer I could be helpful to the German readers and switch to \𝔷𝔷

@DavidCarlisle \blub ?

@barbarabeeton That's neat. I wasn't aware of it. But the beautiful arboretum on campus is named after Beal, the guy who started the experiment.

@DavidCarlisle do you remember - did we have the case to filter out html tags (or especially or only   in the text that goes to the ace editor?

@DavidCarlisle I noticed an issue with   after an update, nor sure if it was there before

@DavidCarlisle texfragen.de/einfaches_latex_dokument that dot comes from an   that the original lexer produced

@DavidCarlisle not sure if you had a fix in the -sk version, I used the original one now

@StefanKottwitz er that's odd it should be populating it with the text content so no markup, I can look later

@StefanKottwitz ah well that's different. so that isn't the html   it is (by then) the literal character U+00A0 I could strip it out but it isn't at all clear that it should be stripped out you can use A0 in text (it is the same as ~ basically once you have the utf8 input enabled which is the case here) so removing it in documents may break examples Having an A0 character in a latex preamble is an error.....

@DavidCarlisle ah, do you have a line for to strip it in a slightly patched version of runlatex.js for that site as a quick fix?

@StefanKottwitz er let me look..

@StefanKottwitz untested but after ` p[i].textContent=pretext.replace(/\s+$/,'');`

@StefanKottwitz add p[i].textContent=p[i].textContent.replace(/\u00A0/,' '); (if I got my JavaScript right)

@DavidCarlisle you got it right, thanks!

@DavidCarlisle I just added a g to have a /\u00A0/g

@DavidCarlisle hm, not sure.

@StefanKottwitz it's probably safe if I always strip out nbsp that are at the start of a line, they probably come from wiki/markdown spftware making emptyish paragraphs, theer is no sensible use for a non breaking space if it isn't being used like ~ in see table~\ref{zzz}

@UlrikeFischer as always it's got a danger of messing up packages, shame you can't easily isolate the top level

@DavidCarlisle yes, it is

@StefanKottwitz OK I'll try to add that later

@DavidCarlisle yes, that is the problem. It is also possible that it gets in definitions by copy and paste, and it wouldn't be good to try to guess if this is wanted or not.

@StefanKottwitz the font sizes are rather odd on this page. the ttfont is very tiny.

@StefanKottwitz I may just do that if it is in "guess the preamble mode" as "fix up the preamble" seems related.

@DavidCarlisle Thank you for the example. It seems you really like the letter z. Does the timing output look like 0.00145 seconds (4.42e3 ops)?

@FaheemMitha yes as I mentioned that's the output in that case

Oh yes, so you did.

I just came across theguardian.com/commentisfree/2021/may/12/…

which sounds quite horrific. I know there are some people from the UK here. So I wanted to ask if things are as bad as they sound.

Or perhaps it's been that way for a while?

But I didn't know the police no longer investigated scams/frauds. Are they all needed for crowd control?

@StefanKottwitz does this work for you? adding the if block before the existing replace()

	    if(runlatex.adddefaultpreamble) {
		pretext=pretext.replace(/^[ \t\u00A0]+$/gm,'');
	    }
	    p[i].textContent=pretext.replace(/\s+$/,'');

dinner

@DavidCarlisle ooh

@DavidCarlisle oh no

@DavidCarlisle works!

@StefanKottwitz OK I'll push that to git in a bit, thanks

when you hope no one accepts any of your answers...

@DavidCarlisle by the way, do you know why the pdfjs output may be blurry?

@StefanKottwitz it isn't the best pdf renderer ever but I don't notice it being particularly blurry?

Firefox:

Edge:

IE rendering, non-pdfjs:

does it look different of you add %!TeX pdf ?

differently odd :-)

FF non-pdfjs:

@DavidCarlisle you need an upvote

firefox is pdf.js anyway but might be a newer version

@UlrikeFischer someone will probably accept an answer and ruin it:-)

@DavidCarlisle I ask because there may be a known issue such as an undesired resizing of the PDF view after rendering

@StefanKottwitz I really do nothing there at all it's an iframe of pre-determined size and it shows the result of the post request, it just does what ot does...

@DavidCarlisle @DavidCarlisle Zooomin in doesn't make it much better, edge pdfjs:

@DavidCarlisle I wonder if the pdfjs output is a bit off because of a resizing to fit to the iframe, for example

@DavidCarlisle I'll check on another computer later, just bringing it up as you may have heard about it

@MarcelKrüger It seems l3benchmark is unhappy with wrapping makeatletter. I haven't investigated further. Time to head bedwards.

@StefanKottwitz not really specifically although there are plenty of reports generally that pdf.js has some quirqs and isn't always that faithful, but really it os the only option I see for cross platform rendering. If you enable the cooke settings options then you could choose to use your browsers default renderer instead (same as adding the TeX pdf comment)

@DavidCarlisle I'm on Windows now, that must be the fault :-)

When doing a \read\filehandle to \fileline, how can one tell from \fileline whether the line ends normally or with a %? Is there an easy test?

@StevenB.Segletes I think that is the wrong question:-) you could set \endlinechar=-1 then the end of line won't make a space, otherwise just test for a space at the end of the definition

@StevenB.Segletes the standart \typein command locally sets \endlinechar to avoid getting a space

@DavidCarlisle Is that the case if I want to process the end of line in the normal way? That is to say, if I read two successive lines, and append the 2nd to the first, how must \endlinechar be set so that a space gets inserted if line one ends in ^^M, but no space gets inserted if line 1 ends in %

@StevenB.Segletes that just happens on its own

@DavidCarlisle It ain't working for me...hmm I'll research some more. Thanks.

@StevenB.Segletes but what you do lose is the two consecutive newlines make a \par so you'd have to fake that

@DavidCarlisle I see my problem...the code I grabbed had set \catcode\endlinechar=9

@StevenB.Segletes ah

@DavidCarlisle Yup. That fixes it.

@DavidCarlisle Thanks.

@StevenB.Segletes commited runlatex.js