One of tex's sillier built in constants finally parametrised!

didn't notice this before (but is already in tl) Umiddle allows you to specify \mathbin or \mathrel spacing

\documentclass{article}

\usepackage{unicode-math}

\begin{document}


$ \Uleft(\frac{x}{y}\Umiddle         | \frac{x}{y} \Uright)$

$ \Uleft(\frac{x}{y}\Umiddle class 2 | \frac{x}{y} \Uright)$

$ \Uleft(\frac{x}{y}\Umiddle class 3 | \frac{x}{y} \Uright)$

$ \Uleft(\frac{x}{y}\Umiddle class 5 | \frac{x}{y} \Uright)$

\end{document}

All night awake because the cold didn't let me breathe, the weekend at home with the flu... fortunately there is TEX.SE!

@DavidCarlisle I can get 1.0.4 once it's in W32TeX

@JosephWright 1.03 should be enough to pass that test I think, meanwhile....

Does anybody have experience concerning posters? I want to enhance the wikibook. Looking at extsizes, would it be better to use a regular A4 page and just scale it up?

@DavidCarlisle Indeed

@Johannes_B I think you need to distinguish "advertising posters" with big text designed to be read at a distance where scaling up is the thing to do from "scientific meeting posters" which don't use particularly big fonts and just stuff loads of information on a big page. extsizes.sty makes no sense really given scalable fonts it makes more "fixed" sizes available for metafont generation

@DavidCarlisle ok. Thanks

@DavidCarlisle I'm probably going to drop the 'stub' font loader in L3 format mode and simply use luaotfload. I'll probably need to rearrange a bit of the allocator but other than that its easy enough: reasonable plan?

@DavidCarlisle There's some requirement in luaotfload to do with allocating a count: very odd

@DavidCarlisle Oh, I see: you have to call the allocator count \e@alloc@attribute@count (and so on): I'll put some 'glue' code in for the present

@JosephWright yes I think so

@JosephWright "high level interface"

\texttt{  a^^^^0302b  !^^^^0302c  ^^a0^^^^0302d }

@JosephWright the ^ combining with ! ^^^^^

@Johannes_B have you looked inside any of my packages? :-)

@DavidCarlisle You are a special case. ;-)

@DavidCarlisle ugh. But one the other side \^{} is the wrong input and pandoc should correct it. Which leads me to the question why \textasciicircum is mapped to "02C6 instead of "005E?

@UlrikeFischer I wondered about that but I did what fontsoec was doing before (which I guess is what xunicode was doing)

@UlrikeFischer yes I think the thing to do in the default is add a U+00A0 if the base is empty. i don't think I can do much about this Sans code pro example at that level, \^{} is no worse than \^{!} there. Not sure about the textascii.... accents it wouldn't take too much to convince me to use the ascii slots

@PauloCereda friend of yours just joined?

@DavidCarlisle Living not far from Bielefeld.

And Hameln, too

@egreg it's all we need: another duck chemist.

@DavidCarlisle ooh

@DavidCarlisle then I'm trying to convice you ;-). Imho ascii should give ascii.

@CarLaTeX -- keep warm, and ... chicken soup! so many people have had such bad colds (and worse) this winter. take care of yourself.

@barbarabeeton yay, chicken soup!

@PauloCereda Duck soup

@egreg OH NO

@UlrikeFischer I blame @WillRobertson ^^^^^^^^

Hi guys o/

How can I find the caption manual and the manual for the rest of commands?

@DavidCarlisle Currently re-writing l3alloc: after lots of clever ideas, I think we'll end up with the 2e version :)

@Cardinal Open a terminal and type texdoc caption. hit enter

I am talking about this answer:

@Johannes_B Unfortunately, I'm a windows guy!

@DavidCarlisle Looks so. xunicode has

\DeclareUTFcharacter[\UTFencname]{x005E}{\textasciicircum}% see also x02C6
\DeclareUTFcharacter[\UTFencname]{x02C6}{\textcircumaccent}

@Cardinal Windows has a terminal as well.

Command Prompt?

@Cardinal Yes.

OK, let me check your code.

@Cardinal Alternatively, use texdoc.net. But it might not be the doc that matches the version on your computer.

Thanks it worked ^_^ @Johannes_B

@UlrikeFischer I have a slight worry that when I first made the latex/base version of tuenc.def I found i got more missing character problems with the ascii slots but that sounds wrong, perhaps I mis-remember

@Cardinal Nice to hear. Have fun. Reading a manual is better than googling.

@Johannes_B exactly. I will do that from now on.

@DavidCarlisle Even then I think the base configuration should be logical. I will be offline for a while. I have to install a new router.

@barbarabeeton Thank you, I've just had a cup of chicken soup!

@JosephWright how are you with fontforge? :-)

@DavidCarlisle I was thinking this is one for GUST!

@DavidCarlisle As TUG is in Poland this year there should be a chance to raise it ...

@JosephWright Frank's going to bachotex....

@DavidCarlisle Yup

@DavidCarlisle So I guess is Hans

@JosephWright any bets that the change won't break xetex?

@DavidCarlisle Well Hans' point is that newer OpenType math fonts don't have those values: as I said, this area is still in flux so stability simply can't be expected

@JosephWright sure I agree, fixing the font is best if it needs fixing.

@DavidCarlisle Meanwhile, l3alloc re-write is interesting: I'm going for a hybrid LaTeX2e-ConTeXT MkIV approach :)

@JosephWright what does context do?

@DavidCarlisle Start at 256 for everything ;)

@JosephWright i was hoping to finalise tu changes so one of us could do do PL2 this weekend but it's a can of worms. I think I'll take it to team mail.

@DavidCarlisle ConTeXt doesn't use the same 'allocator' registers as plain/LaTeX2e and hand-allocates scratch space in the sub-255 space, then \newcount and whatever start above the insert boundary

@JosephWright oh yes, I knew that once

@DavidCarlisle Yes, good plan

@DavidCarlisle I need to write some code comments then will check-in: all this to alter the font loading (which will be separate!)

@DavidCarlisle I wondered about this too. But xetex doesn't seem to have problem with a "correct" font like TeX Gyre Pagella Math so there is hope ...

@egreg @JosephWright @DavidCarlisle Is the optional argument in {s O{#3} m} the new stuff? c.f. tex.stackexchange.com/a/354502/37907

@Johannes_B Yes

@Johannes_B Latest xparse

@JosephWright Great feature. Thanks for that.

@Johannes_B Yes, very handy!

@Johannes_B thank Bruno:-)

@DavidCarlisle A big thanks to the L3 team and the people testing the code, giving useful feedback to the team.

@DavidCarlisle l3alloc re-write checked in: you can see how ConTeXt-like it (doesn't) look

@JosephWright OK finished my opus on the state of TU encoding, so I'll have a look...

@UlrikeFischer I'll give Heiko a chance to comment first (I don't think he tracks the github rep at all so i pinged him under that question) tex.stackexchange.com/questions/253401/…

@DavidCarlisle Thanks. There was a question on texwelt.de about it and I thought it would be good to record the issue. Any thoughts about the starting remark in answer about the allocation of the pdf resources?

@UlrikeFischer hard to add anything in the format just for latex in a way that'll work for pgf if that has to work in plain and context etc, @JosephWright can add a PDF resource allocator to l3alloc, for a clean start....

@UlrikeFischer did you see the |T| outcome. it's a feature.

@DavidCarlisle The answer from Hans? Yes. I wonder why no context users complain. Why is the space only wrong if one use \left/\right?

@UlrikeFischer I don't know, I'm not sure it's worth asking (you could ask, so it's not just me:-)

@DavidCarlisle On my latest L3 commit, perhaps longer-term we might adjust ltluatex so it can 'look for' some expl3 names in place of the 2e-style ones

@DavidCarlisle Here's a test

\documentclass{article}
\usepackage{unicode-math}

\newcommand{\test}[1]{%
  \setmathfont{#1}#1: $\left|T\right|$\medskip
}

\begin{document}
\test{Latin Modern Math}

\test{TeX Gyre Termes Math}

\test{Asana Math}

\test{XITS Math}

\test{STIX Two Math}

\end{document}

@JosephWright I think it would make sense to pull "e@alloc@attribute@count" etc out to variables which could be initialised as something like local attribute_count_name = attribute_count_name_ or "e@alloc@attribute@count" then an l3 version just needs to set them up before loading ltluatex?

@DavidCarlisle ^^^^ Note that only STIX Two seems to behave

@egreg so only stix 2

@egreg post that to the list see what Hans says?

@DavidCarlisle Please, do

@DavidCarlisle Done.

@egreg I meant you:-)

@DavidCarlisle I'd need to subscribe

@egreg @DavidCarlisle My results

I could post to the list, but not now -- have to go to the theater -- "Crime and Punishment", it will be quite depressing imho ;-(.

@egreg it would be good for you, be a bit modern drag you out of the 18th century.

@DavidCarlisle I saw your mail about TU. Sound all reasonable. And I wonder who to blame for the creation of \textasciicaron and \textasciibreve ;-)

@DavidCarlisle Something like that, yes

@UlrikeFischer Will I'm sure:-)

@DavidCarlisle We are not really too worried about this at the moment but hopefully one day, and it seems sensible just to load ltluatex.lua as that's all that's needed to get luatotfload to work

@JosephWright just do it, I'd say. Good practice not to bury magic strings in the middle of the code anyway.

@DavidCarlisle OK, I'll have a look when I get a moment

@DavidCarlisle Hopefully got it right: please look over checkin!

@JosephWright svn diff looks reasonable, I'll try it out later

Related to Double backslashes disappear from code, I've taken the liberty of identifying and correcting some of the posts that still have double-backslash corrections...

...limiting the edits to a handful per day.

This is the query I use to identify a number of these cases:

@Werner Good plan

@JosephWright From @m0sa it seems like "someone did something"...

But that "something" hasn't been identified. Moreover, I don't think it can be reversed, since there seems to be no consistency in the removal. Some threads have only some answers affected, while others remain in-tact.

So, it may be up to us to correct this purely due to the ad-hoc manner that it affected the posts.

Even if the cause is identified, I believe it cannot be reversed.

Knowing that the issue seems to precede Feb 22, 2015, I've been searching for dates earlier to that (changing the created clause to something like created:2010-01-01..2010-12-31 to search for possible affected posts in 2010).

@UlrikeFischer @egreg tug.org/pipermail/luatex/2017-February/006353.html

@JosephWright @egreg and @UlrikeFischer persuaded me to push back a bit on the |T| font thing:-)

@DavidCarlisle I think that Pagella is a “false positive”.

@egreg can't say I'm surprised. I should build a test with more letter combinations?

@Werner -- oops! when i updated that question with dates, i corrected all the q/a that still had the error. (i should have spread it out. apologies for overwhelming the "active" list.) anyhow, thanks for the search cue. i gather you haven't found any. thanks for undertaking the search for earlier ones. very distressing that the cause is still only a conjecture, and really nasty that it can't be undone.

@egreg $\left|T\kern\fontcharic\textfont0"1D447 \right|$ is I suppose what's wanted

@Werner -- align and matrix are also good targets to check. i'll take those on, and limit my updates to a few a day. (since the searches found several hundred potential items, this is going to take a while ...)

@barbarabeeton I know. I started with tabular, but other things will also be at risk. It'll definitely take a while. My search query provides at least means to identify the problem areas. Not everything will be caught, of course.

I have a wrapfig that is close to the bottom of page 1 and messes up page 2 as well. Is there an easy fix? The figure clearly fits well on page 1. I don't understand why it's affecting the next page

It's not like in this question where it simply doesn't fit

But it causes the same problem shown in that question

never mind, found the solution

@Szabolcs When you start fiddling with wrapfig you are on your own: it's up to you ensuring that the figure falls at a suitable position on the page.

@PauloCereda new ringtone? youtube.com/…

@Johannes_B Can you think of any possible reason for something like this? I can't.

@egreg No. It is absolute horror for me.

@Johannes_B I commented it makes no sense.

In general I dislike capitalized titles, but this is pure obscenity.

@egreg He seems to be german, so bad stuff is going to be wanted.

@Johannes_B It's even worse than letter spaced Fraktur!

@egreg You need that at some point. Though there are better ways to highlight stuff.

@Johannes_B -- how would you highlight something in a book that's typeset entirely in fraktur? (i observed the letterspacing convention in german books i had for texts in college, and it didn't really bother me. bold would have been too much, and italic simply doesn't exist. underlining would have really been out of place in the footnotes in "faust".)

@barbarabeeton Another Fraktur font, or ... a roman font.

@Johannes_B or slanted comic sans, in pink

@DavidCarlisle This is only suitable for geese :-)

@Johannes_B -- possibly another fraktur font, but i thnk a roman font wasn't even used in the publication info. i'll have to dig it out and look. (i have a nice little collection of insel books. they have always been eminently well designed. will be interesting to see what's done there. have to say, the books assigned in my german lit classes were always much more attractive than the engineering and math textbooks i left behind.)

Just for @egreg to admire

@DavidCarlisle Your next presentation for work should be in Comic Sans, just to see the reactions of the audience.

@egreg they wouldn't notice

@DavidCarlisle Try Schwabacher, then. Maybe some old Uncial.

@egreg or arabic

@DavidCarlisle Would they notice? :P

is the answer to E492: Not an editor command: Use a proper editor? (@PauloCereda)

@egreg I'm sure the brilliance of my presentation would mask any minor readability issues in the slides

@DavidCarlisle Wrong s ;-)

@Johannes_B if it was German, perhaps. That is "Old English MT" :-)