@WillRobertson, @JosephWright see this unicode-math question...

Ok.

@Ho1 implementing bidi algorithm for example would mean you could that you would not need \rl here tex.stackexchange.com/questions/253507 But it would still be broken as the underlying typesetting is confused.

@DavidCarlisle The error caused by loading longtable doesn't show when book is used instead of scrbook. Let's blame Markus Kohm.

@DavidCarlisle I'd seen that: I assume @egreg has checked it's a macro/Lua level problem not an engine one

@JosephWright: you got mail.

@ChristianHupfer @clemens @ArthurReutenauer youtube.com/… That is the whole problem

@JosephWright Not sure it's a macro level issue: I see no change in luaotfload and unicode-math from 2014 to 2015 and I seem to remember some similar issue with TL 2011. However, I tried also with TL 2012/2013 and no problem showed up.

@Johannes_B GRRRRRRR

@daleif ? :-)

@egreg Might be an engine issue then

@Johannes_B Comment to that video. \Huge{text} !!! Argh

@daleif Oh no!

@daleif that is not as bad as promoting \\[.5cm] to be a paragraph.

@egreg Can we include longtable in the LaTeX kernel?

@PauloCereda no, it's mine get off.

@Johannes_B Das Kleineparagraph. :)

@DavidCarlisle Expect DuckTeX soon. :)

@PauloCereda :-)

@Johannes_B I have a poster on the door to my office of some frequent LaTeX errors. Item 0: any use of \\ or \newline in the text.

@Johannes_B Tried with TL 2011 and, lo and behold, no italic correction is inserted.

@daleif understandable :-)

@egreg what are you referring this to?

@egreg make a plain tex example and take to luatex list?....

@DavidCarlisle tug.org/pipermail/luatex/2011-October/003264.html and tug.org/pipermail/tex-live/2011-October/030295.html

@egreg ah so post saying that's come back?

@clemens youtube.com/channel/UCC-3dzj6dfbWwGzQzhkUS5A/…

@DavidCarlisle Well, not really: here's a Plain test

\input luaotfload.sty
\font\test={file:latinmodern-math.otf}

\newfam\testfam
\textfont\testfam=\test

\tracingoutput 1 \showboxbreadth \maxdimen \showboxdepth \maxdimen \tracingonline 1

\Umathcode`V="0 \testfam "1D449

$\fam\testfam VV$
\bye

with the log

...\mathon
...\test 𝑉
...\kern2.14
...\test 𝑉
...\kern2.14
...\mathoff

While if I try with unicode-math I get

....\mathon
....\l_fontspec_font 𝑉
....\l_fontspec_font 𝑉
....\mathoff

I guess Vafa has a new name?

@egreg So it is macro layer?

@JosephWright Seem so. But I couldn't spot changes in luaotfload or unicode-math.

@Johannes_B Aaah! :-( I stopped after “Zum einen, backslash backslash is der Zeilenumbruch” - which is 23 seconds in.

@egreg So some change in the binary that is only actived by code in unicode-math then?

@ArthurReutenauer :-(

Thanks for your time. I should go for now.

@Johannes_B have you found out who is behind those videos? Have you seen how he explains the difference between \small{<stuff>} and \small <stuff>? (#06)

Oh, you've already commented on that matter... :)

@JosephWright See what Ulrike says

@daleif My computer actually stopped playing the video at that point (OK, it was because I dropped my pen on the mouse pointer and it followed a link to that person’s channel).

@DavidCarlisle Yep, noticed that as well :-)

@egreg Ah, could be

@AlanMunn It has been a long time since the last 100K party.

@egreg Well @percusse and I are in a head to head very heated battle for the next one. Especially given how much we both care about reputation.

@egreg Which char has a catcode change in this range? They are all ASCII

@JosephWright I don't know. But the problem seems just that.

@JosephWright What's the catcode of U+1D449?

@egreg Letter, apparently

It is 11 outside of math, it is 12 in math (with the proposed workaround of doing \luatexcatcodetable\CatcodeTableLaTeX).

  1D449;MATHEMATICAL ITALIC CAPITAL V;Lu;0;L;<font> 0056;;;;N;;;;;

@egreg Looks like a letter to me

@JosephWright And the LuaTeX team seems adamant into not adding italic correction after letters (catcode 11) in math mode.

@egreg That's a bit of a problem: we are for a start looking at fixing \CatcodeTableLaTeX so it has the right catcodes

@JosephWright Perhaps it may be solved by adding some attribute to letters that we want to have the italic correction after them in math mode.

@egreg I guess I'll have to look into this in detail

@egreg Possibly

@JosephWright Or, temporarily, with a \CatcodeTableLaTeXMath

@egreg Indeed

@JosephWright See tug.org/pipermail/luatex/2011-October/003264.html for the discussion about italic correction after letters

@egreg At least it's not a mistake per se in the code we've got, more a problem with the font

@JosephWright No, it's a LuaTeX problem that completely disregards needs for math.

@JosephWright But it was 12 in texlive 2014. And the problem with disappears if one sets them to 12 \catcode\^^^^^1d449=12 \catcode\^^^^^1d447=12.

@ArthurReutenauer answer no apparently:-)

@AlanMunn very apt for our long bidi discussion above:-)

@Johannes_B -- did anybody notice the absolutely grotesquely unmatched quotes for \grqq at 0:51 of the video? bleccch!

@barbarabeeton ooh and rhymes with TeX. :)

@DavidCarlisle Ah OK

@barbarabeeton Yes, I did. It’s ugly - and also proof that he didn’t do his slides with TeX, which was already pretty obvious from their appearance.

@JosephWright Made a CW answer

@barbarabeeton What about the silly advice about the size changing commands? 1:18 mark.

@egreg: Juve bought Neto, nice acquisition.

@PauloCereda I agree.

@egreg -- right! whoever uses that syntax is gonna get a \Huge surprise! but @daleif notwithstanding ("Item 0: any use of \\ or \newline in the text."), i do use those for constructing tests, and they're occasionally used for extraordinary repairs when something is recalcitrant and the to-printer deadline has expired. (but i certainly don't recommend for "everyday" use, and certainly not to a newbie!)

@barbarabeeton The main difference is that you know what \\ does. The casual user doesn't. The other day, a student showed me a first draft of his thesis. Guess what? Every paragraph ended in \\. But he didn't follow my course. ;-)

@egreg: could you help me with my Italian keynote?

@egreg -- something that's on my list of "do this when nothing else is looming, and you're bored" is trying to edit the latex wikibook. in its present condition, it's so uneven that i refuse to recommend it to anyone. and that brings up a problem that i'm facing ... how do you say nicely to someone who's volunteering to help out with some tex stuff, but is so misinformed (it's too impolite to say "incompetent") that accepting the help would take more time than doing it oneself?

@PauloCereda Of course!

@barbarabeeton That sort of collaborative effort shows big limitations.

@barbarabeeton You'll write that paragraphs should only be ended by a blank line just to see that some time later somebody has edited the part saying that also \\ is good because their buddy told them so.

@egreg -- true, but i'm rather tired of seeing stuff sent in to ams that should get major surgery because the author wasn't "properly taught" (or even more depressing, didn't feel it necessary to read any documentation), and then have it glossed over by "editorial lite" so that what goes out the door is something that i certainly wouldn't be proud to put my name on. i know that some authors just won't read, but at least providing decent documentation is a reasonable goal.

@egreg -- if i see that, i will politely tell them why that shouldn't be done. and if i see a recommendation in those words submitted to tugboat, i will edit it out, or not publish it.

oh, i will also be polite and say why it shouldn't be done, and that they shouldn't ignore all those "underfull hboxes" caused by \\ .

@barbarabeeton We do what we can: promoting good documentation is surely something to do.

@barbarabeeton @egreg We’ve had an extended discussion about the topic earlier today - inconclusive of course, but there definitely is an issue here: how to improve the quality of the average TeX advice found on the Web?

@ArthurReutenauer -- i'd sure like to see it improved!!! and i'm willing to put in some effort. maybe this could be a topic for discussion in darmstadt? i think enough of the "interested people" will be there. (i've got a couple of other topics i'd like to discuss too, somewhat marginally related to this.)

@barbarabeeton Sure, happy to have that discussion.

@egreg exactly my point (about the \\, don't know why one cannot reply on the tablet interface). I've seen so many new students do that and I wonder where they have it from. I've made sure it is not even mentioned in my Danish latex book. I've sent students back to their office and fix those before I would help them with the rest if their setup

@daleif My glance at the student was self-explaining, I think. :)

@ArthurReutenauer So, you do speak german?

@barbarabeeton Very very good point.

@barbarabeeton Can you tape record it? I am very very interested here and unfortunately cannot attend.

@Johannes_B -- i'll try to arrange for a recording.

@barbarabeeton Nice, thank you :-)

@Johannes_B me neither :(

@barbarabeeton this won't be a team meeting, but nevertheless, how does stuff like that work?

@clemens Another time, i am sure.

@Johannes_B at least I should manage to go to a Dante meating some time :)

@Johannes_B ^^

@DavidCarlisle I have been told i look good in white.

@Johannes_B well we always need someone to sacrifice

@DavidCarlisle -- why, david, that's an odd place for a meeting of the ku klux klan!

@clemens I hope to get to a TUG meeting as well. So many interesting people to meet there.

@DavidCarlisle Last UK-TUG meeting?

@DavidCarlisle Ah, i rather stay in the backround then.

@barbarabeeton I carefully avoided that lot:-) the druids are somewhat older (and better at TeX)

@Johannes_B -- with a willing scheduler, sometimes there are "bof" ("birds of a feather") sessions arranged and assigned actual rooms to meet. that's not usually so easy in a hotel, more likely at a school. i've met with groups in rooms with tables, in the corner of an auditorium, in a willing participant's room, never (that i can remember) on a lawn under a tree.obviously, the more structured the setting, the easier to make a recording or keep notes. need to see what the options are at the site.

@barbarabeeton Out of interest, are you attending every DANTE meeting?

@Johannes_B -- i had a special invitation to last year's which was "dante 25". i hadn't attended one before that for a long time. way long ago, dante was kind enough to make me an honorary member, which was why i got the invite to the 25th. some years, a "local" group will sponsor a tug meeting at the same time and place as their own annual meeting or a eurotex, and then i'll attend if at all possible.

@barbarabeeton I hope to get to meet you some time. As well as other people around here. And people very active in german-speaking parts, like Markus and Elke.

@DavidCarlisle You should read manuals. ;-)

@egreg what a strange suggestion, I did my bit removing assorted pstricks and font packages to narrow down the cause;-) (I assume this is the ntheorem question?)

@DavidCarlisle Yes, of course. Adding amsmath to the options to ntheorem cures the problem as advertised in the manual. ;-)

@egreg a gentleman would have edited this fact into my answer but I see you try to steal my tick before I even have it:-)

@DavidCarlisle You took the wrong road… ;-)

@DavidCarlisle This one is easy enough for you

@clemens The \listofreactions idea sounds daft: you list schemes for this (if required: I find the whole 'list of X' business a waste of time)

@JosephWright agreed (which is why I'd like to drop it…). But at least a few people seem to actually use it

@clemens Bonkers

@clemens I mean, what does it say: 'Reaction 1: reaction one'?

@JosephWright yes :) Or even only `Reaction 1'

@clemens :-)

@clemens What a great listing that will be

@clemens Could be worse: 'Reaction 1: reaction 5'

@clemens I have seen 1 (in words: one).

@JosephWright lol

@Johannes_B the main site has three questions about such a list IIRC. (One of them being the reason I added that thing to chemmacros in the first place)

@clemens You could drop it in the doc, but actually leave the feature for now. Kind of an easter egg :-)

@Johannes_B Worth a thought

@egreg Despite what Taco says in 2011, I think \catcode11 is correct for these chars: you can't really have 'we will pick and choose which Unicode props to use for what', it's unsustainable

@JosephWright That's what I think, too. In any case, LuaTeX has still big problems with Unicode math (try \[\int f(x)\,dx\], for instance.

@egreg I also think @UlrikeFischer has it spot on in tug.org/pipermail/luatex/2011-October/003302.html

@egreg LuaTeX or LuaLaTeX?

@egreg Probably linked to my statement a couple of days ago about ConTeXt and math mode support: this is not such a big thing for them, so they have not been pushed to address these issues

@JosephWright LuaLaTeX, at least.

@egreg You see my point, I guess? Could be a macro-level issue

@JosephWright The missing italic correction isn't: it's due to the bad choice they made.

@egreg Indeed

@JosephWright so have they reverted the engine change intentionally, or is it a regression?

@JosephWright The bug with \int is long standing, AFAIR.

@DavidCarlisle Comes down to the catcode business: the set up in TL'14 was to have a separate file for LuaTeX from that for XeTeX, meaning that I suspect the LuaTeX one had been partly hand-edited to ignore some letters (according to Unicode definition)

@DavidCarlisle That's really not at all logical

@JosephWright oh but I thought taco said in that old thread that they removed the dependency on catcodes?

@DavidCarlisle I think he said the logic should change, but don't see a message saying it had

@DavidCarlisle OK, he did say (tug.org/pipermail/luatex/2011-October/003319.html)

> I am positive that this issue will be fixed in the engine one way or another before texlive 2012...

@JosephWright i should try myself byt what happens if you have a catcode 11 V with a mathcode that makes it use the 1d415 slot?

@DavidCarlisle Not tested

@JosephWright yes but if the fix was changing catcodes in a tex file, that isn't the engine...

using catcodes for this is just so obviously wrong.

@DavidCarlisle It's not clear to me whether the truncated luatex-unicode-letters.tex file deliberately missed out those problem chars or not: as far as I can tell, it was simply a hack from unicode-letters.tex to miss out XeTeX parts, but from some date ages in the past so ...

@DavidCarlisle Raise again on the LuaTeX list?

@JosephWright one of us should, or egreg or Ulrike or all of us...

@DavidCarlisle Usually it's you or me

@JosephWright I know, I thought they may appreciate a change..

@JosephWright stop it:-)

@DavidCarlisle :-)

@JosephWright Cry 'Havoc!', and let slip the dogs of war!

@JosephWright unicode-letters is your creation so i nominate you

@DavidCarlisle Seconded.

@DavidCarlisle <sigh> OK

Motion adjourned.

@JosephWright remember your old plan of "I don't do 2e stuff" ?

@DavidCarlisle Yes, still stands more or less: I'm not touching things that don't carry forward to L3. We need working Unicode support, and whilst the macros might get different names the ideas will be the same. So if the engines themselves don't work ...

@JosephWright Dinosaur! <3

@PauloCereda Grr

@Johannes_B Yep

@JosephWright meanwhile Karl wants me to resurrect symbols.tex from last century and rename it so it isn't the first hit for texdoc symbols ;-)

@ArthurReutenauer Interesting :-)

@DavidCarlisle Oh yeah, that was fun :-)

@DavidCarlisle On the LuaTeX/kernel stuff: I guess I'll do an update in light of comments then put on GitHub and open up?

@JosephWright yes please

@DavidCarlisle OK: probably tomorrow or Sunday

@DavidCarlisle Reminds me: need to ask about the whatsits business on the LuaTeX list

@JosephWright do you want me to kill my dpctex/luaregisternames github and you include something in a general ltluatex repro or keep potential kernel and potential package stuff separate?

@DavidCarlisle Let me see what I get done first!

@JosephWright yes sure, I just thought I'd throw that in as a possibility:-)

@DavidCarlisle I will probably try to put together some idea of an add-on support package that covers this stuff

@JosephWright I tried to checked the catcode of the math variables in context (not quite easy as you can't use the ^^^-notation directly, but imho it has no special code. The catcode are 12, and if one change them to 11 there is the same faulty correction. I also checked the bug tracker and there is a new bug report 0000933 (and an old one 0000171 looks interesting too)

@UlrikeFischer So nothing to do except wait for some kind of fix?

@JosephWright well the bug 933 is marked as major and assigned to Hans Hagen, and it looks like as it has the same cause. One could link to the tex.sx discussion and the older bug report 852 (mine from 2013) and so perhaps get somewhere.

@UlrikeFischer Perhaps best to raise with Hans in person at the TUG meeting?

@UlrikeFischer, @egreg, @DavidCarlisle Does look like the 'fix' before was simply to hack the catcode table: tug.org/svn/texlive?view=revision&revision=26623

At the very least that should apply to XeTeX too

I am starting to use Python to generate Latex directly from it. Since Python support multi-line raw strings, I could write the whole Latex code inside the Python script, and have access to Python variables on the fly. This makes it much easier to do more complicated things. Similar to using Lua from inside Latex.

I wrote small note to compare the two methods using a small example I found here on this site. Here is the link to the note 12000.org/my_notes/generating_latex/index.htm

One main problem, is that I can't use my TexStudio any more, since the Latex code now sits inside a Python script :) but on the other hand, I can make more complicated things more easily using Python+Latex

@JosephWright don't go that route, it will lead to endless arguments over whether odd edge case characters are letters or not

@DavidCarlisle From what I pick up in the TL checkin, they simply removed anything called MATHEMATICAL ... :-)

@Nasser why can't you write python in texstudio if that's what you want to do?

@JosephWright yes but as you note above sticking with Unicode classes is defendable, anything else is chaos

@DavidCarlisle I did not know I could do that. I thought TexStudio was for just Latex code? I use the Canopy Python editor now. But will look into it. thanks.

@DavidCarlisle Yes, but KB probably went for the pragmatic solution here: something sensible now for the end user

@DavidCarlisle I note @UlrikeFischer has checked ConTeXt and this suggests they've gone the same way

@Nasser never used it but python and latex are both just plain text, you can write them in anything

@JosephWright imho the kernel should keep the current catcodes, but unicode-math could reset them.

@UlrikeFischer Reasonable position

@WillRobertson See discussion above for full details on catcode/LuaTeX situation :-)

@JosephWright only just reasonable, what grounds are there to assume that only that block has the "wrong" catcode (and why does catcode affect italic correction)

@DavidCarlisle I mean 'gives the user at least some usable output at present with a known to be beta piece of software'

@DavidCarlisle that's why I suggested that unicode-math resets the catcodes (they are responsable). As a package it can easier and faster react to further development.

@JosephWright possibly but if you go that route it will be hard to undo it later, too many people will have managed to get ifx or other catcode sensitive code that relies on it. (although I suppose fixing people's documents for this year has some merit:(

@DavidCarlisle Yes, but do you have a better solution for people now? The engine business is complex and they are still defining what Unicode math behaviour should be

@JosephWright no, (unless taco tells you there is some secret switch to flip some internal behaviour somewhere, or perhaps more likely if you can get the effect from lua code without affecting the macro-visible catcodes)

@DavidCarlisle I doubt that many people check catcodes in this part of unicode. I can't even remember the exact number ;-),

@UlrikeFischer I spend my life in that block (they come from mathml, really:-)

@DavidCarlisle No callbacks in this area at present

@JosephWright that's what I suspect is the case, yes, but on luatex list (if i hadn't delegated it to you:-) I'd push hard for not changing catcodes away from UCD settings

@DavidCarlisle Fine, we leave the kernel alone but @UlrikeFischer is probably right that a package-level fix is sensible

@DavidCarlisle Remember of course that UCD says nothing about catcodes :-)

@JosephWright sure but it says something about letters and non-letters

@DavidCarlisle Certainly

@DavidCarlisle Doesn't mean it's right :-)

@JosephWright well no I think they should probably have not been letters in UCD but that's not the point:-) But how can it be right that italic correction for a catcode 12 v is different to a catcode 11 one (even if you don't consider the 1d400 block?)

@DavidCarlisle Certainly the engine behaviour is wrong

@JosephWright I doubt that will be quite the wording Hans/Taco use :-)

@DavidCarlisle Is Taco still working on the code?

@JosephWright no idea, seems to be mainly Luigi these days but in my mind it's those two:-)

@DavidCarlisle We could ask him

@barbarabeeton Do you remember the discussion on differentials? \newcommand{\diff}{\mathop{}\!{d_{}\kern-\scriptspace}} will avoid the italic correction to be inserted.

@DavidCarlisle but do you care about their catcode? Does it matter if it 12 or 11?

@UlrikeFischer it matters if it changes

@UlrikeFischer and why should italic correction in abc be different than the italic correction in abc from \meaning or \detokenize with catcode 12 characters?

@UlrikeFischer @JosephWright why don't I see the problem in plain tex:

\input{luaotfload.sty}
\font\lm="Latin Modern Math"

\textfont0=\lm
\textfont1=\lm

$V^*$

$\string V^*$

%\showthe\catcode"1D415

\catcode`\!=7


\catcode"1D415=11

$𝐕^*$

\catcode"1D415=11

$𝐕^*$

\catcode"1D415=9

$𝐕^*$

\bye

or @egreg :-) ^^

or the above with 1D449 and 𝑉

Ah sorry I do see it in plain now

\input{luaotfload.sty}
\font\lm="Latin Modern Math"

\textfont0=\lm
\textfont1=\lm

$V^*$

$\string V^*$

\catcode"1D449=11

$𝑉^*$

\catcode"1D449=12

$𝑉^*$

\catcode"1D449=9

$𝑉^*$

\bye

I just typed "latex vs. html" on google search, and I get page full of stuff about Mattress and sofas and sleeping. Google search is getting from bad to worst.

This is from a company that is supposed to be the expert on search engines.

@Nasser you can't blame google that latex has such a stupid name

@DavidCarlisle I just tried it again, and now it brings better results. May be google was in the middle of database backup or something. Very strange. sometimes I noticed I get different results from same search. It could depend on time of day or something.

or maybe someone who works at google saw my complaint above and fixed the bug in the google database.