00:00 - 21:0021:00 - 00:00

12:05 AM
@egreg I do not check my reputation carefully but my impression is that I lost less than what the system claims.That is, I think you are right.

8 hours later…
7:36 AM
@JosephWright er If it was earlier today I was asleep, let me look...
@FaheemMitha sure but the problem is not well defined in that case. is a fraction one line or two?
@JosephWright the end document one?

7:49 AM
@DavidCarlisle Yup

8:35 AM
@DavidCarlisle One line. IMO.

@FaheemMitha and a six line matrix also one line? what about six displayed equations that get marked up as a array the same as a matrix, .... I'm interested to know what effects you were seeing that caused you to say that line numbering didn't work well for displayed math, it seems to me that so long as it doesn't give spurious tex errors more or less any arbitrary numbering is Ok.

@DavidCarlisle It's a long time ago. I don't remember the details.
I haven't tried doing it recently. Things might have changed.

@FaheemMitha I suspect not:-)

@FaheemMitha latex is open source, all behaviour is customisable

8:41 AM
Personally I'd go with the mathematical semantics I think. If something is a single equation or math expression, or part of an equation or math expression, it ought to count as one line.
@DavidCarlisle I mean, without extensive low level fiddling.
Not everyone is a TeX expert.

@FaheemMitha exactly but mathematical semantics is one thing that is not available to tex at all

@DavidCarlisle Agreed. So some way to override things locally would be helpful, imo.

@FaheemMitha the line number is just a latex counter so if a line numbering package counts a display as 1 line and you want to think of it as two you can always trivially \addtocounter{..}{1} to get the numbering back on track even if it does not display the numbers next to the equations

@DavidCarlisle Can you get it to display against some line it's not displaying on?
I guess it would depend on how it was set up...
I see there are a lot of questions about line numbering (relatively speaking), so people do do it.

@FaheemMitha there are 294 out of 175,570 questions tagged and most seem to have an answer so I don't see much evidence of major unsolved issues in that area.

8:51 AM
@DavidCarlisle I don't know whether there are or not. Like I said, it was a long time ago.

@FaheemMitha I was replying to your "I see there are a lot of questions" there are not many questions here on this.

@DavidCarlisle I just meant that people do number their pages using TeX. As opposed to using something else.
There are other ways of numbering pages of a TeX document than using TeX, but by definition they are even less intelligent.
To return to my question yesterday, per @JosephWright's answer, the \luastringN includes an \unexpanded... "token". Is that the right word? How about if you want it to be expanded? Do you just omit the \unexpanded?

@FaheemMitha yes as I said in chat here you can use \directlua(.... "#1") if you have control over #1 and know it expands to something that is valid content for a lua string
@FaheemMitha "number pages"? I thought we were talking of numbering lines?

@DavidCarlisle Ok. Is it still worth including the \luaescapestring, since (as I understand it) that helps with corner issues?
@DavidCarlisle Sorry, I meant numbering lines.
I was being imprecise.
Would \luaescapestring interfere with expansion?

@FaheemMitha it depends, possibly, possibly not. If you do then lua characters that might not be safe get replaced by things that might be safe, but in some cases you may want to handle that yourself
@FaheemMitha well it is expandable

9:01 AM
@DavidCarlisle To be clear, what I meant was, if I wrap the argument from TeX in \luaescapestring, will the argument still expand?

@FaheemMitha luatex does have a manual and luaescapestring is documented there as "Most often, this command is not actually the best way to deal with the differences between TeX and Lua.
"
@FaheemMitha yes as I said, it is expandable

@DavidCarlisle I read that section. I didn't really understand what it was talking about...
@DavidCarlisle The argument?
@DavidCarlisle So you'd recommend not using \luaescapestring by default? Handling on a case by case basis instead?

@FaheemMitha it means "Hans added the command as some people thought it might be useful, but he doesn't like it much" :-)

@DavidCarlisle Huh.
Still trying to understand it.

@FaheemMitha there are not many real cases where you want to pass arbitrary tex token lists to lua, so the situation does not come up much in practice I think.

9:05 AM
@DavidCarlisle Expansion, you mean?

@FaheemMitha no, luaescapestring

@DavidCarlisle Ah.
Well, passing \foo to Lua and expecting the value at the other end, as opposed to the string "\foo" is not so unreasonable, I think.

@FaheemMitha take the example in your question, that is concatenating two arbitrary tex token lists in lua and passing the result back to tex, so you need all kinds of quoting to make sure the lists are safe in tex and in a lua string, but you would never need to do that in real code, it is simply \def\foo#1#2{#1#2} so no lua involved.
@FaheemMitha that is unrelated to luaescapestring

@DavidCarlisle Yes I'm aware. I was talking about expandability.

@FaheemMitha expansion is the default so that's what you get unless you suppress it.

9:09 AM
@DavidCarlisle Yes, I'm not trying to write a routine for arbitrary anything.
@DavidCarlisle Yes, I understand. Sounds like a reasonable default to me, too.

@FaheemMitha except that edef-expansion is generally not safe for arbitrary latex text (it will destroy any accented characters for example)

@DavidCarlisle Oh. Hang on. Is edef-expansion the default kind of expansion?

@FaheemMitha it is the only kind of expansion

I don't know what that is. I do have one argument/field that might have accented characters (human names). So I could selectively disable expansion for that argument?

@FaheemMitha \unexpanded{#5} presumably

9:13 AM
@DavidCarlisle Ok. It sounds like you think luaescapestring is not worth bothering with.
I thought LuaTeX supported UTF-8. Shouldn't it be able to cope with accented characters?

@FaheemMitha as I said above luaescapestring is unrelated to this, it is addressing an entirely different issue.

@DavidCarlisle I realise. I'm speaking generally.

@FaheemMitha yes I was going to correct myself there accented characters will be OK

I mean, it's supposed to help with some corner cases, so the question is where it's worth keeping it around in case those corner cases make an appearance.
@DavidCarlisle So UTF-8 is ok?

@FaheemMitha luaescapestring is only an issue if you have single or double quotes or newlines or backslashes in the argument, no other characters are affected
@FaheemMitha yes, unless someone has used newunicodechar package to give characters an active meaning, or you have characters like ~ that are active by default.

9:17 AM
@DavidCarlisle Single or double quotes could happen. Newlines are very unlikely in my use case. Backslashes are also unlikely.
@DavidCarlisle Ok.

@FaheemMitha well take the example of double quotes, if you know that you are using \directlua{[[#1]]} then double quotes in #1 are not a problem so \luaescapestring won't help. Conversely ]] in #1 is a problem but luaescapestring does nothing about that. If on the other hand you are using \directlua{"#1"} and #1 might contain a double quote, \luescapestring is a convenience function to quote the " in the string.

@DavidCarlisle I'm mostly using [[, ]] for strings. It's a bit weird, but at least I know it won't be appearing in "normal" text.

@FaheemMitha then luaescapestring does nothing useful at all

@DavidCarlisle Single or double quotes might appear in the text, though.

@FaheemMitha As long as you are confident ...

9:23 AM
@FaheemMitha single and double quotes are allowed in [[ strings

@DavidCarlisle Did you see my team list mail about the open PRs? Really would appreciate your thoughts on github.com/latex3/latex3/pull/560

@JosephWright Would a single [ or ] be a problem?
@DavidCarlisle Hmm. Right.

@FaheemMitha only at the end.

@DavidCarlisle Ok.

'f you have [[#1]] and #1 is [abc] then you will not get [abc] as the string

9:25 AM
@DavidCarlisle Oh?

@DavidCarlisle I know Hans is not keen on \luaescapestring, but if you use either " or ' for the 'outer' string mark-up, it works as far as I know robustly

@FaheemMitha just basic lua syntax nothing to do with tex, you would have [[[abc]]] which is the string [abc followed by a trailing ] so most likely a lua syntax error.

@DavidCarlisle Yes, it says:
> return  [[[abc]]]
stdin:1: <eof> expected near ']'

@JosephWright yes, but usually I think you are passing numbers or known tokens to lua, arbitrary text runs normally better handled by callbacks so the need to quote unknown token lists is not that common I think

@DavidCarlisle I guess, certainly since they added all of the token-grabbing stuff

9:29 AM
@FaheemMitha i mean it's definitely an error if that is all it is, but as a fragment of a larger lua expression it coul dbe valid depending what else is going on.

Is there any reason to prefer [[]] to "", or vice versa?

@DavidCarlisle We perhaps should alter a few bits of expl3 to use token.scan_argument(): would speed some stuff up

To represent strings, I mean.

@FaheemMitha er well they have different quoting requirements...

@DavidCarlisle For normal text.

9:31 AM
@FaheemMitha not an answerable question. if you use [[ you can have multi-line text if you use " then it has to be a single lua line and you need to quote newlines as \n but which is more convenient depends on all kinds of factors, which is why lua provides both

@DavidCarlisle Ok. I'll stick with [[]].

@JosephWright possibly not, was out all day yesterday, let me look..
@JosephWright oh the box stuff, sorry I had a long standing action to look at that... I just looked over the PR and it all looks reasonable but not sure if I am going to have any tine today for a real review, I should go with what you have

@JosephWright I would be happy to help with that if you want that. Do not expect any huge improvements through, even luaescapestring is pretty fast.

@MarcelKrüger Any speed improvement is worth having
@MarcelKrüger At present, we'd have to have a version conditional, so may be not worth it: I've asked the team as a whole
@MarcelKrüger There's one place I'm really thinking:
\cs_new:Npn \@@_if_eq:nn #1#2 { \tex_strcmp:D {#1} {#2} }
\cs_if_exist:NT \tex_luatexversion:D
{
\cs_set_eq:NN \lua_escape:e \tex_luaescapestring:D
\cs_set_eq:NN \lua_now:e    \tex_directlua:D
\cs_set:Npn \@@_if_eq:nn #1#2
{
\lua_now:e
{
l3kernel.strcmp
(
" \@@_escape:n {#1} " ,
" \@@_escape:n {#2} "
)
}
}
\cs_new:Npn \@@_escape:n #1
{
\lua_escape:e

Lua now? :)

9:43 AM
@PauloCereda as opposed to lua_in_a_bit

@DavidCarlisle ooh semantics
\cs_and_now_for_something_completely_different:NN

@JosephWright You mean because scan_argument only came in 1.10 (or 1.09?)? Maybe you could just use scan_string, they should do the same stuff in this context. That should be there for quite some time already.

@MarcelKrüger From v0.95, it seems, so TL'16

10:10 AM
@JosephWright I just did some experiments and certainly the much bigger impact would come from using \luadef. This would require version tests, but it is faster by a factor of ~5. (See gist.github.com/zauguin/520b6dd6e4fb3cbc5776e0aed68f282e , I do not know how to post code here)

@MarcelKrüger That is wroth considering (@DavidCarlisle). I suspect we should do the same for all of the Lua-based stuff we need to emulate TeX primitives, so in \char_generate:nn, \file_get_timestamp:nN, etc. Would you be 'up' for making a PR?

@JosephWright Of course, I will prepare a PR.

10:32 AM
ooh a marmot, a koala and three ducks

@MarcelKrüger Cool: I could do it, but I'd prefer we had the 'history' of where it comes from

10:52 AM
Greetings! I have problem with installing getnonfreefonts in Texlive 2019 on Ubuntu 18.04 LTS
then sudo texlua install-getnonfreefonts
sudo getnonfreefonts --sys
and then
@JosephWright
@Joseph

@MYaseen208 I use vanllia TL, sorry, plus have never worried about the non-free fonts

Thanks @Joseph
@DavidCarlisle
@DavidCarlisle

@MYaseen208 you really shouldn't ping random people:-)

@DavidCarlisle ping

@MYaseen208 if there is a problem make a proper example, show what error message you get and contact the maintainers of getnonfreefonts

10:57 AM
Sorry for this

@DavidCarlisle, here is the question
2

I switched from Windows 7 to Ubuntu 14.04. The following code was producing this output. \documentclass[oneside,english,a4paper, 12pt]{book} \usepackage{a4wide, fullpage, color, fancyhdr, lastpage, changepage, multirow, calligra, graphicx} \usepackage[T1]{fontenc} \usepackage[utf8]{inputenc}...

Getting same error in Texlive 2019 on Ubuntu 18.04 LTS

@DavidCarlisle oh no

@MYaseen208 I don't have ubuntu nor have I tried installing getnonfreefonts so can't help, as i say if you think something has changed ask a new question referencing that or (better) ask the maintainers

5 hours later…
4:08 PM
Hello!
Is it possible to use \caption* without \usepackage{caption} inside a figure environment?

@manooooh \caption doesn't have a star form by default. So you need some redefinition somewhere.

@DavidCarlisle yes I mean that, we need to define something new
But if it is hard I can load the package...

@manooooh the entire definition of caption is only 23 lines, so you can copy it and add some definition for a star form (latex.ltx starting at line 6135)

@DavidCarlisle ok, I will try to do it, thanks David! Do you remember where I can find such file? (I'm on Windows)

@manooooh look in any log file, same folder as article.cls etc
@manooooh or the comment below
\def\caption{%
\ifx\@captype\@undefined
\@latex@error{\noexpand\caption outside float}\@ehd
\expandafter\@gobble
\else
\refstepcounter\@captype
\expandafter\@firstofone
\fi
{\@dblarg{\@caption\@captype}}%
}
\long\def\@caption#1[#2]#3{%
\par
{\protect\numberline{\csname the#1\endcsname}{\ignorespaces #2}}%
\begingroup
\@parboxrestore
\if@minipage
\@setminipage
\fi
\normalsize
\@makecaption{\csname fnum@#1\endcsname}{\ignorespaces #3}\par

4:21 PM
@DavidCarlisle found!
@JouleV are you The old JouleV?
@DavidCarlisle if a caption has >1 lines, would you center the text? Or would you let it in default mode (aligned to the left)?

Can someone tell me why the list of symbols does not appear if use compile the file using the following

@manooooh the default for multi-lined captions is justified not left aligned, certainly I wouldn't center the lines, just centre the parbox that contains justified text (which is what the above does)

@DavidCarlisle ok, thanks for the correction

I hate tex and latex
It is so unintuitive and it never works the first time
\clearpage
\printglossary[type=\acronymtype]
\printglossary

@nbro better to ask a question onsite rather than use an external link, but there is no information there about lists of symbols

4:33 PM
And I have this before the document
\usepackage{glossaries}

\makeglossaries

\newglossaryentry{latex}
{
name=latex,
description={Is a mark up language specially suited
for scientific documents}
}

\newacronym{jtvae}{JTVAE}{my description}

I have the following lines at the end of my document

@DavidCarlisle would you end a caption with a "."?

have you generated your glossary with makeindex or whatever?
@manooooh it depends

@DavidCarlisle Am I supposed to use makeindex?

@DavidCarlisle what are the different situations?

I just used the commands in the pastebin link that I linked you to above
pdflatex > bibtex > makeglossaries > pdflatex > pdflatex
I have read posts. People suggest to use this combination

4:39 PM
@DavidCarlisle I have a diagram of a simple graph, and I am specifing some important notes of the diagram. With this context I think I need to add a ".", right?

@nbro well you need to generate it with something, oh actually glosaaries does have a (not recommended) option to do all the sorting in tex but usually you would use makeindex, see texdoc glossaries page 15 or so
@manooooh if the caption is long enough to be set as a paragraph then I'd use . if it is a one or two word caption then it varies, I just searched all instances of \caption in the thousands of tex files I have used for examples here and some have . and some not, for no real reason
@nbro oh makeglossaries is a wrapper/alternative for makeindex that should work
@nbro impossible to debug disconnected fragments, make a small complete example and ask a question on site.

@DavidCarlisle ^^^^ I think it is long enough, so I think we should add "."

@manooooh I can't really help, in english these things are very variable and individual publishers have guidelines for what they want (which they probably just made up one day in the pub) I have no idea at all about the conventions in non-English documents

@DavidCarlisle oh, we don't have a convention, so I want to adjust to an English convention!

4:47 PM
@manooooh really that is not a figure (it is part of the surrounding list) so the text isn't a caption, it is just a part of the text for item (b) so it should be set as a normal paragraph and end with a . I'd say.

@DavidCarlisle ok, that makes sense

@nbro I'll leave that to Ulrike she probably knows more about that than me:-)
@manooooh our convention is to make things up as we go along.

@DavidCarlisle well, I don't like that, I prefer to be consistent, so I don't have an English convention right now! Hahaha jk

@manooooh sure, be consistent in any given publication, but there are very few global rules

@DavidCarlisle in LaTeX, when we are writing indented code and we need to insert a blank line i.e. an Enter, this new line which is empty has to be with indentation or not?
I think not
I think all blank lines do not have to be indented

4:55 PM
@manooooh all white space at start of a line is removed (except in verbatim) so you never have to indent the source.

@DavidCarlisle ? I mean in source code, I don't have automatic indentation:
Should we keep that blank space with indentation or should we delete its indentation?

@manooooh none of the white space there makes any difference, add it if you like or delete it, tex can not tell the difference

@DavidCarlisle I know...

@manooooh so what is your question? You are asking which of two identical options to pick, what answer do you want????

But I have read in a web page that we must delete all indentation in blank lines ¯\_(ツ)_/¯
@DavidCarlisle an analogous question: would you indent the code or not?
Altough the output is the same, you will answer "Yes of course"

5:00 PM
@manooooh you can probably find a webpage that says brexit is a good idea, don't believe everything you read on the web.

@DavidCarlisle so that's why I'm asking... I want to read a professional opinion

@manooooh look at latex.ltx, there is a lot of my code in there, look at the indentaion style....

@DavidCarlisle the code is indented

@manooooh I'm am amateur tex has never been part of any job I had.

@DavidCarlisle the experience in this site gives you the qualifier of "professional" in almost everything :)

5:03 PM
@manooooh ah it is more than it used to be:-)

@DavidCarlisle lol

5:27 PM
I could check on this myself, I guess, but are there scoping issues when writing Lua functions for use with LuaTeX that I should be aware of? I know that Lua encourages one to declare local variables so that they global namespace is not contaminated, but the downside that local variables may not then be visible elsewhere?
My LuaTeX question made HNQ for some reason.
Though I suspect not many people will be clicking on it.

@manooooh -- The decision can be based not on "is it long enough?", but on whether or not it is a sentence. Although I have trouble reading this in the exceedingly small type that shows up on my screen, it looks like a sentence. So yes, end it with a period.
@manooooh -- If I interpret this correctly, the blank line is purely for formatting, i.e., making it easier to distinguish the \nodes from the \draws. Thus, the extra vertical line indicating the indentation is meaningless; I would delete whatever caused it, in this case (probably) multiple spaces -- but only enough of them to remove that single vertical, not any of the ones to the left of it.

@barbarabeeton but, as @DavidCarlisle lighted me up, in this context we don't need to put it as caption but as normal text, because 1) it is inside an enumerate, 2) this diagram ends \item

@manooooh -- That's absolutely true, but the suggestion I gave holds also for a regular caption, so might be useful to you in the future.

@barbarabeeton yes, you are right! But sometimes inside enumerate, to make a new paragraph, we need to add Enter, so as it is indentated (because of \begin{enumerate}), we have generated a new blank space with indentation. So my question is, should we delete its indentation?
@barbarabeeton all you said actually is useful for me :D
@barbarabeeton we don't need to discuss whether a blank space is necessary or not, I am discusing whether when a blank space appears we add spaces to align it with the rest of coude

5:45 PM
@manooooh -- The example you showed looks more to me like a formatted algorithm than the output of enumerate. I'm not willing to extrapolate this to an enumerate environment. However, my instinct is to add whatever blank spaces may be necessary to give a result that shows the intended "level" of the line, and no more. In the example, there appear to be more spaces than necessary, resulting in a "visible level" indicator, which here is incorrect.
You will probably have to check the output to make sure the input is correct.

Hmm, this Lua function is behaving differently in Lua then when run from an Emacs buffer/script, which suggests there may indeed be a scoping issue. Time to try a simple TeX MWE.
Hey @barbarabeeton. How are you doing?

@barbarabeeton forget about the 2 or 3 spaces added, this problems appears when copying from Overleaf to TeXnicCenter. If you want, you can analyze this code:

@FaheemMitha -- Still suffering withdrawal symptoms from retirement. But, for a change, the sun is shining, and that's a nice plus.

@barbarabeeton I was meaning to ask what you are doing now. I had the impression you had stuff lined up. Though I forget what that stuff is. Isn't some conference in the offing?

@FaheemMitha In a buffer the scope will be just that 'run'

5:52 PM
@JosephWright I don't follow.

@manooooh -- That's pretty awful. Both of these are wrong, in my opinion. I'd repair the second one to add just enough spaces to line that extra vertical up with the second long vertical, and hope it aligns properly. Or, consider ending the previous line with \\[1\baselineskip]. Since I don't know what package is being used, it's impossible to know whether this is allowed.

@manooooh you always need a blank line to end a paragraph (whether not a paragraph should be indented should always be set as a general style never on an individual paragraph.

@FaheemMitha -- Well, TUG 2019 is happening in August, and I expect to attend (barring disasters). Unfortunately, I haven't yet found (in my myriad boxes) the notes I made before I retired with the intention of giving a talk, so ... this year no talk.

@DavidCarlisle yes. But if we are inside an environment then all the code is shifted to the right, isn't it?

@barbarabeeton You don't have it on a computer somewhere?
Paper is passe.

5:58 PM
@manooooh how can you worry about this when it makes no difference, aren't there enough things to worry about that do make a difference? :-)

@DavidCarlisle the problem is that I already worried about all the rest, now I just have to worry about being consistent :)

@barbarabeeton @manooooh is showing the latex file source and asking about indenting style there, not about typesetting. (or at least, I think that is the case:-)

@DavidCarlisle yes! My question is about style

@manooooh I would say it really is wrong to worry about things, that kind of thinking leads to complicated cut and paste rules that people end up following, without realising that none of the complication is needed. the rule is really simple, white space at start of line is ignored (it is not even reported to tex) anything else you say about that is making things more complicated for no good reason.

@FaheemMitha -- I have it on backup media, but the only computer I have access to is a really out-of-date macbook, and trying to read the particular media (created on a linux box) on that is more than 50% likely to reformat it so that it can't be read again later on another *nix box, which is what I want to acquire as soon as some serious hardware/software/connectivity questions are answered. (I don't want to be my own sysadmin; my husband has agreed, but his experience is all IBM mainframe.)

6:02 PM
@DavidCarlisle yes, there is a reason: I want to be consistent, so I spend time to learn about styles, norms.....

@barbarabeeton Sounds less than ideal. Won't the AMS let you use their systems?

Altough the output is the same, I want to be consistent in the source code!

@manooooh as I say, I think that is completely wrong, but it's your document, do as you wish.

@DavidCarlisle why is it completely wrong? Should not we be consistent?

@manooooh you should follow the design and the design here is that white space is ignored.

6:06 PM
@DavidCarlisle -- Okay. I'm not at all used to verticals of this sort showing in input, so I interpreted it (apparently incorrectly) as the output. Under the circumstances, @manooooh, you really don't need the indentation.

@barbarabeeton yes, the image is not an output but the internal code of my document. I know that I don't need the indentation, but it is useful for me or for any programmer that needs to read the source code. So, with this in mind, should we keep spaces in a blank line or not?
@DavidCarlisle I don't understand. I know the spaces are ignored, but what does it have to do with the fact that we need blank spaces to make a new paragraph?

@manooooh if you share authoring with another person using a different editor, you would find that different people indent by different amounts, but in any case people should just go with the flow and do what there editor does. It makes no sense to indent by hand or spend ages trying to customise different editors to match.

@FaheemMitha -- It's against AMS rules to let anyone not an employee use their systems. It actually took me an extra week to pack boxes, but the plug was pulled on my workstation as soon as I logged off on my last day. (Auditors. At least, that is the excuse they give.)

@manooooh you need a blank line, you never need spaces space characters at the start of a line are not reported to tex.
@barbarabeeton Here we introduced a system that people could be kept on a "retainer" (less than 100 pounds a year, so not enough to live on:-) but it allows us to give people who retire machine access/email etc without worrying the lawyers.

@barbarabeeton That's really not very friendly.
Anyway, sorry to hear that.

6:13 PM
@DavidCarlisle -- While that's true, some mail systems, when forwarding files, strip off final blanks and if there's nothing left on a line, may also remove the "blank line". This is undoubtedly OS dependent. Unfortunately, I've seen too many files where this happened to trust anything without careful examination.
@DavidCarlisle -- Your management is much more enlightened than that at AMS. (Things have gotten less and less "friendly", and more and more bureaucratic through the years. When I was first hired, the exdir was a "benevolent despot", accent on "benevolent".)
@FaheemMitha -- Thanks for the commiseration.

@barbarabeeton Are you planning to look for different work?
TeX consulting, perhaps?

@FaheemMitha -- I've already reached fossilhood. I shall look around for some nice volunteer opportunity, preferably one to which I don't have to commute a great distance. My TeX background (as @DavidCarlisle will tell you) is too grounded in history, and not up to date with current practices; AMS procedures are (for stability reasons) several years behind the times, and that was a very constrained system.

@FaheemMitha also, @barbarabeeton struggles with the English language:-)

@barbarabeeton It's possible you are, in common parlance, selling yourself short. But I don't enough about TeX to offer an informed comment. I don't even know viable TeX consulting is. My impression is that relevant parties think that both software and support should be free.

@FaheemMitha You are probably running 'interactively' in the buffer: in that case, each line forms a scope

6:23 PM
@JosephWright Do you mean the REPL? It was explained to me in #lua that each line in the REPL forms a scope.
But I'm seeing failures in LuaTeX.

@FaheemMitha Yes, exactly
@FaheemMitha Details?

@JosephWright Ok.

@FaheemMitha In LuaTeX, the Lua scope is the whole TeX run

REPL is Lispy terminology, but seems to be used elsewhere.
@JosephWright If I define a function as local, is that a problem with seeing it on the TeX side, for example?

@FaheemMitha -- "Selling myself short", maybe. But I haven't yet learned LaTeX3, have a very biased view regarding the stability of luatex, and aside from being able to guess in many situations where to look for an input error, may not be able to fix it expeditiously.

6:25 PM
@barbarabeeton It's possible to learn new things. What do you think is the problem with LuaTeX?
I was learning it this week, so I'm curious.

@DavidCarlisle -- No worse than your typing. (Or mine either, but at least I know how to proofread. Usually.)

Journal support for TeX is terrible. Some years I spent months going and forth trying to get help from the journal regarding a style file. Finally I came here (not sure why I didn't do so initially), and I got a diagnosis and a workaround in, as I recall, half an hour.
Bottom line, if you use TeX, you should be willing to support it. Morons.

@FaheemMitha -- It's my understanding that backward compatibility is not guaranteed. Coming from a background where a journal article may have to be resuscitated after 20+ years to be used in a "collected works" volume, that is anathema to me.

@barbarabeeton Yes, I actually has a question about this a little while ago. Probably here in chat.

@FaheemMitha Makes no odds for TeX, though if you pass stuff back to TeX remember that scoping is then TeX's rules

6:29 PM
@JosephWright I guess I'm missing something. Working on a MWE.

@FaheemMitha Right, looking forward to it

@FaheemMitha -- Well, the AMS does support TeX, though not as thoroughly as I pleaded for, for 20+ years. It took them more than ten years to admit publicly that they would no longer accept submissions in AMS-TeX.

@barbarabeeton Fair enough. I meant everyone else.
It's not like journals don't have money.
What package is \directlua defined in? luacode? Though I thought that was optional.

@FaheemMitha -- I'm learning to be careful in ascribing terms like "moron". In a certain current situation (which has nothing to do with TeX), I feel that using that term is insulting morons.

@barbarabeeton :-)
Never mind, I'm still running pdflatex on this file. Bugger.

6:36 PM
@FaheemMitha \directlua is the marker primitive of LuaTeX
@FaheemMitha Depends

@JosephWright Right. So does not require any package support.

@barbarabeeton I still find this so weird, particularly the idea that nothing changes in terms of the design ....
@FaheemMitha Exactly, it's the defining characteristic of LuaTeX

What is a "marker primitive".

@FaheemMitha When LuaTeX starts up, it's the only primitive it adds to TeX90

@JosephWright The only one? Ok.
Why wasn't it just called \lua?

6:39 PM
@FaheemMitha Initially: there are lots more, but they are activated from \directlua

@JosephWright -- As I'm fond of saying, math has a long shelf life. Not nearly so in other disciplines.

@FaheemMitha There's also \latelua

@JosephWright Ok.
@barbarabeeton LuaTeX (or something like it) strikes me as a good idea.

@barbarabeeton Yeah, but that just means you reprint the same text, not that you reprint on a new volume with no change (certainly with no change in design: that's rather the point of new editions)

But I agree that backward compatibility is a good thing.

6:40 PM
@barbarabeeton That's what second-hand book sales are for ;)

Though my understanding is that Lua is relatively stable. No big shakeups are planned.

@JosephWright -- Is that (\latelua) spelled correctly?

@barbarabeeton For example, amazon.co.uk/Chemistry-Elements-N-Greenwood/dp/0750633654 is pretty much never going to change, so I think it will stay in print 'for ever'
@barbarabeeton Yes, it's the one for shipout
@FaheemMitha Things are getting to be stable, yes

This looks like a nice overview - tug.org/TUGboat/tb30-2/tb95mahajan-luatex.pdf

@barbarabeeton Yup, late not latex or anything of the sort

6:42 PM
To teach LaTeX in 2019, do you like the video-on-YouTube way?

@barbarabeeton Something like that is badly needed so TeX can talk to all the other stuff out there.
My specific concern right now is databases, but there is lots of other stuff too.

@JosephWright -- It's considered an honor for a noted mathematician to have a volume of "collected works" published. It's also a mechanism to get more income, though typically not a lot. If most of these "works" are journal articles, a used bookstore isn't of much use. Having a stable input file cuts down enormously on the proofreading cost.

@barbarabeeton I see the point, I guess I wonder how people get away with it ;) Both in terms of copyright (multiple publishers ...), and in terms of style (book vs journal), it seems incredibly strange to me

@JosephWright -- Permissions can be a nightmare, but they're only rarely rejected, at least among the more "academic" publishers. I think that AMS may be getting more submissions because of the behavior of Elsevier et al. (Possibly the same with Annals of Math, Pacific Journal, other journals not yet farmed out to someplace commercial.) And books usually aren't included in "collected works" except by reference.

@barbarabeeton The Journal of the AMS is supposed to be super prestigious. Have you worked on it?
JAMS for short.

6:55 PM
@barbarabeeton I'm sure permission is achievable, like I say it's just out of my area. I meant more style: to me, the way I write journal articles and the way I'd write a book chapter are rather different (I've only done one book chapter, mind, and lots of papers).

@JosephWright -- AMS, at least, has very few monographs with different authors for each chapter; that's not particularly relevant in math. Articles for proceedings of seminars and symposia aren't particularly different from journal articles; again, perhaps a math idiosyncrasy.

@barbarabeeton I do see all of that, I guess I still find it odd that people expect original research to be re-printed, rather than summarised in textbooks by different authors (summing up the work of multiple people)

@FaheemMitha -- I wasn't involved in the design, but I did do an investigation for the editor at one time. (The editor also happened to be president of the society.) He wondered why some articles in a couple of issues looked so dreadful. Turns out that the authors of those articles didn't trust TeX, and inserted their own spacing. Over time, most authors have grown either more used to the TeX conventions, or have been browbeaten into using standard practices.
@JosephWright -- Mathematical theorems usually aren't easily summarized. Every step of a proof is essential. So it's often essential to examine the full original (even if it's well known), especially when trying to build on it or present an alternate proof.

7:15 PM
@barbarabeeton Sure, I get that, but I suppose it's all the rest of the text I'd expect to see altered, for example ordering between material, introduction, context, ...

@JosephWright -- If I remember correctly, it took Andrew Wiles four extensive articles, equivalent to an entire issue of the Annals, to prove Fermat, and I think there wasn't much "fluff" in the lot. But a lot of the practices in math publishing can probably be attributed to the established culture of mathematic[ian]s. It's a very different culture.

@barbarabeeton Sure, yes, I do get that: like I say, part of what I find odd is the desire to reprint at all except as facsimile (so from plates)
@barbarabeeton In any case, provided the tools work, I'm happy: with my team 'hat' on, I'm all for stability

@JosephWright -- Sometimes there are later corrections that need to be merged in. Sometimes the original format has a different page size (and that almost guarantees the necessity of a new proofreading). Provided the files are in good condition, it's usually easier to set up a comprehensive TeX run than to handle each item individually -- computers are (now) cheaper than people. AMS records the environment (engine, package/version number, etc.) and preserves that in the driver file. Ready to rerun.

@barbarabeeton Think you'd go mad with what happens in my area :)
@barbarabeeton For example, a colleague (and former boss) recently wrote amazon.co.uk/…. It's basically a second edition combining two earlier books ...

7:31 PM
@JosephWright -- Since I've already been accused of being a bit mad, it might be an effort to make the adjustment.

@barbarabeeton ... amazon.co.uk/… and amazon.co.uk/… ...
@barbarabeeton ... but he did not have the computer files for those (I think probably Mac Writer or similar originally, now in Word)

@JosephWright -- Well, the one reviewer gave the combined edition a higher rank than either of the previous separate books, so there must have been some other improvements. (Or a somewhat biased reviewer.) But it's my understanding that chemistry "moves faster" than math, so a full update was warranted. (Hope it wasn't just pro forma, like many textbook "updates".)

@barbarabeeton Oh, the re-write was definitely worth it. One review says not much has changed, but that's not really true. Also, the older 'primers' look inexpensive, but they add up (the whole series is nearly 100 books).
@barbarabeeton I'm biased, perhaps, as the content more-or-less aligns with the course I teach, oddly :) But I recommend it to my students

@JosephWright -- AMS does not return the files as published to an author, except in preparation for a contracted revised edition. So an author will be fortunate in that respect. (Although I'm sure that some authors will be appalled at many of the changes made in the course of production to their precious code.)

@barbarabeeton Totally get that, didn't expect they would :)
@barbarabeeton If you are doing a second edition, you probably 'put back' all the annoying things the publisher 'fixed' as a first step ;)
@barbarabeeton In the case of my colleague, the publisher managed to ruin all of the figures in the new edition, in the sense the font got messed up. He was not too pleased, as he'd set up exactly as they'd asked, drawn everything, then had to correct them all ...

7:44 PM
@JosephWright -- Maybe true. I was never really involved in the production of a revised edition; that's mostly in the hands of the production editors. I did, though, as author support, get occasional questions about what something meant, and was usually able to explain why it was done, so it didn't get undone unless a change in meaning was involved.

@barbarabeeton I'm basing on some of the 'fun' I've had with proofs ... some odd stuff happens in chemistry

@JosephWright -- Figures are one of the reasons that AMS still uses LaTeX --> dvips --> Acrobat distiller to produce the print version (pdflatex for the electronic, to get the correct links). That permits automated graphics checking, which includes a thorough font verification.

@barbarabeeton I'm of course thinking of a Word-based workflow here, so the figures were pasted into Word (Windows I think), and stored inside it using OLE (they were drawn in the mighty ChemDraw); I'm an unusual chemist, remember, in using LaTeX when I can

@barbarabeeton Oh, that was interesting. I thought people just used pdflatex.

@mickep -- The authors usually do use pdflatex. The two versions (print vs. electronic) are subjected to a page-by-page check to make sure that all line and page breaks are identical. And that is the kind of consistency and backward compatibility that I've been used to and insist on.

7:52 PM
@barbarabeeton i'm still surprised by that: I'd have thought someone in the serious PDF preflight business would have written a suitable tool
@barbarabeeton I (and the rest of the team) are well-aware that dvips is still important: @DavidCarlisle and @UlrikeFischer know I've done a lot of work recently on driver stuff and dvips

@JosephWright -- As I understand it, things have gotten better. But authors tend to scale figures, without taking into account the fact that font sizes can get too small and lines will become thinner. Any line less than about .75pt thick (especially if colored or not solid black) may not be clear (or even visible) when printed on an offset press. After a document is in pdf form, it's not always easy to tell what's part of a figure. As a colleague once put it, "it's all hamburger."
@JosephWright -- If you know of a super-good, super-reliable preflight tool that can isolate such problems, I'm sure the AMS production crew would love to learn about it.

@barbarabeeton Sure, but don't you have one 'hamburger' per figure? We always have to supply each graphic separately, including (usually) details of how we made it. I have fun saying 'pdfLaTeX + GhostScript' for mine :) (I always re-process PDFs)

@JosephWright -- No. Everything is run in-line. dvips just gives references to the separate figures, so they're isolated when they need to be. The article or book in pdf form with embedded figures is meatloaf.

@barbarabeeton Guess that comes from accepting LaTeX input ;) In Word, you really do need the figures separately 'just in case'

8:07 PM
@JosephWright -- Most of the time, tikz graphics are in-line. In case of problems, they may be extracted, run separately, and then included as graphics. But those aren't usually the ones that are scaled inappropriately. Separate .eps and .pdf figures are often submitted, and the scaling done in the tex file. So checking the separate figure without that information doesn't provide useful results.

@barbarabeeton Don't you just ignore scaling by the authors?
@barbarabeeton I'd (naively) imagine the first step of proper typesetting would be strip out any author scaling so the professional typesetter can actually see how they look

@JosephWright -- What if the graphic, as supplied, is ten inches wide? Without scaling, that can't be fit onto the printed page. No. The author's scaling can't be ignored. (Often it may be changed, but not ignored.)

@barbarabeeton You must do a lot better than we do: I know that figures should be either one column (8.8 cm) or two columns wide, but most authors don't. I imagine chemistry typesetters re-scale every single figure to make them actually the right size ...

@JosephWright -- I think that authors are getting the hang of things. Even if the scaling factor isn't exactly appropriate, the positioning in the incoming files is usually quite good at making the author's intent clear. And that is really important; avoiding changes in meaning is the primary goal of the copyeditors, and figure placement is often critical in that respect.

@barbarabeeton That can come out very interesting for chemistry

8:24 PM
@JosephWright -- No doubt true in any scientific/technical area. The copyeditors need to have at least a smattering of knowledge in the discipline, as well as be able to spell.

@barbarabeeton I mean that most papers are still done as manuscripts, with 'FIGURE 1 NEAR HERE' instructions ... which then get more-or-less ignored or which are almost impossible to follow due to figure size

@JosephWright -- Oh. Well, LaTeX does have the advantage there. Good luck, is all I can say.

@barbarabeeton You have been spared the pain of two-column Word templates and trying to place figures :)

@barbarabeeton I just got a question from a copyeditor who had noticed that copy&paste of an url in a bib broken at a hyphen doesn't work as the hypehn is lost - I wondered why someone even thought to check this ;-)

@UlrikeFischer Hopefully they are using one of the automated tools that are around nowadays for ref checking

8:32 PM
@JosephWright well but such tools would use the url in the annotation and not try to copy&paste imho (the url was fine, it is really only in the adobe reader when you select and copy).

@JosephWright -- TUGboat does get submissions with two-column figures, which sometimes have to be placed at the bottom of the first page. We've got a reasonable, if hackish, solution for that; a question about it exists somewhere on the main tex.sx site.

@barbarabeeton I wonder if you ever have to do one-and-a-half column floats :) I have at least one paper where that was the solution the typesetter picked
@UlrikeFischer Hmm, true

@UlrikeFischer -- Hmmm. Good crystal ball? I would assume that if the first line of a broken url ends with a hyphen, it's a real one. On the other hand, it's likely that Adobe reader doesn't know (or doesn't bother to check) that it's a url, so disappears the hyphen. I'll have to remember that.

@JosephWright it was a quite interesting question. I learned that the adobe reader doesn't have suitable tools to copy an url (with sumatry and the adobe pro you can use the context menu).

I've just learned that Adobe Reader has a limit on the number of files it will let you open ...

8:39 PM
@JosephWright -- In TUGboat, we'd treat then as full-width floats, and center the content within the float area. We don't set text to half-column width. (That's the easy way out.)

@barbarabeeton Oh, this wasn't how the source was: they were making sure we did not use too many pages :)

@barbarabeeton the adobe reader only allows to copy the url "as text", and so hyphen disappears. But using hard hyphens doesn't help: then the hyphen doesn't disappear but the link doesn't work as it is the wrong hyphen.

@JosephWright -- Neither TUGboat nor AMS journals have a strict page limit, only suggestions. So, mercifully, things like this are avoided.

@barbarabeeton Example on it's way: not 'our' fault at all

8:56 PM
@b
@barbarabeeton Talking parenthetically about DVI vs PDF workflows, looks like we are almost (finally) sorted with a robust way to break links in dvips :) (@DavidCarlisle, @UlrikeFischer, @PauloCereda)

I'm seeing \catcode 95\active in my aux file. I'm guessing this has something to do with the underscore package.

@FaheemMitha Yes, so it will read back properly

@JosephWright Ok. Thanks for the confirmation.

00:00 - 21:0021:00 - 00:00