TeX, LaTeX and Friends

8:25 AM

If you'd just agree to hold your meetings in here, you'd have PLENTY of time to figure things out before the deadline.

@UlrikeFischer at least that is a plan

8:46 AM

So, quick question: I'm trying to make a table with 65 columns, but latex seems to cut off the last one, only displaying 64. Is there anything obvious that might be happening there? (Like, for example, the designers of tabular believing a person would never try to make a table with that many columns?) Also, is there an easy work-around to get the last column to show up?

@Rithaniel There's no built-in limit, but the usual concern here is that the table simply won't fit on the page ...

@JosephWright Some of the table packages impose a limit, IIRC. This needs an MWE...

@DavidCarlisle ooh

@DavidCarlisle but not The Secret Plan

@JosephWright Yeah, this was an issue to begin with, but use of \resizebox helped get everything to fit

Well, this means that it might be something I've done to cause that last column to not display

9:06 AM

@Rithaniel As @yo' says, we need an example

I am reluctant to copy the whole thing, because 64 columns will make it a wall of text, but I'll copy the top bit:
\definecolor{Gray}{gray}{0.8}
\definecolor{Grey}{gray}{0.85}
\begin{document}
\pagenumbering{gobble}
\newcolumntype{g}{>{\columncolor{Grey}}c}
\begin{table}
\scriptsize
\begin{center}
\resizebox{550pt}{360pt}{
\begin{tabular}{ c|g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c g c }

3

New trailer :)

9:26 AM

@Rithaniel is it actually cut off, or simply off the page as the table is too wide?

@Rithaniel We need some form of demo, I'm afraid

@Rithaniel applying \resizebox to tables is evil, every time I see it I regret that we added the command to latex... (if you must do it, don't leave space tokens around teh table)

It's cut off. I've resized to multiple different widths, none of which work. (Should I link a pdf?)

Also, space tokens?

@JosephWright am I doing something silly or have the old test2e commands \PACKAGE, \AUTHOR etc been dropped from regression-test.tex ?

@Rithaniel no post an example here (or better post it as a question)

@PauloCereda I DIDN'T WANT TO KNOW THIS!!! AND I FEAR THAT IT WILL BE BAD ANYWAY :( :( :( :( :( :( :( :(

9:30 AM

\documentclass{standalone}
\begin{document}

\begin{tabular}{*{65}{c}}
  1 &  2 &  3 &  4 &  5 &  6 &  7 &  8 &  9 & 10 &
 11 & 12 & 13 & 14 & 15 & 16 & 17 & 18 & 19 & 20 &
 21 & 22 & 23 & 24 & 25 & 26 & 27 & 28 & 29 & 30 &
 31 & 32 & 33 & 34 & 35 & 36 & 37 & 38 & 39 & 40 &
 41 & 42 & 43 & 44 & 45 & 46 & 47 & 48 & 49 & 50 &
 51 & 52 & 53 & 54 & 55 & 56 & 57 & 58 & 59 & 60 &
 61 & 62 & 63 & 64 & 65 \\
\end{tabular}

\end{document}

@Skillmon OH NO

@PauloCereda sorry for my emotional outbreak.

@Skillmon /duck hug

@Rithaniel See my very small demo above

@JosephWright some rabbits seem to be a bit concerned about The Secret Plan... :)

9:31 AM

@PauloCereda I was just so very disappointed by those so called movies.

@Rithaniel you have \resizebox{550pt}{360pt}{ \begin{tabular}...\end{tabular} } which is forcing a space either side of the table, compare \fbox{a} and \fbox{ a } also really never use \resizebox while specifying both dimensions, that distorts the font with unequal scaling in each dimension.

@Skillmon ah the movie. :) I stopped at the original trilogy. :)

@PauloCereda I even liked the prequel trilogy, not as much as the originals, but they were enjoyable. The sequel trilogy is disappointing at least. I used to go into the cinema multiple times for every star wars movie released in my life span, not so for the new trilogy.

@Skillmon I personally don't like Natalie Portman. :) I watched the prequel, it was okay, I guess, but the Rey arc is confusing and enraging. And I cannot believe they killed Han!

Well, the overall goal of the table is to provide a cayley table for a group of 64 elements. Each cell of the table is just going to be an abstract symbol, so distorting the font isn't too much of an issue, truth be told.
So, is the little excerpt I posted above not enough to constitute a demo? Should I post the whole thing?
Also, if \resizebox is undesirable, what would be an alternative? I'm very much a latex amateur, so I'm just using what google searches and experimentation can lead me to.

9:38 AM

@PauloCereda Harrison Ford would only do it on condition they killed Hans: he felt the character was 'done'

@Rithaniel distorting fonts is always an issue, there is no point typesetting information if you then distort it so it's painful to read:-) If you provides an example, then someone would suggest reasonable markup but without that it is really hard to say. You can use 30000 characters in code sections posted in a question, I am sure you can make an example that is smaller than that.

@JosephWright it makes sense, but it's sad to see him go... :(

@PauloCereda watch the original trilogy again, then by magic, he's back

@DavidCarlisle ooh

\documentclass{article}
\usepackage{graphicx}
\begin{document}
\resizebox{\textwidth}{!}{%
\begin{tabular}{*{65}{c}}
  1 &  2 &  3 &  4 &  5 &  6 &  7 &  8 &  9 & 10 &
 11 & 12 & 13 & 14 & 15 & 16 & 17 & 18 & 19 & 20 &
 21 & 22 & 23 & 24 & 25 & 26 & 27 & 28 & 29 & 30 &
 31 & 32 & 33 & 34 & 35 & 36 & 37 & 38 & 39 & 40 &
 41 & 42 & 43 & 44 & 45 & 46 & 47 & 48 & 49 & 50 &
 51 & 52 & 53 & 54 & 55 & 56 & 57 & 58 & 59 & 60 &
 61 & 62 & 63 & 64 & 65 \\
\end{tabular}%
}

\end{document}

@Rithaniel ^^^ Example using resizing

@DavidCarlisle Yes, I tided some of them up

\documentclass{article}
\usepackage{graphicx}
\begin{document}
\noindent
\resizebox{\textwidth}{!}{%
\begin{tabular}{*{65}{c}}
  1 &  2 &  3 &  4 &  5 &  6 &  7 &  8 &  9 & 10 &
 11 & 12 & 13 & 14 & 15 & 16 & 17 & 18 & 19 & 20 &
 21 & 22 & 23 & 24 & 25 & 26 & 27 & 28 & 29 & 30 &
 31 & 32 & 33 & 34 & 35 & 36 & 37 & 38 & 39 & 40 &
 41 & 42 & 43 & 44 & 45 & 46 & 47 & 48 & 49 & 50 &
 51 & 52 & 53 & 54 & 55 & 56 & 57 & 58 & 59 & 60 &
 61 & 62 & 63 & 64 & 65 \\
\end{tabular}%
}

\end{document}

@Rithaniel ^^ Better one ...

9:42 AM

@DavidCarlisle watchitbackwards.com

See, I just replaced the height specification with ! so that it scales vertically the same as horizontally, and that has made it unreadable, in my opinion. Though, here is a truncated version of what I've put together.

\documentclass[10pt]{article}
\usepackage[margin=25pt,bottom=25pt,footskip=0pt]{geometry}
\usepackage{color}
\usepackage{graphicx}
\usepackage{mathtools}
\usepackage{amssymb}
\usepackage{amsmath}
\usepackage{comment}
\usepackage{array}
\usepackage{multirow}
\usepackage{tabu}
\usepackage[table]{xcolor}
\usepackage{cancel}
\definecolor{Gray}{gray}{0.8}
\definecolor{Grey}{gray}{0.85}
\begin{document}
\pagenumbering{gobble}
\newcolumntype{g}{>{\columncolor{Grey}}c}
\begin{table}
\scriptsize
\begin{center}

@JosephWright I blame @UlrikeFischer I was reading the manual. (section 2.1 ....)

> If you watch Star Wars backwards, it’s about a planet-constructing space station that builds itself instantly and immediately shoots lasers from an exhaust port.

> If you watch Star Wars backwards, it’s about this guy that kisses his sister and then cope with it by forgetting she’s his sister. Then he runs away to live with his Aunt and uncle in the desert.

> If you watch any Star Wars film, when Yoda speaks, he sounds like a normal being.

My favourite description is still Jaws:

> If you watch Jaws backwards, it’s about a shark that throws up so many people, that they have to open a beach.

@Rithaniel The final column shows up fine

@DavidCarlisle I was just considering if I can blame you for the wrong encoding I got with a biblatex file but I think it is the fault of biber ;-)

9:46 AM

Not for me, I've counted the columns. Perhaps it's my environment, then?

\documentclass[10pt]{article}
\usepackage[margin=25pt,bottom=25pt,footskip=0pt]{geometry}
\usepackage{color}
\usepackage{graphicx}
\usepackage{mathtools}
\usepackage{amssymb}
\usepackage{amsmath}
\usepackage{comment}
\usepackage{array}
\usepackage{multirow}
\usepackage{tabu}
\usepackage[table]{xcolor}
\usepackage{cancel}
\definecolor{Gray}{gray}{0.8}
\definecolor{Grey}{gray}{0.85}
\begin{document}
\pagenumbering{gobble}
\newcolumntype{g}{>{\columncolor{Grey}}c}
\begin{table}
\scriptsize
\centering

@Rithaniel ^^

(I'm using Texmaker, by the way)

@Rithaniel I just changed the entry in the last one so it showed up easily ...

@Rithaniel that's just the editor so has no effect on the typesetting

@DavidCarlisle What's the change?

9:48 AM

Wow, that looks really good, @DavidCarlisle I still have the last column getting cut off, though

@JosephWright oh just usual, I changed resizebox to use \small instead and used \centering not \begin{center}

@DavidCarlisle Ah, right

Yeah, it looks to be mainly the
\scriptsize
\centering
\small
\setlength\tabcolsep{2pt}
bit

@Rithaniel hang on I'll show an image, (as Joseph said, your test file should have something different in the last column, to make it easier to see, but as it is above you get:

@Rithaniel You've not used the final c column in the demo: if you add an extra entry, it shows up

9:54 AM

Okay, that appears to have worked. So the thing I was failing to do was to count the copy of c in any particular row of my actual code

Thank you guys for the help

(I thought it threw an error when the number of columns mismatched the number of columns specified in the tabular declaration. Though I suppose it's just if the number of columns is greater than that declatation)

Another package complaining about \IfFileExists. :)

@PauloCereda ?

@UlrikeFischer songs

@PauloCereda what is that package doing?

@UlrikeFischer Oh I have no idea. :) I just ran an old code of mine and it complained about that macro. :) I decided to use leadsheets instead. :)

10:15 AM

@PauloCereda hm, it is doing (shortened) \IfFileExists{ifpdf.sty}{\RequirePackage{ifpdf}\ifpdf \fi}{} and this explodes if ifpdf hasn't been loaded yet.

@UlrikeFischer Ah interesting!

@PhelypeOleinik As per @PhelypeOleinik's comment above, what is the proper method for supporting more that one release? Basically I'll update the current version of \InputIfFileExists but keep the old one for those strange users that does single target updates (e.g. users on overleaf that only updates their own copy of memoir)

\ifcsname InputIfFileExists \endcsname
  \DeclareRobustCommand \InputIfFileExists[2]{%
    \IfFileExists{#1}%
      {\expandafter\@swaptwoargs\expandafter
        {\@filef@und}{#2\@addtofilelist{#1}\@@input}}}
\else
  % Old definition
\fi

@daleif I think this one works good ^^^

@UlrikeFischer and the behaviour changes with the new definition?

10:31 AM

@DavidCarlisle \IfFileExists does:

@DavidCarlisle I don't think so, it fails also on tl18 for me.

\ifx\reserved@a\today
  \typeout{#1 found}#2\relax
\else
  \typeout{BAD: old file \reserved@a (should be \today)}%
  #3\relax
\fi

@DavidCarlisle So if #2 contains an unmatched \fi it will break...

@PhelypeOleinik erh, but InputIfFileExists is already defined!? It is the version from the kernel we are looking at.

@PhelypeOleinik as I pointed out re bidi patching the other day:-)

@daleif he was checking for the version with space at the end of the name (robust internal version)

@DavidCarlisle ohh, clever

10:34 AM

@daleif OOH

@DavidCarlisle ^^ it's catchy

@daleif Are you getting used to have \ExplSyntaxOn spacesignoredinthecode? ;-)

@PhelypeOleinik arguably a latex bug (we could re-arrange IfFileExists not to have #2 and #3 inside the if test...

@PhelypeOleinik but why does is fail only if the \RequirePackage{ifpdf} is executed? This here doesn't fail:

\documentclass{article}
\RequirePackage{ifpdf}
\IfFileExists{ifpdf.sty}{\RequirePackage{ifpdf}\ifpdf \fi}{}

\begin{document}
blbb
\end{document}

@UlrikeFischer \ifpdf \fi is not a mis-matched \fi in that case

@DavidCarlisle ah, \ifpdf is not a \if command yet ;-).

10:40 AM

@DavidCarlisle Forget my comment above. That's the temporary \IfFileExists defined for loading the format. It's redefined later.

@PhelypeOleinik I hardly ever work in expl3 code, still gives me a headache ;-)

@PhelypeOleinik yes but main definition has same issue

@daleif It's enjoyable, once you get used to its wordiness

@daleif Why do you need to patch \InputIfFileExists?

@PhelypeOleinik @JosephWright @UlrikeFischer something like:

\makeatletter
\long\def \IfFileExists@#1#2#3{%
  \openin\@inputcheck#1 %
  \ifeof\@inputcheck
    \ifx\input@path\@undefined
      \expandafter\expandafter\expandafter\@firstofthree
    \else
      \expandafter\expandafter\expandafter\@secondofthree
    \fi
  \else
    \closein\@inputcheck
    \edef\@filef@und{#1 }%
    \expandafter\@thirdofthree
  \fi
  {#3}{\@iffileonpath{#1}{#2}{#3}}{#2}%
}
\long\def\@firstofthree#1#2#3{#1}
\long\def\@secondofthree#1#2#3{#2}
\long\def\@thirdofthree#1#2#3{#3}

10:48 AM

@DavidCarlisle Looks good. No conditional dancing ;-)

@UlrikeFischer memoir adds some hooks similar to at begin/end package. I'd like to totally remove them, but that may break existing documents.

@daleif well you should perhaps make a suggestion how to add them in the kernel directly so that we can get rid of this filehook and similar packages.

@PhelypeOleinik but dare I define the xxxofthree commands or should I juggle things round a bit more. Has anyone in the last 30 years gone \newcommand\@firstofthree[1]{#1} in another world we could use \use_whatever:nnn ...

\long\def \IfFileExists@#1#2#3{%
  \openin\@inputcheck#1 %
  \ifeof\@inputcheck
    \ifx\input@path\@undefined
      \expandafter\expandafter\expandafter\@thirdofthree
    \else
      \expandafter\expandafter\expandafter\@firstofthree
    \fi
  \else
    \closein\@inputcheck
    \edef\@filef@und{#1 }%
    \expandafter\@secondofthree
  \fi
  {\@iffileonpath{#1}}{#2}{#3}%
}

@DavidCarlisle Perhaps this to avoid duplication at the end

@DavidCarlisle Or even remove #2 and #3 altogether

@DavidCarlisle Let me grep that. Hold on

@PhelypeOleinik getting rid of 2 and 3 sounds good, (oh I see the change you made)

@PhelypeOleinik but you have better have firstofone not firstofthree if you are not duplicating the arguments:-)

11:00 AM

@DavidCarlisle Oh yes. That's what I meant ;-)

@PauloCereda see what you have done ^^^^

@UlrikeFischer oh no

firstofthreehundredandtwelve

@PauloCereda don't worry, if we change that command it potentially changes every \input or \usepackage or \documentclass in every latex document. But if anything breaks we can blame you.

@DavidCarlisle oh no

@PhelypeOleinik ^^ see it is always good to have a backstop plan.....

11:08 AM

@DavidCarlisle can I at least have ice cream?

@PauloCereda no

@DavidCarlisle Seems good :P

@DavidCarlisle OH NO

/quacks in despair

@DavidCarlisle But we need ice cream and nutella pizza

3

@DavidCarlisle Except for \long vs. non-\long all but one definition of \@firstofthree and \@secondofthree seem reasonable.

@DavidCarlisle There's this one though: tex/latex/churchslavonic/cu-num.sty

\def\@firstofthree\cu@repeat@i #1\stop@ #2#3#4{#2}%

@PhelypeOleinik nice!

11:14 AM

@DavidCarlisle Exquisite. Funny thing is that the author uses his package's prefix for every internal macro, but this one :-)

I was just looking in textcase (removing the embedded copy from memoir), the code \expandafter\let\csname MakeUppercase \expandafter\endcsname \csname MakeTextUppercase \endcsname is that again because of the robustness? instead of \let\MakeUppercase\MakeTextUppercase`?

@PhelypeOleinik I may ask them to change that anyway (even if we don't add it to the format it breaks all the other packages that do define it)

@PhelypeOleinik YES

@PauloCereda I know how to make "Eistorte" - with selfmade vanille and rawsberry icecream and meringues in between.

@UlrikeFischer ooh <3

11:37 AM

Hmm, are there situations where one does not want \MakeTextUppercase instead of \MakeUppercase? I want to remove the embedded textcase from memoir. But I did not emulate textcase so users could just use \usepackage[overload]{textcase} in their preamble and everything is fine. If I move to \RequirePackage{textcase}, \usepackage[overload]{textcase} breaks with an option clash. I'm tempted to just require the overloaded version

@daleif sounds reasonable to me, I suppose you could have a memoir option not to use overload if that proves really necessary?

@daleif Use \PassOptionsToPackage{overload}{textcase}\RequirePackage{textcase} instead.

@daleif Or drop it altogether, typographical pretty upper casing can't really be done automatically, imho.

@Skillmon what is the benefit?

@DavidCarlisle magic!

@daleif no option clash if someone loads it earlier without it.

@Skillmon memoir has always used uppercase headers as the default (mimicking book), I always disabled it

11:45 AM

@daleif :) that one wasn't really serious, I know that you can't just drop it.

@Skillmon it only have one option and AFAIR \usepackage[option]{pkg} \usepackage{pkg} would not give an error, so I don't really see the benefit here

@DavidCarlisle I'll try to keep the original versions (perhaps in "undocumented" macros)

@daleif yes, that's correct, my bad :)

@daleif give a class option which version to use, change the default to textcase. Why move it to undocumented? That way backwards compatibility is better, imho.

@daleif well you are then changing the behaviour of memoir, so it is mainly the question if you think it will affect many documents.

sollniss

is it possible to disable jumping for references?

@Skillmon the manual is already cluttered enough

@UlrikeFischer one problem is that users do not read manuals, so even if this change is documented they go to stackexchange to complain

sollniss

11:53 AM

nvm \ref* does the job

@daleif even though users don't read manuals, you should add it to the manual.

2

@daleif yes, but then it is much better if one can answer "this change has been documented and you can revert it with this option" than "it is a secret change that only chat users know about" ;-)

@daleif there is a reason we write those. And if you write it in the documentation those of us who read documentation can answer, too, and not only you. ~50% of the answers on TeX.SX are from the documentation.

@UlrikeFischer I just don't want to add class options to handle package options

@Skillmon rewrote which?

@daleif that's not a class option that handles a package option, it is a class option that lets you select whether you want to use \MakeUppercase or \MakeTextUppercase (and in the latter case the textcase package is loaded).

@daleif "rewrote"?

@daleif if you mean "we wrote" it is meant as "we (the people publishing packages and classes) write documentation"

12:02 PM

@Skillmon no, we already use \MakeTextUppercase, it is a question of whether or not \MakeUppercase should be overwritten or not.

One could just save the originals (which is convoluted now they are robust) and add a macro that can restore the originals if needed.

@daleif ah, I misunderstood the initial question...

@daleif that's the same as adding a class option to control package options, isn't it?

@daleif well you could ask @DavidCarlisle to add an \undooverloadtextcase command to textcase (and a \undoundooverloadtextcase to reactivate them ;-)) . Imho that is a much better place for such a command than a class or some other package.

@Skillmon it does not clutter the class options

@daleif and adds a macro instead. I don't see the real advantage of this.

@Skillmon I do, in terms of documentation. I find it strange to have a class option saying: "Do you want to overwrite a macro or not"? IMO that does not belong in the class options.

12:12 PM

@daleif ok, your choice. I can only say that I'd add an option with appropriate documentation instead of a macro with appropriate documentation.

@DavidCarlisle I think you have a typo on elocalloc, it says Source File: textcase.dtx in the dtx and in the sty, as does ucharcat

@daleif oh almost as if I cut and pasted a random small dtx file to get started..... Thanks will fix...

@DavidCarlisle yasniplet can be your friend in Emacs

@daleif possibly although the number of times you start a new package from scratch isn't that great, and not always regular enough to be templated...

@DavidCarlisle So far I only use it for smaller things. But when I go on vacation and want to write a now for my office door: FERIE + TAB and it makes and tabs be through a template.

12:24 PM

@daleif fixed in github not sure I'll push to ctan (they would want me to up version number) but it will be fixed next time.

Interesting slow tv: they are moving a light house. About 12meters an hour. Stood on a dine cliff soon to be eaten by the sea. Now being moved 80-100m in land.

12:42 PM

@daleif (Just as I was going to reply, energy went down; sorry for the delay) Yes, \MakeUppercase (and \MakeLowercase) is defined with \DeclareRobustCommand, so the actual definition is in \MakeUppercase<space>. There are rumours of an upcoming \let for robust commands in the kernel :-)

@PhelypeOleinik I was just going to request that.

@daleif Nothing's done yet as we haven't decided what to support, but the implementation will probably be similar to the letltxmacro package. You can use that meanwhile

@PhelypeOleinik for now I stole the syntax from the end of textcase plus that \ifcsname + space test

@daleif That's good too :-)

12:58 PM

@PhelypeOleinik the main issue with it is that we have to do it twice. But for now that is probably what we have to do (new and old formats)

@daleif Twice?

@PhelypeOleinik github.com/slavonic/cu-tex/issues/61

@DavidCarlisle Let's hope someone sees that...

@MarcelKrüger in view of all the fonts for the harf-tests, I would suggest to create again a texmf in the repo and add a texmf.cnf to the support folder to use it.

@PhelypeOleinik that github organisation has had relatively recent activity

1:08 PM

@DavidCarlisle They were fast!

@PhelypeOleinik less than 20 years for responding to a bug report, they would not fit into the latex team at all...

4

@DavidCarlisle Any specific case you have in mind?

@PhelypeOleinik pick any from latex-project.org/cgi-bin/…

@DavidCarlisle Ooh. I didn't know that page. I think I was okay with not knowing it :-)

@DavidCarlisle I think we could close latex-project.org/cgi-bin/ltxbugs2html?pr=latex/4477. Resolved with change to TU ...

1:22 PM

@DavidCarlisle The important thing is: My name is not in the list ^^^

@PhelypeOleinik the people not in the list are the one who gets the blame ...

@UlrikeFischer Drat! ;-)

@PhelypeOleinik don't we have to test whether or not they are rebust and then use one set of macros if they are and a normal let construction if the robust test fails.

@daleif Ah, right. Although I don't think it's necessary for \Make(Upp|Low)ercase. I think these are robust since a long time ago

@PhelypeOleinik we could simplify that drop down menu and only have your name

1:31 PM

@DavidCarlisle Oh no!

My university had a short internet outage just now. First time since I'm here...

@UlrikeFischer done

@DavidCarlisle ;-) I assume that I can't change or comment there?

@UlrikeFischer email comments are public on that page if they include the original OP in the email header, or I could tell you the highly secret password so that you can log in and edit the status directly.

@DavidCarlisle why is then your closing comment not public? In the mail copy I got the OP seems to be in the header.

1:50 PM

@PhelypeOleinik I think your're right, ahh well no harm done

@UlrikeFischer I think if I had constructed an email to the OP, CC'ed gnats and added the magic gnats syntax into the email body to close the ticket, then the email would be public, but because I just closed it from the back end of gnats (which knows nothing about the web interface that we added) the mail isn't seen by the incoming mail filter. Gnats is @JosephWright's favourite bug tracker, that's why he stopped us switching to github issues for so long.

@DavidCarlisle :)

latex-project.org/cgi-bin/ltxbugs2html?pr=latex/2333 is really an amusing anomaly ;-)

@UlrikeFischer have you seen who submitted that?

@DavidCarlisle all hidden and confidential ;-( and I was so eager to see something funny.

1:56 PM

@daleif Seems to be so for at least 12 years: github.com/latex3/latex2e/blame/master/base/ltfinal.dtx#L927

@UlrikeFischer someone quite famous in the tex world

@UlrikeFischer Fine with me. I thought about doing that through l3build's texmfdir, but when that didn't work with fonts I gave up.

@UlrikeFischer you have mail

@MarcelKrüger I can do it, is the harfnode-dev tree the "main" currently? Then I would do it there and you can move the commits to the other trees.

@JosephWright could you try this example github.com/plk/biber/issues/292 on windows? I wonder if it is OS dependant.

@DavidCarlisle aha;-)

2:29 PM

@PhelypeOleinik lines saved ;-)

@PhelypeOleinik oh no

@UlrikeFischer we ducks are very smart :)

2:50 PM

@PhelypeOleinik the comparing feature in github is actually called blame??

@daleif it's actually a git feature:

GIT-BLAME(1)                      Git Manual                      GIT-BLAME(1)

NAME
       git-blame - Show what revision and author last modified each line of a
       file

@PauloCereda love it

@Skillmon <3

@daleif as it is in svn

I assume no svn user ever does svn praise

@UlrikeFischer It's currently the tree where most development is going on, but I think we should try to keep it on-topic, especially since the tests there don't run on Travis currently anyway. So maybe you could add it on dev? They will merge in the next day's anyway.

3:04 PM

The following might be a stupid question, but I never needed it myself and the mathematicians use it quite often: Is the following ok, or should I input it differently? \lVert {}\cdot{}\rVert

@Skillmon I would make the {}\cdot{} into a macro, so it is controllable from the preamble. Just in case you change your mind.

@Skillmon I'd do \lVert {\cdot}\rVert I think as that's more naturally three things, rather than 5

@MarcelKrüger I just made in the harfnode branch as there are the fonts. It works okay and it also find e.g. ,"LuaotfloadDummyFont.otf" (I added a small test file) but at least on my system tests prefer fonts in C:\windows if availabe. Is the init.lua need for anything if the fonts are in a texmf?

@DavidCarlisle seems more reasonable.

Having one symbol that means many things is bad IMO

3:05 PM

@daleif and yes, it's in a macro.

@DavidCarlisle I often find that to be too cramped, and go for the {}\cdot{} route.

@daleif as I said, I never used it myself, just reading a script from a maths professor who uses it quite often.

@daleif hang on let me look at the output....

@Skillmon I tend to use mathtools+etoolbox to get a \norm where, if given a blank arg just typesets \lVert{}\cdot{}\rVert

@DavidCarlisle No I think your're right, that would give the same output

@daleif that norm usage was just an example, he uses it in stuff like \sum_{i=1}^{n} \alpha_{i} k(x_{i}, {}\cdot{}), too.

3:08 PM

@UlrikeFischer ok. Yes, the init.lua is necessary because we use temporary files. They get random names by default and LuaTeX logs these names, so I added the init file to get predictable file names.

@MarcelKrüger ah ok. I let it in the folder - or is there a sensible place in the texmf structure?

@DavidCarlisle {}\cdot{} gives a bit more spacing, making the dot more prominent.

@Skillmon so does mine, I change it to \argm to keep the usage of \cdot to a single thing. Many people never give that an extra thought until you explain to them why it is not a good idea

@UlrikeFischer I don't think so. Also I think LuaTeX doesn't do a kpse search for it given that it could decide if kpse is loaded.

@Skillmon yes sorry I see what it's doing now. It's still logically dubious, ensuring that you get the mathbin spacing with two empty arguments but that's OK, this is tex, not mathematics:-)

3:12 PM

@DavidCarlisle would it make sense to also have immediate versions of the toc writing macros? One can get into trouble if you are trying to insert a line into the toc at a very specific line combined with using \include.

@MarcelKrüger ok, I push the commit. Lets look if travis likes it ;-)

@daleif they would be rather hard to document behaviour, if you put anything there with a page number the number may be wrong, and if you put something without a page number the timing of when it gets added to the file with respect to delayed non-immediate writes would be "interesting" . \include is actually an easier case as it has the forced page breaks so the timing is more controlled.

@DavidCarlisle right, in all the cases where I had issues we were using \addtocontents so page numbers were not really an issue.

@daleif yes but usually for a toc the thing you write with \addtocontents includes \thepage somewhere (except when it doesn't:-)

@UlrikeFischer You made Travis sad

AlexG

3:27 PM

A quick question to the TikZperts here, which are the available anchors of a text node in TikZ? I just cannot find a comprehensive list in the PGF manual

@AlexG doesn't that depend on the shape of the node? or what do you mean by anchor in this case?

@MarcelKrüger schluchz ;-(. The issue47-mac-only-font-family failed for me too locally, but I thought it is because I'm not a mac. I have to go now, and will look later.

@AlexG there are lists and pictures in the shape library section.

AlexG

@daleif Ok, for rectangular nodes with text inside.

@UlrikeFischer Thank you! Exactly this I was looking for.

@Skillmon twitter.com/SteveStuWill/status/1186653595518652416

@AlexG I probably have the names wrong: are we talking \node[anchor=...] or (A.north east) on a named node?

AlexG

3:37 PM

@daleif The former.

@PauloCereda "unducky valley"

@Skillmon ooh

3:55 PM

@AlexG that is an interesting one. I just now discovered that you can use any angle as an anchor. I can understand your confusion. So at least all angles plus what the shape defines?

Philosophical question: Why is an empty parbox empty? \fbox{\parbox{\textwidth}{}}, why isn't it \textwidth?

@daleif because that's what tex does.

@daleif the argument just sets \hsize, the length used for linebreaking, so if a line breaks then it makes an \hbox of that width, which forces the surrounding \vbox to be that wide. But if there is no text no \hbox is created so nothing sets the width of the vbox.

@DavidCarlisle fair enough, one reason why I tend to always add struts. Is going through some of my TODO list, one item was to explain some very strange behavior, and that was exactly it. And there were no struts in this particular case.

@daleif the danger with adding struts always is that if the parbox contains display material such as \[ or \begin{itemize} etc, then the strut ends up on its own in a spurious paragraph taking up vertical space.

4:10 PM

@DavidCarlisle this was headers. If users think it is a good idea to add displayed math or list to the headers... then that is their problems.

@daleif if you withhold secret information from me, then it's not my fault if my answer doesn't apply:-)

@DavidCarlisle ooh a secret

In memoir we automatically add struts to all the headers except to the empty one (for reasons). A user then added material to the empty header and got some very strange results. Proper method in memoir is to make an actual pagestyle (with a different name) and then overwrite empty with it. Just wanted to find the source of the strange results.

@daleif tabular p columns try to be clever adding the struts with \everypar but it's a tricky area and it doesn't always do the "expected" thing, and when it goes wrong it's hard for the user to see what happened or to debug. so "strange results" are sadly rather expected in this area...

su.figshare.com/articles/Writing_science_with_LaTeX/3205753/3

4:20 PM

@JosephWright ew Overleaf

5:13 PM

@PauloCereda Don't you like Overleaf? Even if they gave you their green duck?

@CarLaTeX not the idea, naturally. :) It's amazing what they did. What really grinds my gears is the fact that now the number of people writing bad TeX code is growing almost exponentially... :)

@PauloCereda You can console yourself with Nutella: milanotoday.it/eventi/casa-nutella-biscotti-gratis.html

@CarLaTeX OOOOOOH

@CarLaTeX I could find some pizza dough and have a Nutella pizza which it is good. :)

@PhelypeOleinik ^^

barbara beeton

5:45 PM

@JosephWright -- This is behaving very strangely. After about three seconds, the readable text disappears and a grayed-out "source code" takes its place. Is this intended, or is it my laptop just too aged to show it properly?

@barbarabeeton I'd recommend downloading the PDF: I find the web view ... odd

@JosephWright you chose.... wisely

barbara beeton

6:03 PM

@JosephWright -- Better, thanks. The first "page" of the pdf (it's hard to distinguish pages) still looks like the transmogrified browser view. But it looks to be worth reading. (Side note: Stockholm University math department is (still) a TUG institutional member, continuing from when the late Roswitha Graham was on the TUG board. Support much appreciated.)

6:19 PM

@PauloCereda Nutella is good and pizza is good. Nutella pizza is not a pizza.

@PauloCereda *¬*

@CarLaTeX But it's good nonetheless ;-)

@PhelypeOleinik I don't doubt it :)

@CarLaTeX :-)

6:43 PM

@DavidCarlisle latex-project.org/cgi-bin/ltxbugs2html?pr=latex/3869 can be closed too.

@JosephWright Is there a way to tell l3build that a test should use both logfile and PDF based testing?

@MarcelKrüger Er, no: we originally tried that, but it seemed tests would need to be different. I'd use an approach like the driver tests: two thin wrapper files, one 'core' file

@JosephWright Ok, thanks.

@UlrikeFischer done thanks

What's the difference of putting a package in, say, texmf-dist/tex/latex or texmf-dist/tex/generic (other than organisation, of course) when said package works in both Plain and LaTeX?

6:50 PM

@PhelypeOleinik Not everything searches both, they are checked in a fixed order

@JosephWright Hm... And if a package has a package.sty and a package.tex, both of which load package-code.tex, should all of them go to the same place or the .sty goes to tex/latex and the rest to tex/generic?

@PhelypeOleinik Er, that's complicated!

@JosephWright Is there any guide for me to read?

@PhelypeOleinik /usr/local/texlive/2019/texmf-dist/web2c/texmf.cnf

@PhelypeOleinik I mean that today there are only really three formats, and they all search basically all the places (ConTeXt II was different ...)

6:54 PM

@PhelypeOleinik for example context searches generic but not latex, some formats search generic before latex, some search latex before generic

@JosephWright @DavidCarlisle But as long as there are no duplicate names, the precedence shouldn't matter, right?

@DavidCarlisle Er, only MkII I think

@PhelypeOleinik well context doesn't _explicitly) search latex but it seaches everything at the end so...

@DavidCarlisle ConTeXt loads expl3-generic from tex/latex

@PhelypeOleinik Like I said, MkII doesn't ...

6:56 PM

@PhelypeOleinik eg uptex has TEXINPUTS.uptex = $TEXMFDOTDIR;$TEXMF/tex/{uptex,ptex,plain,generic,latex,}//

context has TEXINPUTS.context = $TEXMFDOTDIR;$TEXMF/tex/{context,plain,generic,}// so no latex but the trailing comma means it searches the whole tree.

@DavidCarlisle @JosephWright Oh, that's a good start. Thank you both!

7:16 PM

@JosephWright on git policy, If I'm going to make those callbacks data not simple, do I branch and push to both develop and master (develop is ahead of master currently) and if I branch and merge back without pushing the branch to github, can you really tell I did that, and didn't edit master directly?

@DavidCarlisle As a hotfix?

@JosephWright yes, essentially not necessarily a patch release just for that but it would be good to fix it sooner rather than later as you can't patch it in-document as we don't expose the lua table

@DavidCarlisle Well that is a hotfix ;) You are right that as we don't go through formal review, etc., you could commit direct to master and it would be the same

@JosephWright it would be good to have a reliable way of testing those callback classifications but it's hard I think, especially the more obscure ones

@DavidCarlisle Yes, I know: it's all a bit fuzzy

7:22 PM

@JosephWright the last change I made I did make a branch and merged it everywhere then deleted the branch, but as I never pushed the branch I realised afterwards that apart from a commit comment along the lines of "merge from zz branch" the branch never existed as far as the outside world was concerned.

@DavidCarlisle Yes, all true: the workflows tend to be written around everything being formally reviewed ...

@JosephWright I'm planning to make page_objnum_provider and process_pdf_image_content data not simple

@DavidCarlisle Cool

@JosephWright whi needs reviw? shory you trust me not to mak teh typos

3

7:34 PM

@DavidCarlisle ooh

8:05 PM

@DavidCarlisle latex-project.org/cgi-bin/ltxbugs2html?pr=latex/3388 can probably be closed too - there are enough newer bug reports ;-) Does one say that pathes with spaces don't work in luatex currently?

overleaf.figshare.com/articles/… - done with LuaTeX

2

@UlrikeFischer I think I'll leave that one while the current issues with quoting remain, that reminds me I was going to send a team mail about that....

@JosephWright interesting, you'd think most cern people would have a local tex installed, but perhaps it's the collaborative editing they need?

8:25 PM

@DavidCarlisle Yes: it's all huge groups of people I think

@JosephWright oh you mean like when when I first was given a spec for revtex and they wanted the frontmatter to cope with several hundred \author commands ? :-)

@DavidCarlisle Something like that

@JosephWright in mathematics a 4 or 5 author paper has "a lot of authors" but Physics seems a bit different,

@DavidCarlisle :)

@DavidCarlisle For chemistry, 4 or 5 is roughly 'normal'

9:36 PM

@DavidCarlisle latex-project.org/cgi-bin/ltxbugs2html?pr=latex/3797 has been resolved at some time, \multicolumn is \long.

@UlrikeFischer actually you should be able to understand the password to get into gnats (based on some curious space-free script:-)

@DavidCarlisle and this one latex-project.org/cgi-bin/ltxbugs2html?pr=latex/2635 ;-)

@UlrikeFischer I wonder if Jörg's mainz email address from 1997 still works?

@UlrikeFischer gnats won't let me close it :(

9:55 PM

@DavidCarlisle perhaps you can at least suspend it ;-)

@UlrikeFischer no I get a permissions error saving it so I can't change the state at all, I'll try a different browser or I suppose we have to channel into Rainer's subconscious memory of how this is supposed to work....

@DavidCarlisle what is different for this than for the other?

@DavidCarlisle It's supposed to work?

@UlrikeFischer no idea but it's gone in now (I changed responsibility from team to me, as it was complaining about having no change reason for "responsible" even though it hadn't changed)

8 hours ago, by David Carlisle

@UlrikeFischer I think if I had constructed an email to the OP, CC'ed gnats and added the magic gnats syntax into the email body to close the ticket, then the email would be public, but because I just closed it from the back end of gnats (which knows nothing about the web interface that we added) the mail isn't seen by the incoming mail filter. Gnats is @JosephWright's favourite bug tracker, that's why he stopped us switching to github issues for so long.

@DavidCarlisle this one looks resolved too: latex-project.org/cgi-bin/ltxbugs2html?pr=latex/4497. At least the definition is different now and I get a clear error from the example.

10:06 PM

@JosephWright I think @UlrikeFischer has just volunteered to be told the gnats password, don't you?

@UlrikeFischer On a mission?

@DavidCarlisle Yes: I can't actually remember it ... would have to ask FMi

@JosephWright let's say it would be easier for @UlrikeFischer to remember it (or me as I am naturally fluent in these things:)

@DavidCarlisle I have some vague idea, but then I love Gnats so much ...

@JosephWright not really, only thought to reduce them a bit. But some of them are so old and curious that is quite difficult to know what to do about them. But this e.g. can be closed too. Whatever is asked for time has passed over it: latex-project.org/cgi-bin/ltxbugs2html?pr=latex/2284

@UlrikeFischer Most of the old ones should I suspect be gentlely put down

@PauloCereda GitKraken has 'move commit up/down' :)

10:18 PM

@JosephWright yes, perhaps another "handle bug session" at some time. It would be good to know the real open bugs in this lot. @DavidCarlisle this is resolved too latex-project.org/cgi-bin/ltxbugs2html?pr=latex/3021

@UlrikeFischer do you want to test you new found information?

@DavidCarlisle I took a quick look, but when I click on edit there are lots of fields ;-(. It is too late now to decide which one I should edit.

@UlrikeFischer status has a drop down that you can set to close, and the "reason for change" you can write zzzz (if you can find that on your keyboard) or any coherent reason, probably not worth touching the other fields unless as happened with one of the above the form returns a status error when you try to save.

@DavidCarlisle Any idea what 'Sponsorship Tiers' are for GitHub?

10:35 PM

@JosephWright as I say in the first day I had dozens of messages from people discussing what tiers to offer. I just unwatched the repo so I stopped getting messages other than official status changes. As far as I understand it it's like crowd funding, you offer different perks for different levels of sponsorship, the more they pay the more help the sponsor gets,

... so if they offer you 1000 dollars a month , you offer free on-site support anywhere in the world, and if they make a one-off payment of 1 dollar, you don't. the idea of just saying "if you like the software, donations would be appreciated" doesn't really fit this model, I suspect

@JosephWright help.github.com/en/github/…

11:30 PM

@PauloCereda facebook.com/overleaf.editor/photos/a.1516210008631298/…

11:42 PM

@yo' I don't suppose "dress them in soy sauce and Chinese pancakes" is the answer you look for?

@DavidCarlisle you can try it :)

ooh ... cooked rubber ducks ...

@yo' oh! after all these years, now you tell me the ducks aren't real?