TeX, LaTeX and Friends

8:16 AM

I see the spammers are about again

@JosephWright Spammers?

@eiterorm Yes: they post very long 'questions' that get flagged within seconds

@eiterorm You probably won't be able to see the questions

@eiterorm Once they get enough flags they disappear from the front page, but I can still see them

@JosephWright Ah, ok. =)

Brent.Longborough

8:51 AM

@egreg I think ffarwel is a bit more permanent... More like addio than ciao

9:24 AM

Good morning all. I'm almost fine. Thanks for caring. :)

@Brent.Longborough Did English copy farewell from Welsh ffarwel or conversely?

@egreg :-)

@egreg Yay, glad to hear you are close to 100%. :)

11:14 AM

I've broken LaTeX again. =(

A: Alignment in Siunitx with Comma separators

5

You are using \num inside an S column: that's not how it works. The column contents are parsed 'directly': \documentclass{article} \usepackage{siunitx} \sisetup{ input-decimal-markers = .,input-ignore = {,},table-number-alignment = right, group-separator={,}, group-four-digits = true } \b...

Hmm, I wonder what I did wrong here (for the downvote)

@JosephWright probably you are supposed to make any possible input work, rather than saying the input is wrong

@egreg are you sure we did?

@DavidCarlisle Not you, quite certainly. :P

@DavidCarlisle Anyway, I'm rep capped nonetheless.

@eiterorm You've probably just broken it for you When Joseph says this, he meant for everyone :-)

May 12 at 10:48, by Joseph Wright

@DavidCarlisle I see you broke LaTeX again!

@DavidCarlisle What an honour. :)

11:24 AM

@DavidCarlisle Of course I meant for me. It even works for everyone else. =(

Apparently, the TL .rpm stuff for Fedora is sync'ed with the 2014 dev branch.

Question: I can suggest edits to posts by editing them instead of flagging them. I saw one post I wanted to add a tag to, but I don't think it needs any other edits. (Not by me anyway.) What should I do? Flag it instead?

11:38 AM

@eiterorm think it counts as an edit

@DavidCarlisle I could edit it, but adding a tag isn't enough. I also have to change at least 6 characters in the question, which I think is unnecessary.

@eiterorm ah Ok I forgot about that (after a while you can just change tags without asking:-)

@eiterorm do you not get an edit tags menu if you hover over the tags?

@DavidCarlisle No, I just get the tag details.

I was thinking of adding a {biblatex} tag to this question: tex.stackexchange.com/q/179267/3465

11:55 AM

@eiterorm it's got it already:-)

@DavidCarlisle That's why I pasted it here. ;-]

@DavidCarlisle It does work: I was just pointing out that siunitx ignores separators in the input entirely

@DavidCarlisle :-)

Brent.Longborough

2:02 PM

@egreg All that SE-Karma is a burden; should should offer a few bounties! :-P

@egreg Sounds Germanic, so I'd guess the English got there first.

Hi, what's the internal command makeindex uses to print an index element? Including the line spacing etc.

2:18 PM

@1010011010 It uses \@idxitem; in the theindex environment, LaTeX does \let\item\@idxitem

@egreg Thanks. Strange though, considering I already tried that . . .

Gonzalo Medina

@egreg I thought about you when I saw the definition of the commands in tex.stackexchange.com/q/179301/3954 ;)

@GonzaloMedina WOW!

2:45 PM

@PauloCereda Re: options to Lua scripts (TL mailing list). What's the difficulty: why is arg[1] (etc.) not enough?

@JosephWright I think you want --help or --repository=foo arguments in any order

@JosephWright Do you think that the “multiple \par” question should be tagged tex-core? ;-)

3:13 PM

@egreg I thought output routines were too strange so I gave a center environment answer:-)

@egreg Err, yes

@DavidCarlisle Not that tricky, surely?

@JosephWright for any particular case no (which is why taco indicated he writes it from scratch each time) but unix scripts traditionally call on getopts to sort it out and offer a named argument interface. otherwise you end up wondering if --repository = foo is one arg or 3 in the argv array so I gather from the conversation that's there is a lua library that may or may not be there that does same

@JosephWright :)

@JosephWright Depends on how complicated you want your code to be. :)

@JosephWright it's like asking why use l3keys when you could split up the list yourself with custom macros

3:24 PM

@DavidCarlisle You got a mention in my email. :)

@PauloCereda I saw

@DavidCarlisle A good mention, of course. :)

@PauloCereda At least you didn't accuse me of cheating

This is cheating! — egreg 15 mins ago

Q: Why does potassium react the most vigorously with water?

3:44 PM

@JosephWright This is for you

2

Link to exam Which metal reacts most vigorously with water? (A) Ca (B) K (C) Mg (D) Na The given answer is B and the reasoning is that is is the most electropositive since it the furthest left and down according to periodic trends. However, isn't $\ce{K}$ the conjugate acid from a st...

I would have answered B even knowing just a little chemistry. ;-)

What should I do when my MWE, which doesn't provide the intended result on my end, seemingly works perfectly for others? Both systems appear to be equally up to date.

@eiterorm cry

@DavidCarlisle Are there any less obvious things I might try? =P

@eiterorm usual response is "have you run latex enough times"

@DavidCarlisle Several times. No errors, warnings, undefined references, bad boxes or anything else. I've even turned the computer off and on again (because I went to sleep). =P

3:59 PM

@eiterorm let me try your mwe

@DavidCarlisle Ok, latest question. It will probably work perfectly for you as well. =P

@eiterorm running biber as its first time Ive used it (with tl2014) it'll probably take an hour doing whatever it does the first time.

@DavidCarlisle It appears to be a system-specific problem, though. Two people have already commented, saying the MWE works for them.

what on earth does biber do on its first run (still going....)

@DavidCarlisle It was rather quick for me, even the first time.

4:11 PM

Okay to my understanding the process of printing the index is an automated cycle (correct me if im wrong) of <determine first letter of item> <if first letter not in index yet (true) --> print heading_prefix<letter> print heading_suffix>, <if false --> print nothing> <place item> <print delimiter e.g. delim_0> <print page number><next item> ... In case you're wondering, yes I want to change that cycle

@1010011010 it isn't the most customisable program in the world (hence xindy aims to be a replacement) but usually you can fool makeindex to doing what you want by wrapping the items it is sorting with "inventive" macros so they print something perhaps not obviously related to the sort keys all depends (and would need a example doc in a question:-)

@eiterorm you get to the stage thinking should I kill the process....

@DavidCarlisle For me, even the first run could be measured in seconds. I think you can kill the process.

@DavidCarlisle I think I'll run into trouble trying to put the index in longtable format .... I'll take a look at xindy but the manual does not look welcoming :-(

4:27 PM

@eiterorm yes but then I have to find out where it was trying to write its cache and clean up:-)

@1010011010 longtable should be easy enough, what do you want to align?

@DavidCarlisle Well, I sure hope I didn't break your system as well. =P

4:52 PM

Hmm... should pdflatex overwrite (clear) the contents of the .bbl file every time I compile?

5:10 PM

@DavidCarlisle I believe I may have found the problem, which appears to be something entirely different from what I've asked. I don't have a solution, so I could still use some help from this site, seeing as my Google Fu hasn't provided me with answers yet. Before I start a new question, should I close, delete or answer my previous one? Which is the appropriate action?

The problem appears to be related to the compilation process.

@DavidCarlisle I want to change the index in a split index environment, slightly altered to enclose arraystretch of the tabular as an argument to makeindex. So \index{tableindex}{#1}{#2}{#3}{#4}{#5}{#6} and create an index item in a longtable format so \begin{longtable}{<arguments etc.>}
<some header>
\endhead
#1 & #2 & #3 & #4 & #5 \tabularnewline (next item)
#1 & #2 & #3 & #4 & #5 etc.
#6 would be a special case which it easiest to code for a case by case scenario (not recursive - I doubt makeindex has the customizability for it given you mentioned it's lack thereof), it'll determine the …

@DavidCarlisle Your previous answer to recursively split up elements (the one with the registry counter) would probably serve fantastically here for the layout part, with as input #3 (just with the regular , delimiter) to completely circumvent the need of #6.

5:28 PM

LOL existentialcomics.com/comic/29

7

percusse

@AlanMunn :-) You find the most unexpected twists.

5:41 PM

@eiterorm any of those. If you think you just made a one-off error delete but if you think there was an error that someone may gain benefit from seeing an answer self answer and accept your answer

@percusse I really like this comic.

@AlanMunn Are you thinking to SE badges?

@DavidCarlisle This is certainly something that might benefit others. However, the "answer" will be little else than a link to a new question. Will closed questions still be available? If so, I'll just edit in the link to the new question and close the old one.

@1010011010 I have no idea what you mean:-) as predicted a working example and a real question are likely required. But anyway @egreg is the index expert he even has a package for that purpose:-)

@1010011010 I guess you should look at the glossaries package.

5:48 PM

@eiterorm if you delete it I think it can't be seen by people with les than 10K rep

@DavidCarlisle Is 'close' the same as 'delete'? They show as different options.

@eiterorm Closed question are different

@egreg For those that hunt them, perhaps.

@eiterorm probably not (I have no idea how this site works:-)

@JosephWright So, will closed questions work as other questions, the only difference being that it's inactive?

5:51 PM

@eiterorm No, 'close' is not the same as 'delete'. Closed questions remain on the site and viewable (e.g. duplicates show up with a link to the duplicate question). Deleted questions are invisible to users with less than 10K reputation.

@AlanMunn Ok, thanks. Then I'll post a new question, link the old one to the new one, and close the old one. But first: gym.

6:04 PM

@egreg: tex.stackexchange.com/questions/179332/…

7:04 PM

Where can I find a list for makeindex keys that go into the *.ist or \jobname.mst file?

7:15 PM

Never mind, found it.

7:35 PM

In retrospect, this request may be more complicated than anticipated.

Hi everybody

8:03 PM

@DavidCarlisle Posted the full concept on front page, if you're interested.

@MarioS.E. Hi!

@egreg Gregorio, since you are Maestro... and since I've playing with LaTeX since a couple of years now... how do you get to write in the margins? hehehehehehe

@egreg I'm thinking specially in books like this from Jean Luc Duomont: treesmapsandtheorems.com/pdfs/TM&Th-samplepages.pdf, or even this one from our very own Nicola Talbot dickimaw-books.com/latex/thesis/thesis-report.pdf Even the Koma Script Documentation seems to have some words written in the margins (ftp.ist.utl.pt/pub/ctan/macros/latex/contrib/koma-script/doc/…)

@MarioS.E. \marginpar (although lots of variants such as marginnote or todo package)

8:20 PM

I've come to love \marginpar because it so seamlessly aligns the marginpar with the anchor, without leaving any horizontal space in the body text.

@1010011010 and sometimes it even manages to put the text in the correct margin:-)

@DavidCarlisle Show the old school marginpar some love!!

@DavidCarlisle and how do they align it to some section title

@MarioS.E. depending on the format of your section title you can probably just put \marginpar{foo} in the title, but make sure to use the [] to supply a title without marginpar for the toc

@DavidCarlisle Nice :)

8:45 PM

@PauloCereda bbc.co.uk/sport/0/cricket/27491216

@DavidCarlisle Oh for pity's sake

@JosephWright just flicked over to bbc news site to see if there was any good news:(

@DavidCarlisle Have you looked at the Lua script updates?

@JosephWright got a bit sidetracked in fixing europecv (and trying to find an email address of the author that works..._) (in between unicode math discussions:-)

@DavidCarlisle :-)

@DavidCarlisle No problem: I will add a few more bits in a little while

8:50 PM

@JosephWright woah lots of new tlg :-)

@DavidCarlisle Not yet: just modified ones!

@JosephWright that's what I mean:-) not looked at the normalisation code yet but seen your log msg and email

@DavidCarlisle Cool

@DavidCarlisle I've spotted a few issues in the generic loader: hope to fix them and sent to CTAN :-)

@JosephWright so where is the source of lualatex format (and who does it?)

@DavidCarlisle Huh?

8:55 PM

@JosephWright initex input for lualatex format is it just latex.ltx or is something inserting new allocation macros etc?

@JosephWright found it : kpsewhich lualatex.ini

@DavidCarlisle Same as for all engines: look for the .ini file

@JosephWright I think last time I looked there wasn't an ini file :-)

@JosephWright we really should pull this kind of stuff "in house"

% some commands in the kernel need patchin to work with luatex
\input lualatex-patch-kernel.tex

@DavidCarlisle Yes, agreed in principle

@DavidCarlisle See also xetex.ini :-)

9:18 PM

@DavidCarlisle Right, CTAN updated so back to Lua :-)

@HeikoOberdiek I've sent CTAN the fix for using expl3 with custom formats: should make the TL2014 DVD

I've tried running the experimental accessibility package and at about 90% of the document it told me input stack size exceeded... now I know that input stack size is generally speaking caused by a poor macro, but it really felt like I was so close :-(

What's the worse that can happen from increasing it?

@JosephWright Thanks.

@MarioS.E. \marginpar?

@DavidCarlisle Just nine runs? Not bad. ;-)

@JosephWright I should never have looked: it only patches one command and breaks it:

$ lualatex zz127
This is LuaTeX, Version beta-0.79.1 (TeX Live 2014) (rev 4971)
 restricted \write18 enabled.
(./zz127.tex
LaTeX2e <2014/05/01>
Babel <3.9k> and hyphenation patterns for 78 languages loaded.
tyoe here:

\foo =#
! Illegal parameter number in definition of \foo.

@1010011010 nothing

@egreg It was the confidence the BBC had in our national team "England's home summer campaign began in familiar fashion " that was particularly depressing:-)

9:33 PM

@DavidCarlisle What are you up to?

@JosephWright the redefinition of typein in lualatex-patch-kernel.tex puts the input through an extra \def so I gave it # and bang :(

@DavidCarlisle Question is what's it for?

@JosephWright oh because we (I, I think) added \@m to endlinechar so no character gets added at end of line and that's not enough in luatex

@DavidCarlisle Probably a LuaTeX bug: see recent bug reports/fixes!

@JosephWright no not a bug in this case 1000 is a legal character code

9:40 PM

@DavidCarlisle Ah

@JosephWright but could add "1000000 or something instead

@JosephWright LuaTeX limits the valid range for \endlinechar to ≤ 127.

@DavidCarlisle Or something like

  \expandafter\edef\expandafter#1\expandafter
    {\expandafter\unexpanded\expandafter{#1}}%

instead of the \def

@JosephWright yes you know we added \@m and subtracted it again rather than saving and restoring the old value in a group because it saved a couple of bytes, and we needed the space....

@DavidCarlisle :-)

9:46 PM

I wonder how many people have passed # to \typein over the years ....

2

@HeikoOberdiek Why did they do that?

@egreg :-)

@egreg I do not know, perhaps it simplifies the implementation by using a byte instead of a Unicode character for the end of line character.

@HeikoOberdiek Well, it's a big regression. And it breaks existing code.

@egreg At least it is documented behavior for quite some time.

9:53 PM

@HeikoOberdiek is it documented that negative values = -1 ?

\makeatletter


\showthe\endlinechar
\advance\endlinechar-\@M
\showthe\endlinechar

@DavidCarlisle Why not the following?

\def\@xtypein[#1]#2{%
  \typeout{#2}%
  \chardef\reserved@a\endlinechar
  \endlinechar=\m@ne
  \read\@inputcheck to#1%
  \endlinechar\reserved@a
  \@typein}

@egreg well it leaves reserved@a changed (which isn't a very reserved command despite its name:-)

The documentation only says:
> 9.1 Changes from TEX 3.1415926
> The upper limit to `\endlinechar` and `\newlinechar` is 127.

@egreg try passing in \reserved@a as [#1] :-)

@DavidCarlisle Use anything else. ;-)

@DavidCarlisle Big problem, however.

9:59 PM

@David: Wins: Another triumph for Great Britain. / Losses: England loses again.

@DavidCarlisle "Nice", now this is not documented AFAIK.

@HeikoOberdiek I can see running the latex test suite over luatex and xetex is going to be generating a lot of mail:-)

@DavidCarlisle I have just written the next email to the luatex mailing list.

10:15 PM

@egreg probably it simplifies passing stuff to lua (which doesn't really believe in unicode) as \endlinechar gets added to \directlua stuff (yeh @HeikoOberdiek's not the only one who can read manuals:-)

@HeikoOberdiek and I have just mailed tex-live list about the typein patch..

@DavidCarlisle Wrong address. You could fix \typein in LaTeX2e. ;-)

1 hour ago, by David Carlisle

@JosephWright we really should pull this kind of stuff "in house"

@HeikoOberdiek I was just looking what was in the format other than latex.ltx, so far I've looked at one macro...

10:47 PM

This one can be closed now: tex.stackexchange.com/questions/179172/…

Manuel

11:19 PM

I've been reading a little bit about LaTeX3 in the last few weeks… and I asked myself many many times about something that I don't fully understand: shouldn't in TeX something like \math{…} or \math …\endmath be the mathematical environment, and $…$ just a “front end” rather than $…$ be the math itself?

@Manuel Well, $...$ is the primitive way, the only one in “naked TeX”.

@egreg: how are you feeling tonight?

@PauloCereda Much better. The pain is much less now.

Manuel

I know, but still, I don't fully understand. I think I would prefer to know that \math{…} is the math environment, and then something like \def$#1$… was done. And something similar to those category codes like “alignment”, “superscript” or “subscript”. Is there a necessary reason to give them their own category code rather than let them be “active” or something like, e.g., “just a front end to \sb{…} or \sp{…}”?

11:45 PM

@Manuel That's what Knuth decided. In LuaTeX there are \Ustartmath and \Ustopmath that can be used instead of $ for opening and closing math; also \Ustartdisplaymath and \Ustopdisplaymath with obvious meaning; \Usuperscript and \Usubscript are substitutes for ^ and _.

There's also \aligntab that can replace & in alignments.

@Manuel Interesting, but I'm going to bed. I don't know your time zone, in mine it's almost 2am. ;-)