TeX, LaTeX and Friends - 2016-12-05

« first day (2227 days earlier) ← previous day next day → last day (2966 days later) »

Maeher

08:50

@cfr Surprisingly, they are defined by tikz (in tex/generic/pgf/frontendlayer/tikz/tikz.code.tex) using the \alt macro of beamer without declaring any dependency for beamer (seems wrong somehow).
Also means that its apparently not possible to customize the behavior, since \alt ist hardcoded in there.

It somehow would make a lot more sense if the overlay aware path macros would be defined by a separate tikzlibrary that required beamer. And you should be able to control their behavior with something similar to \setbeamercovered.

At the moment you can actually use \node<1>{} in any document using tikz and it will somewhat surprisingly result in an error about the undefined control sequence \alt.

Joseph Wright

09:08

@Maeher How odd

@Maeher I guess comes from the fact Till originally wrote both pieces of code 'together' (or at least the beamer/pgf bits: beamer doesn't load TikZ)

Joseph Wright

09:21

@Maeher Reprot to the TikZ maintainers, I guess

David Carlisle

09:54

@JosephWright get the beamer maintainers to offer to maintain tikz as well, to keep things in sync?

David Carlisle

10:12

@JosephWright I suppose I should make latexrelease load luaotfload when moving the date forwards, so I need to put the everyjob code into latexrelease with some suitable guards :(

Maeher

10:42

@JosephWright I guess I should. Though looking at the open issues from 2013 in their bugtracker, I'm not exactly confident that it will be addressed. ;)

3 hours later…

percusse

13:43

@Maeher TikZ is trying to be beamer compatible. Not the other way around.

Since beamer is not changing (in the eyes of old pre-Joseph Wright development team) you only need overlay awareness.

Joseph Wright

@percusse Certainly not changing much even now (small steps where they are safe/sensible)

percusse

\node<1>{} is syntax error for any TikZ code sans-beamer anyhow. But you can alter what \alt does to hack in.

@JosephWright I'm sure there will come a time when you will say, you know what? I can convert this spagetti into L3keys beauty.

so when do we see l3ynote package on CTAN ?

Joseph Wright

@percusse There are about 8k lines; certainly a rewrite is tempting, but I feel far too risky (also I'm busy enough)

@percusse There's the whole style sheets business to solve for that

percusse

@JosephWright Ah yes.

Maeher

@percusse Sure, it's just not the syntax error I would be expecting. I'm not certain what the best solution would be, but defining macros using another package without making sure its available seems wrong to me.

But I don't really care that much.

What I really wanted was to modify the behavior of overlay aware tikz nodes. Because the current behavior is suboptimal for my use case.

percusse

13:51

@Maeher Why not ? Don't you enjoy slowly revealing TikZ pictures in your presentation?

Maeher

Turns out that's not possible at the moment.

yes, in fact I do

percusse

What is your use case that doesn't fit?

Maeher

you can't do that well at the moment, because a node that is not currently on the slide simply does not exist. That causes a lot of problems. The picture changes size, so I have to take care that it does not jump around. I cannot use relative positioning anymore, because the nodes suddenly don't exist.

David Carlisle

@JosephWright I'm sure you are relieved to know where that 0.00002pt came from.

Joseph Wright

@DavidCarlisle Indeed

percusse

13:54

@Maeher That's a different problem about frames jumping around right?

Maeher

no, not really

I can get the result I want by putting \visible<> around my nodes

and I would like to be able to configure tikz/beamer to behave that way with the overlay aware path commands

percusse

So if I understand correctly, you refer in the current slide to a node that becomes visible in the later slides. Is that about right?

Maeher

I want all nodes to always exist just not be visible

yes

that happens quite often

percusse

That's by design of beamer. Not a TikZ problem. If you make things overlay-aware. Beamer reads only the stuff that are defined at that slide. And makes multiple copies by advancing the slide number

Maeher

hm, I think you are misunderstanding

percusse

13:58

possible. But I feel that you want to hack beamer not TikZ.

Maeher

well, not exactly. The problem is the definition of the overlay aware node command. I will have to replace them.

percusse

With what for example?

Maeher

\documentclass{beamer}
\usepackage{tikz}
\usetikzlibrary{positioning}
\begin{document}
\begin{frame}
\begin{tikzpicture}
\node (a) {a};
\node<3->[left=of a] (b) {b};
\node<2->[left=of b] {c};
\end{tikzpicture}
\end{frame}
\end{document}

This should work in my opinion, but does not

because b is undefined when c is placed for the first time

percusse

Are you familiar with overlayarea?

This is exactly why they have been invented

Maeher

not in detail, how would it help with the fact that the node is undefined?

maybe I'm misunderstanding what it's for

percusse

14:09

@Maeher That won't help with the undefined behavior. But if you change the order of definition and dependency you won't get any jumpy behavior and the graphics will stay the same.

You have to do that when designing the figure anyways. A burden on the user unfortunately. I would kill for that feature too but it's not there. I wasted a good chunk of time for my phd defense presentation for similar stuff.

But again, I would have wasted that time on other things anyways in the name of procrastination

Maeher

To me that seems like a (suboptimal due to harcoded sizes of the overlayarea) workaround for a design flaw :)

percusse

That's not a design flaw in my opinion you look for a unicorn :)

You can define dummy nodes in the beginning of the picture and draw on top of them later for example.

Maeher

Well, not really, I have found out how to fix it to behave like i want it to by now

it's a single line actually

\long\def\tikz@eargnormalsemicolon<#1>#2;{\visible<#1>{\tikz@@command@path#2;}}%

it's just that it messes with internal definitions of tikz

percusse

So if you don't want the visible part what do you do redefine it ?

Maeher

that line I just wrote is the redefinition

the original line is
\long\def\tikz@eargnormalsemicolon<#1>#2;{\alt<#1>{\tikz@@command@path#2;}{\tikz@path@do@at@end}}

percusse

14:16

Yes, but if I want to go back to \alt behavior what do I need to do ? Reredefine it ?

Maeher

Well, that's more complicated, yes :)

In my opinion, it would make sense to make this configurable with a command like \setbeamercoveredtikznodes{invisible} or something like that

percusse

Note that your solution is actually processing the nodes at each slide. But only switching it's visibility.

So from bounding box calculation to other details you are actually altering the definition of the picture.

\alt on the other hand completely ignores the code itself as if it is not there.

Maeher

that's kind of the point :)

I'm not suggesting that only switching visibilty is a solution to a problem, I'm saying I want to have the ability to choose which behavior I want.

percusse

Well, I wouldn't really want that but you can make your definition a bit more flexible

\makeatletter
\def\setbeamercoveredtikznodes#1{\long\def\tikz@eargnormalsemicolon<##1>##2;{\csname#1\endcsname<##1>{\tikz@@command@path##2;}{\tikz@path@do@at@end}}}
\makeatother
\setbeamercoveredtikznodes{visible}

Then you can you use with visible or alt

Maeher

Maybe I'm the only person in the world who wants invisible nodes, I don't know.

percusse

14:25

Well you are one too many to ignore that feature :)

David Carlisle

@Maeher despite the fact that the thing you are redefining is defined in tikz, looks more of a beamer feature request than a tikz one

Maeher

I was just checking how other covered settings are done in beamer so I could maybe write a short package that (consistently with other beamer settings) adds this feature by redefining the internal tikz command.

percusse

I think you just don't like typing the extra \visible stuff.

Maeher

yes, that's pretty much it :)

percusse

Ooh soon you will feel the wrath of emacs and vim knights ...

I have defined shortcuts for that on texniccenter. So that's for me the psychological limit in terms of keyboard wizardry.

Maeher

14:31

It just also seems inconsistent with other overlay stuff in beamer. I mean, the standard overlay for lists and stuff is not \only... they always take up space :)

percusse

@Maeher No. The blocks also take space as they appear.

That's why you need \begin{frame}[T] to align them etc.

@DavidCarlisle ^^^^ You have really strange colleagues I have to say. I have to parse the syntax of ?SYTRF of LAPACK. I have been unpacking that packed output since last 3 days :)

David Carlisle

@percusse in python?

Maeher

@percusse I guess I do not use blocks very often, at least not (dis)appearing ones

percusse

@DavidCarlisle Yep. I know that output is intended for use with ?SYSV, ?SYTRI but still that's quite some clever way of storing output heh.

@Maeher Yeah. Being a package author is really difficult. You have consider everyone.

Well almost everyone

user image

@DavidCarlisle This is just one of the sanitizers it goes on and on :) Hopefully I'll dump it to SciPy repository and never look at it again

Maeher

@percusse For me blocks seems to take up space whether they are visible or not

\documentclass{beamer}
\begin{document}
\begin{frame}
a
\only<2->{
\begin{block}{aaa}<3->
\end{block}
}
\end{frame}
\end{document}

^^^ the a jumps between the first and second slide, not between the second and third

David Carlisle

14:50

@percusse at some point I should contact you offline about python interfaces....

daleif

15:24

Strange question: I have a truetype font that only contains four chars. (1) how do I figure out which slots in the font these char are, (2) how do I then ask for these specific slots in a doc? Currently testing using xelatex (the font contains four logos)

David Carlisle

@egreg slowcoach:-)

@daleif if you know the glyph names you can access them by name and don't need to know the numbers at all

Joseph Wright

@daleif At they in the PUA? (So 'anything goes')

egreg

@daleif otfinfo -g NAME.ttf shows the glyph names also in TrueType fonts.

David Carlisle

@egreg seems like referendums are a dangerous game to play not just for British governments?

daleif

@egreg uniF030 through uniF033

what is the interface for accessing the glyf (can this work on pdflatex as well, going to truetype means I'm loosing latex+dvips users in my application, no big deal, but loosing pdflatex users is something I'd rather not)

David Carlisle

15:32

@daleif so if you know that you know the answer to your question in this case?

daleif

@DavidCarlisle I still don't know how to access a glyf by name

David Carlisle

@daleif well by number you could just use \char"F030 but for pdftex you must have remapped the font to some 0-256 range so it's wherever you mapped it, by name in xetex you use...

daleif

@DavidCarlisle Ok, for a xelatex test I could just use \XeTeXglyph\XeTeXglyphindex"uniF030"

David Carlisle

@daleif yes that (was just checking the syntax:-)

daleif

@DavidCarlisle interesting, the \Xe... works in xelatex, `\char"F030" does not (should it work in xelatex?)

egreg

15:38

@daleif \symbol{"F030}

David Carlisle

@daleif well assuming that the glyph called uniF030 is encoded in slot hex F030 yes (do you get a missing character warning in the log?)

daleif

@egreg nope (still testing through xelatex), guess it is non-standard

@DavidCarlisle no errors, just a box in the PDF

$ otfinfo -g AULogoReg.ttf
.notdef
space
uniF030
uniF031
uniF032
uniF033
nonmarkingreturn
NULL

egreg

@daleif So the font has strange names that don't correspond to the slot.

David Carlisle

@daleif doesn't that mean they are encoded in slots 2,3,4,5 so \char2

daleif

@DavidCarlisle nope. If I open it in fontforge it says 258 (0x102) through 261 (0x105)

BTW: is is complaining about missing chars in the log, just not the console log. Will look again tomorrow.

egreg

15:46

@daleif Something like this might give you the numbers

\font\test=AULogoReg.ttf

\count255=0
\loop\ifnum\count255<"10FFF
\advance\count255 1
\iffontchar\test\count255 \the\count255={\test\char\count255}\endgraf\fi
\repeat

\bye

David Carlisle

@daleif yes that's the tex way

@daleif oh ok but not hex F030 though in anycase:-) so \char258 should work

Toby

16:14

Persuant to this question tex.stackexchange.com/questions/342494/… I want to see if XeLaTeX would make things easier for me but dont have much of a clue how all this fits together so have a few small question to get things set up

I am using Doxygen that produces LaTex output, currently I use MiKTex to ccompile

Google tells me MiKTeX supports XeLaTex

egreg

@Toby I added a possible way for getting automatically the names.

Toby

So is it possible to alter the Doxygen LaTeX output such that MiKTeX h"handles" it with XeLaTeX? or is XeLaTeX much different ?

Thanks @egreg

(Im currently operating under the idea that XeLaTeX is just another variation of Tex, like LaTeX is...)

egreg

@Toby On the other hand, if a symbol is not in amssymb, you're again with the problem of finding a suitable font supporting it.

Toby

Yep

I guess the thing was I looking for a way to reduce the amount of TeX script in my documentation comments. As it is the comments are easy to read in source AND in PDF, with lots of TeX script in there it makes it harder to read in the cource, without it the PDF won't compile >.<

Joseph Wright

@Toby If we assume reasonably common maths symbols, they will be available in the AMS fonts

Toby

16:24

ta

Joseph Wright

@Toby XeTeX is like TeX, XeLaTeX is like LaTeX: yes, we know this is not straight-forward!

Toby

LOL, thanks @JosephWright my TeX foo is very very small ATM

Joseph Wright

@Toby We've got something explaining it, but in a nutshell XeLaTeX allows you to use all of Unicode with TeX. You still need a font with the right symbols though ...

Toby

OK.

And XeLaTeX is like LaTex, but would LaTeX compile with a XeLaTex compiler?

Joseph Wright

@Toby Practically, you run xelatex instead of latex

Toby

16:28

Hm, OK - ATM I don't really "run" latex, Doxygen produces a bunch of tex files and whatnot and I run the make.bat (Im on windows) it also outputs...

Ill check what's inside that now

Joseph Wright

@Toby 'Like' as in 'you can treat as identical other than in certain specific places'

@Toby Presumably somewhere there's a setting that says latex or perhaps pdflatex

Toby

Cool

OK so the bat is about 30 lines, with a few pdflatex calls. Guess Ill try replacing them with xelatex and see what happens! :-)

Toby

16:54

Think Im slowly getting there with it, when compile it says it cant find xetex-def though, and that I have the wrong DVI mode. I googled on the former issue and found, from some pretty old posts that xetex-def is not used any more so I used graphics-def. Now it still complains about it but carrys on anyway. Hunting for info on DVI mode now (currently ps2PDF)

Hm no luck unfortunately, the Doxygen LaTeX requires too much modification

Sebastiano

17:11

Hi to all.

Dear @egreg excuse me lot lot for saturday night. Unfortunately I have a bad internet connection and often falls on the network. Excuse me again also because I have written in Italian language.

I will write soon. Goodbye and have a nice day. I hope that his day yesterday went well. With estimates. Sebastiano.

Regards to all.

StrongBad

17:48

Is there something like a \let for lengths? For example, I would like to do \setlength{\textwidth}{\textheight} and have \textwidth get updated whenever \textheight gets changed, but I would still like to be able to do \setlength{\textwidth}{4in} to break the link between the two.

Joseph Wright

@StrongBad That's not how \let works ...

StrongBad

@JosephWright you are right. I guess it is closer to how \def works. Is there a way to do it?

egreg

@Sebastiano Quite well, thanks. 653 voters showed up, about 80%

3 hours later…

David Carlisle

20:27

@StrongBad it's not really how def works either (it's just weird:-)

@StrongBad you could do \textwdith=\textheight or \def\textwidth{\textheight} or \let\textwidth\textheight which all have some effect but none of them have the effect that you said. You really want to use grouping to "break the link"

Canageek

21:03

Question: anyone know what this error message means? ! Argument of \@xfloat has an extra }.
<inserted text>
\par
l.400 \begin{figure}[h]

I'm guessing the RSC packages have done something dumb....

StrongBad

@Canageek Does this help you track it down: tex.stackexchange.com/questions/315905/…

Also eng.fsu.edu/~dommelen/l2h/errors.html#argex

Canageek

@StrongBad Looking.

This is why LaTeX should suppresses some of this pointless font stuff; there are hundreds of lines of it preventing me from finding what is actually wrong.

David Carlisle

21:20

@Canageek usually it means the class hasn't been updated for a change I made in 1993 to \@xfloat to support colour.

Paulo Cereda

@DavidCarlisle: missed me? :)

David Carlisle

@PauloCereda no, who are you?

Paulo Cereda

@DavidCarlisle some duck. :)

David Carlisle

@PauloCereda dinner!

Paulo Cereda

@DavidCarlisle oh no

David Carlisle

21:23

@Canageek do you load color (or xcolor) ?

Canageek

@DavidCarlisle Yeah, the template does. \usepackage[usenames,dvipsnames]{xcolor}

David Carlisle

@Canageek I had to add an extra \color@vbox to \@xfloat and ` \color@endbox` to \@endfloatbox to keep colours in the float, these are \relax unlesss colour is loaded when they are an extra group, so if a class redefines \@xfloat but not \@endfloat and bases it on a latex2.09 definition from 1985 you get mis-matched {} when you load colour. that was perhaps the major incompatibility between latex2e and 2.09, but doesn't come up so often these days....

@Canageek is the class available? i could have a quick look...

Canageek

@DavidCarlisle It is the template from here: rsc.org/globalassets/05-journals-books-databases/…

David Carlisle

@Canageek dare I look? :-)

Canageek

@DavidCarlisle But I've commented out mhchem, times,mathptmx

@DavidCarlisle Apparently it is a copy of @JosephWright's work

@DavidCarlisle But it is the template that loads xcolor, so it probably isn't broken by default, probably something I've done

David Carlisle

21:36

@Canageek well I'd blame him even if it wasn't, being as it's Chemistry.

@Canageek i was just going to say same, I had to fake a widetext.sty but template loads color and has a figure so...

Canageek

Oh, that is in the packages folder

Removing the comments

\NeedsTeXFormat{LaTeX2e}
\ProvidesPackage{widetext}
%% Package required for equal height while in 2 columns format
\IfFileExists{flushend.sty}
{\RequirePackage{flushend}}
{\typeout{}
\typeout{Package widetext error: Install the flushend package which is
a part of sttools bundle. Available from CTAN.}
\typeout{}
\stop
}

%% Package required for onecolumn and twocolumn to exist on the same page.
%% and also required for widetext environment.
\IfFileExists{cuted.sty}
{\RequirePackage{cuted}}
{\typeout{}

David Carlisle

@Canageek didn't matter I just made a one line file that did \newenvironment{widetext}{\par}{\par}

@Canageek my version is simpler:-)

@Canageek well I'll look if you can tell me what to change to make it break:-)

egreg

@StrongBad You can define a property list for lengths linked to a particular one

\documentclass{article}
\usepackage{xparse}

\ExplSyntaxOn
\NewDocumentCommand{\declaredependentlengths}{mm}
 {
  \tl_set:Nx \l_strongbad_dependent_length_tl { \cs_to_str:N #1 }
  \prop_new:c { g_strongbad_dependent_ \l_strongbad_dependent_length_tl _prop }
  \keys_set:nn { strongbad/dependent } { #2 }
 }

\keys_define:nn { strongbad/dependent }
 {
  unknown .code:n =
   \prop_gput:cVn { g_strongbad_dependent_ \l_strongbad_dependent_length_tl _prop } \l_keys_key_tl { #1 }
 }

\NewDocumentCommand{\setdependentlengths}{mm}

Canageek

So far I can't break it copying more and more of my document into it

ffffff

David Carlisle

21:56

@Canageek see perhaps I fixed it:-)

Canageek

@DavidCarlisle But my document does not work at all yet

David Carlisle

@Canageek well there is always the direct user error possibility \begin{figure} ... { \end{figure}

Canageek

@DavidCarlisle Checked that. Easy as both are highlighted in red

David Carlisle

@Canageek well you or a package has got a mismatched brace in a float such as figure, other than that I can't really say!

Canageek

@DavidCarlisle I'm going through with a machete now

"Is this 100% needed to compile? No? Delete"

David Carlisle

22:08

@Canageek good thing is we can blame @JosephWright for the time it's costing you.

Canageek

22:20

Found the code block that breaks it

There is an unmatched \twocolumn[ somewhere

David Carlisle

@Canageek or any optional argument inside the [] of \twocolumn since latex doesn't match [..] use \twocolumn[{....}] to make nested [] safe

Canageek

@DavidCarlisle No, as in, when I was copying text in, I missed the end of that environment.

I really think helpful error messages would be a big boon to LaTeX

People would lynch the GCC people if they gave this cryptic feedback...

Joseph Wright

@Canageek Not exactly what I meant: the 'print like' stuff is their own!

Canageek

@JosephWright Print like?

David Carlisle

@Canageek don't compare tex to gcc compare it to the C pre-processor, make 1000 lines of nested #define and see how clear the error messages are if you get one wrong.

Canageek

22:32

@DavidCarlisle Yeah, but normal humans have to use LaTeX

Not programming demigods ;)

Joseph Wright

@Canageek The bits that I recognise are more about the 'advice' side of things (cf. achemso, where there is advice but not really an attempt to look like the printed journal)

David Carlisle

@Canageek well they shouldn't be using a macro processor if they want high level diagnostic output.

Canageek

@JosephWright Ah, OK

@DavidCarlisle Is there a way to get high level diagnostic output?

David Carlisle

@Canageek yes write a typesetting system in a compiled language and using a modern scripted customisation layer rather than a 1970's text macro expansion paradigm......

Canageek

22:50

Now I just need to figure out why my text runs off the bottom of the page, on every single page

Canageek

23:01

mhchem? Disabling mhchem was the problem?

David Carlisle

@Canageek not done any chemistry since 1979, so not a package I've used much :-)

Canageek

@DavidCarlisle Lets just say it isn't normally used for layout to my knowledge, so I was rather surprised when turning it back on fixed a page layout issue.

David Carlisle

23:18

@barbarabeeton :

looks like a bug to me, you could report it to stix.. — David Carlisle 1 min ago

barbara beeton

@DavidCarlisle -- yup. thanks. there's also the bug with the slashed zero. i've been told to post that to the sourceforge issue tracker. which i will do as soon as i can find my password.

Canageek

user image

Last thing I need to figure out: what the heck is wrong with this table

If I use table* then it looks even worse then this.

David Carlisle

@Canageek it's over-full by two columns, but hard to say just from the output

cfr

23:43

@Maeher How odd. [I realise this was already said, but it bears repeating.]

Canageek

Decided to take the easy route and split it into two tables

Then it decided to insert a ton of space before the final column and I don't know why

David Carlisle

@Canageek tabular* ?

Canageek

@DavidCarlisle Yes.

David Carlisle

@Canageek so presumably that was the only place where you have glue added (\extracolsep` aka \tabskip)

Canageek

@DavidCarlisle Probably, but I wouldn't count on it due to template

@DavidCarlisle I don't use that deliberately anywhere.

user image

At least before the spacing was even

David Carlisle

23:52

@Canageek using tabular* without adding \extracolsep is basically an error it specifies the table should stretch while giving it no possibility to do so

Canageek

@DavidCarlisle Ok, where do I put it?

David Carlisle

@Canageek usual form is {@{\extracolsep{\fill}}ccccc@{}}

Canageek

So like \begin{tabular*}{\textwidth}{@{\extracolsep{\fill}}l*{5}{c}@{}}

David Carlisle

@Canageek yep

Canageek

@DavidCarlisle That didn't change anything

David Carlisle

23:55

@Canageek but if those lines are \hline then the table is overfull so tabular* won't do any stretching

@Canageek as I was writing:-)

Canageek

@DavidCarlisle OK, so one less column or use \small

egreg

@DavidCarlisle My Phoenician may be rusty, but the alphabet is easy to remember.

David Carlisle

@Canageek something like that

@egreg :-)

Canageek

@DavidCarlisle Got it! Was the \multicolumn I had at the bottom; was too long and stretching things

« first day (2227 days earlier) ← previous day next day → last day (2966 days later) »

Transcript for

Dec '165

$TeX - LaTeX$ TeX, LaTeX and Friends

\begin{chat}

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

site design / logo © 2025 Stack Exchange Inc; legal

mobile