TeX, LaTeX and Friends - 2024-11-26

« first day (5140 days earlier) ← previous day next day → last day (89 days later) »

Werner

00:00

@cfr Yes. Comment flags can't be handled with a response from a mod. It's just delete, edit, decline, delete & decline.

So, I just wanted to let you know that a decline of that comment flag was approval of me not thinking anything needs to be done...

...hmmm... lots of negatives in that sentence.

Alan Munn

00:47

@cfr no no need I think. Hobby does everything I need. One thing that would bee useful would be to automatically figure out the coordinates of the bottom left corner of a box around a node and its terminals and use that as the path through with hobby

Alan Munn

01:03

@cfr So for example, in the tree I gave in my example, it would be intuitively nice to say "draw a curve from ddamwain to disgrifiad avoiding the DP containing y gyrrwr"

@cfr This would require naming the start and end nodes, and the node to avoid.

@cfr Right now I'm naming the lowest node in the "avoid" node, and working with that.

cfr

01:54

@AlanMunn so that's not the arrow in the example, right?

Alan Munn

@cfr Yes it is the arrow in the example

cfr

@AlanMunn that's not obviously an arrow from ddamwain to disgrifiad ...

Alan Munn

@cfr Oh sorry, the version I've been playing with is different from the one I posted here. Yes, the one here is draw from \TR{disgrifiad} to disgrifiad avoiding the same DP

So and intuitive interface would be to give that DP node a name and have the relevant point(s) in the hobby path be calculated from its terminals.

Right now I can do \draw[->] (N.south) to [curve through={($(driver.south)+(0pt,-10pt)$)}] (Num.south);

7 hours later…

daleif

09:01

@mickep dejavu-otf is rather strange, while load it stops for a little while at the message "Using symbolic names for the DejaVu font", and then loads. In contrast fontspec+setmainfont runs just fine. Not sure if come cache needs to be rebuild.

samcarter

@PauloCereda I need sun glasses with so many suns!

daleif

09:13

@mickep the math setup for dejavu-otf looks pretty ugly, and I'd like to avoid unicode-math for this (sorry, find it very annoying when you're mixing fonts).

daleif

10:12

Oh that message, comes when loading via font names instead of file names, file names seems safer, but then I get errors about DejaVuSans-ExtraLight-Bold.ttf missing.

And yes, that font does not exist. Probably better to just steal the relevant code instead.

daleif

10:34

Am I missing something. Say I have this
\RequirePackage{fontspec}
\setmainfont{DejaVuSerif}[% main rm
Scale=0.9,
Extension = .ttf,
UprightFont= *,
BoldFont = *-Bold,
ItalicFont = *-Italic,
BoldItalicFont = *-BoldItalic,
]
Why might this end up having a different line spacing than when I load a different font? The scale option? \normalsize is defined earlier than loading the font in this case.

Sigh, need to remember that the original font has \linespread{1.05}

2 hours later…

gusbrs

12:21

@samcarter Damn, found it! The font package is lh. But I used unorthodox methods and, in part, was just lucky, so I'll leave the question there a while to see if anyone has better ideas. Thanks anyway. ;-)

samcarter

@gusbrs good to hear you found it! It was one of the packages linked from ctan.org/tex-archive/fonts/ps-type1/cm-super , so a good lead :)

gusbrs

12:37

@samcarter Now that you mention it, indeed cm-super says it is a "replacement" for LH. But, well, I did try cm-super as I told you. It was not enough get \usepackage[russian]{babel} happy (I presume I'd have to be explicitly loading the font). And, well, the "lh" name did not call my attention at first glance, I wonder why such a conspicuous name failed to grab my eyes. :-D

samcarter

@gusbrs :)

2 hours later…

daleif

14:28

Does or doesn't lualatex have the ability to go online? See tex.stackexchange.com/q/731702/3929, I would hope not as I see that as rather dangerous.

1 hour later…

David Carlisle

15:30

@daleif it has luasocket.http compiled in but you can't use it from luatex (as opposed to texlua) without shell-escape and it can only do http not https, so for exanple for l3build upload we used a shell escape to curl rather than native Lua connection

daleif

15:57

@DavidCarlisle thanks.

1 hour later…

cfr

17:01

@AlanMunn something like this?

\documentclass{standalone}
\usepackage[linguistics]{forest}
\usetikzlibrary{calc}
\usetikzlibrary{hobby}
\newcommand\TR[1]{$\langle$#1$\rangle$}
\newcommand\1{$'$}
\forestset{%
  /tikz/dadfygio/.style={},
  debug me/.style={%
    before drawing tree={%
      /tikz/dadfygio/.style={draw,help lines},
    },
  },
  name fit to/.style 2 args={%
    tikz+={%
      \node (#1) [fit to={#2},dadfygio] {};
    },
  },
  fit tree/.style={%
    name fit to={#1}{c,tree},
  },
  Hobby to/.style={%
    split={#1}{/}{temptoksa,temptoksb},

user image

Paulo Cereda

@Rmano ooh squirrels are very good at starring repositories :)

cfr

@AlanMunn so Hobby to=<dest>/<skirt> goes from the current node to <dest> and tries to avoid <skirt>. If you want to be sure of avoiding it, you can route through the .south west and .south east anchors, but it will produce a looser loop when that's not necessary.

@AlanMunn but now I'm curious about the identifiers in the tree ...

@AlanMunn the fit tree style will let you do this directly. so fit tree=<name> as an option to the relevant node will let you use <name>.south west in a path. but you may need to delay drawing the path by a cycle because I think if you just do \draw after the tree spec, <name> does not exist yet. if I remember how forest does this, I think the stuff after the tree spec is added to tikz for the root node. (but I didn't check this properly.)

@AlanMunn alternatively, you could alter the processing order at that point. or you could use .max and .min to get the max and min x and y of the subtree rooted at <avoid>. but that would not (directly) yield the interface you wanted. there's a similar example in the forest doc, if I remember correctly.

2 hours later…

cfr

18:58

is e.g.

\usepackage[utf8x]{inputenc}

actually harmful or merely pointless? (I think it is merely pointless, but I'm not sure.)

Alan Munn

19:46

@cfr Wow thank you! That is really nice. I'm not sure I'm a fan of putting the move command as a style though; I prefer to have a separate command. Also, I have a style that adds a decoration along the midpoint of the path, and I'd have to figure out a way to use that with your code. It will take me some time to digest it.

@cfr Curious about the node labels themselves?

cfr

@AlanMunn yes, I realise. I tend to prefer options, but I think your preference is often clearer. but I think you may need delay or to alter the processing order in that case. that is, I take it you'd prefer e.g.

\documentclass{standalone}
\usepackage[linguistics]{forest}
\usetikzlibrary{calc}
\usetikzlibrary{hobby}
\newcommand\TR[1]{$\langle$#1$\rangle$}
\newcommand\1{$'$}
\forestset{%
  /tikz/dadfygio/.style={},
  debug me/.style={%
    before drawing tree={%
      /tikz/dadfygio/.style={draw,help lines},
    },
  },
  name fit to/.style 2 args={%
    tikz+={%
      \node (#1) [fit to={#2},dadfygio] {};
    },
  },
  fit tree/.style={%
    name fit to={#1}{c,tree},
  },
  Hobby to/.style={%
    split={#1}{/}{temptoksa,temptoksb},

[ignoring the fact this tries to draw the arrow twice] - that is, you'd prefer to say

  \begin{scope}[use Hobby shortcut]
    \draw [->] (N.south) .. (fitted at driver.south) .. (Num.south) ;
  \end{scope}

after specifying the tree @AlanMunn?

Alan Munn

@cfr Yes something like that, although I'd wrap the whole thing in a command to hide the innards.

@cfr No you said you were curious about the tree itself (which I assumed you meant what it is trying to represent)

cfr

@AlanMunn the thing is, that will give you an error because fitted at driver does not yet exist.

@AlanMunn yes, I am. but I don't understand the 'No'?

@AlanMunn so you either have to specify it as an option with something like delay={tikz+={...}} or you need to change the processing order. or you need to change the strategy more fundamentally .... can you think of an attractive interface which would get the ordering right?

@AlanMunn does that make sense? (do either of those things make sense?)

Alan Munn

@cfr Well at the moment I have a command that seems to work. But I'll need to play with your code to see if I can combine the two or not. It's a bit hard to share right now because it's part of a package in progress and turning it into a MWE means copying a bunch of other code.

@cfr As for the tree itself, what do you want to know?

@cfr One limitation of your approach over what I'm trying is that I guess you can only skirt a single node right? Because the processor takes only two nodes. Depending on the particular tree it might be necessary in practice to skirt multiple nodes.

cfr

20:29

@AlanMunn yes, I wondered about that. I think it would be relatively easy to extend and also to make it more generally flexible, but I wasn't sure if it was something you'd want to use.

Alan Munn

@cfr Let me think about it. I need to digest the code a bit. One thing that's also in the back of my mind is a command that would also work with tikz-qtree and not be completely forest dependent. I don't know how many linguists still use it, but if it could be accommodated easily that would be a plus for the package.

cfr

@AlanMunn I really wanted to know what the abbreviations stood for. (well, a couple I can probably guess.) beyond that, is it showing something about what?

@AlanMunn since I've never used that, I can't help you there. only qtree and forest.

Alan Munn

@cfr So this is an analysis of the adjective ordering facts we discussed a while ago, combined with some other properties that make Welsh different from English, but in a consistent way. So DP=determiner phrase, NumP=number phrase. NP=noun phrase, PP = prepositional phrase.

If you look a the English translation of the phrase it would be "The driver's detailed description of the accident". 'description' is the nominal version of 'describe' and in English the word order of the clause is SVO, and the word order in the noun phrase is SNO.

cfr

@AlanMunn good. (I guessed right.) (well, I don't know about the other facts.) & thanks. but what's SG?

Alan Munn

In Welsh, the word order in the clause is VSO and the word order in the noun phrase is NSO

@cfr SG=singular

cfr

20:36

@AlanMunn ah!

Alan Munn

@cfr And the <...> denotes that phrase being "there" in the structure, for interpretation but not pronounced.

@cfr So for example in a simple question "What did you buy" we would like to say that "what" is simultaneously at the top of the tree, but also in the object position of the verb.

@cfr tikz-qtree uses qtree syntax, but with tikz so you can add explicit nodes etc. It was a nice transition when forest didn't exist.

cfr

@AlanMunn I don't really follow the NSO part ...

Alan Munn

@cfr So you can think of "the driver" in the "the driver's description of the accident" as the subject of the noun phrase, just like "the driver" is the subject of the clause in "The driver described the accident"

@cfr So just like the clausal syntax in Welsh, "y gyrrwr" is to the right of the noun it's the subject of, just like it would be to the right of the verb it's the subject of.

cfr

@AlanMunn so what's AP?

Alan Munn

@cfr Adjective Phrase

cfr

20:45

@AlanMunn so how does it show NSO for an NP?

Alan Munn

@cfr I'm using "noun phrase" analagous to clause, NP is not a noun phrase in this sense, the whole DP is the noun phrase, i.e., the noun plus all its dependents

@cfr So "disgrifiad" is N and "y gyrrwr" is the S, and the PP is the O.

cfr

@AlanMunn oh, ok. sorry.

@AlanMunn why do you split between disgrifiad and manwl rather than between manwl and y gyrrwr?

Alan Munn

@cfr This is because arguments of words always combine before modifiers. So "y gyrrwr" is semantically most closely related to "disgrifiad" and so combines first.

@cfr So in a deverbal noun like "disgrifiad", the describer and the thing described are arguments of the noun, but "manwl" is not; it's a modifier.

@cfr We see the same pattern in English: you can say "a maths student" = student who studies math (argument) and "a Canadian student" = student from Canada (modifier). When you combine the two together the argument has to combine first: "a Canadian maths student" not "a maths Canadian student"

cfr

@AlanMunn but that doesn't really explain why 'manwl' ends up in one chunk rather than the other. intuitively, I'd think it is modifying 'disgrifiad', so it would get grouped that way.

Alan Munn

@cfr It's not modifying "disgrifiad" alone, but "disgrifiad" plus it's argument. And then "disgrifiad" has moved. But it is a syntactic constituent with the NP headed by "disgrifiad".

@cfr Just like an adverb modifies the whole V+Object even in Welsh when the V+O aren't actually adjacent. So you get V Subj Adv Obj

cfr

21:02

@AlanMunn do you have an example? what counts as an adverb?

Alan Munn

Maybe "yn anffodus"? It doesn't have to be literally an adverb. But maybe I'm not getting the facts right and it always shows up pre-sententially

cfr

@AlanMunn huh. maybe I just don't understand what 'adverb' really means.

rhedodd y bachgen yn gyflym i'r siop ('the boy ran quickly to the shop') ?

Yn anffodus, rhedodd y bachgen i'r siop ?

Alan Munn

@cfr I don't either. :) Maybe "adverbial" is a better term. Things that would translate to something with the –ly suffix in English. But could also be prepositional phrases etc.

@cfr Yes that would work

So "yn gyflym"is modifying the verb phrase even though it's not adjacent to the verb.

cfr

21:20

@AlanMunn but isn't the object 'the shop'? I mean isn't yn gyflym modifying the (verb - subject) rather than the (verb - object)? wouldn't V+O be V+ the shop?

Alan Munn

@cfr Well I think it's modifying both in reality. But certainly not just the verb plus the subject. Think about "He unfortunately drank the poison". It's not the drinking itself that's unfortunate, it's the drinking of the poison.

@cfr Or "He rarely eats meat" doesn't entail he rarely eats.

David Carlisle

@cfr it was actually harmfull but now it has been rendered a no-op so it's just pointless unless you load the ucs package which re-enables it and then what happens happens (it might be needed for some custom compatibility issues, but in general, stuff will break)

cfr

@AlanMunn I can see 'the whole thing'. it was the V+O part I didn't get.

@DavidCarlisle ok, thanks. I removed it in a patch I submitted, but that part of my patch didn't get accepted. so I was wondering if I should try to convince them to accept that bit as well, but it doesn't sound like it matters.

David Carlisle

21:35

@cfr patch to what? whatver it is that is doing that will generate this message on the terminal and log of every run:

Package ucs Info: utf8x disabled, assuming standard utf8 processing
(ucs)             load ucs package to force utf8x processing.

Alan Munn

@cfr Ok, yes the whole thing is a better description.

David Carlisle

@cfr so if they don't believe you they should believe latex. "load ucs to force utf8x" is just a polite way of saying "we have prevented you from shooting yourself in the foot, but if you are determined to do so, you can"

cfr

@DavidCarlisle well, I tried:

% inputenc obsolete

David Carlisle

@cfr what package is this?

cfr

@DavidCarlisle it's not in the package, it's just documentation.

David Carlisle

21:38

@cfr ah so only self inflicted not inflicting it on others so not so bad. documentation of what?

Rmano

@PauloCereda hmmm, yep, two today ;-)

cfr

@AlanMunn thanks.

@DavidCarlisle does it matter?

David Carlisle

@cfr no

cfr

@DavidCarlisle thanks.

Alan Munn

@cfr Always a pleasure. And thanks for the hobby code. I may have some questions once I've had time to digest it properly.

David Carlisle

21:48

@cfr just vaguely interested as at the time that message was added I did put some effort in to tracking down public uses of utf8x and getting them changed

cfr

@AlanMunn I think it would be easy to extend it. however, it is trickier if you want to create the fit node in the tree spec, but draw the arrow later. perfectly possible, but slightly more juggling.

@DavidCarlisle hmm. that's odd. did you check documentation or just packages?

David Carlisle

@cfr oh almost certainly just ran grep over the texlive input tree so just packages, I don't remember exactly

cfr

@DavidCarlisle or, maybe, when was that? if it was early enough, you can probably blame me.

David Carlisle

@cfr around this time github.com/latex3/latex2e/issues/24

cfr

@DavidCarlisle too late. I don't know whether I wrote that line in the first place but, if I did, it was considerably before that, when it was still (I think) needed. so I'm assuming you didn't grep the doc tree.

David Carlisle

22:01

@cfr i usually forget the doc tree exists

cfr

@DavidCarlisle 38 hits for \usepackage[utf8x]{inputenc} ;).

David Carlisle

@cfr yes finding them isn't the same as and getting them changed. Some of them may even be "needed"

cfr

@DavidCarlisle some are also spurious results. I didn't grep at all carefully.

doc/latex/egpeirce/egpeirce-doc.tex:\usepackage[utf8x]{inputenc}
doc/latex/chemexec/chemexec_de.tex:\usepackage[utf8x]{inputenc}
doc/latex/chemexec/chemexec_en.tex:\usepackage[utf8x]{inputenc}
doc/latex/apprends-latex/Apprends_LaTeX.tex:\usepackage[utf8x]{inputenc}
doc/latex/thaienum/thaienum.tex:\usepackage[utf8x]{inputenc}
doc/latex/thaienum/thaienum.pre.tex:\usepackage[utf8x]{inputenc}
doc/latex/lithuanian/testlt.tex:\usepackage[utf8x]{inputenc}
doc/latex/latex-brochure/brochure.sty:\usepackage[utf8x]{inputenc}

@DavidCarlisle 29 ^^

some are still probably spurious, though.

David Carlisle

22:18

@cfr ooh doc/latex/hyperref, I wonder who maintains that?

David Carlisle

22:35

@cfr ah it's in a verbatim section showing code that "is not recommended" :-)

Ulrike Fischer

23:00

@DavidCarlisle uff.

« first day (5140 days earlier) ← previous day next day → last day (89 days later) »

Transcript for

Nov26

$TeX - LaTeX$ 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