TeX, LaTeX and Friends - 2015-06-07

« first day (1680 days earlier) ← previous day next day → last day (3262 days later) »

9:08 AM

Hi all,

I have a font with an absurd amount of ligatures, different ornaments, swash alfabets, small caps variants, etc.

I am considering making a new encoding for it (I don't see another way to somehow use all that the font has to offer within TeX), but the comment made in ftp.dante.de/tex-archive/macros/latex/doc/encguide.pdf is actually somewhat valid:

We recommend that new encoding names should not be introduced
unless careful consideration and discussion in the user community
has confirmed the need for the encoding. If encodings have to change
from font to font, a number of problems arise, so it is best to develop
encodings that can be used with a large number of fonts in parallel.
This allows documents to be typeset using different fonts without
problems.
The TS1 encoding is a good example of a bad encoding (even though
it was developed with the best intentions) as a huge number of fonts

Any pointers are greatly appreciated. I'll happily document what I did and why I did it, so others can use the same font in the future. I'd rather just do it by the book anyway. Need to find out how to, first.

(Note: I've tried a few encodings and none of them really cut it to use the font to its fullest extent.)

@1010011010 with classic tex normal way to make swash and small caps available is probably to offer them as a new font not a new encoding. Symbols are rather different and there you need an arbitrary encoding for the font but you can probably just use U as the latex encoding, and access the symbols by number \symbol{...}

@DavidCarlisle I have an OTF file containing 2000 glyphs. I need to split them up somehow, right?

Looking at the otftotfm manual I'd need to take a random name, and assign it to a glyph slot. To create a new font, I suppose I'd need to tell TeX how to do that translation ?

@1010011010 yes but similar thing to ways the CJK package provided fonts for classic tex, just took the unicode encoding and mechanically chopped it in to 256-character fonts so that you could automatically work out from the utf8 encoding of an input character what slot in what font it should be in. But just use xe/lua tex now.

@DavidCarlisle How can I do that chopping part, would I decide to go that route?

I know virtual fonts, but creating a virtual property list manually is madness.

Or is it? I don't know. It seemed madness.

@1010011010 no idea really, but there are tools around. But that is the way to go if you have a big font with essentially no real relation to a European latin alphabet. If it is mainly styled latin alphabet I wouldn't split it that way but instead a sa sequence of T1 encoded fonts, for each style

9:22 AM

@DavidCarlisle What do you mean by "style"? You mean a weight? Or a variant? This is just one weight. I have 9000+ glyphs across the family.

@1010011010 you said you had swash, small caps, ..

@1010011010 why don't you use xetex?

@DavidCarlisle For learning purposes, compilation speed, and to make it usable for any system. (Some people don't like or can't use XeTeX or LuaLaTeX, I don't know why either...)

@1010011010 I use pdflatex almost always too but I wouldn't use a 9000+ glyph font with it. You could make it work but it's a lot more work to set up, and the the amount of documentation you'll need to tell anyone where each of the characters is in one of the 30-40 fonts you'll need means that it will be much harder for anyone to use.

@DavidCarlisle It's just 6 font files, but yeah they're all pretty big when it comes to the glyph set

9:42 AM

@1010011010 with 6 fonts in classic tex that's just 1536 glyphs not 9000+

@DavidCarlisle 6 OTF files, I meant.

@1010011010 yes but you said you want to make it easier for people to use so you are using classic tex which means at most 256 characters per font.

@DavidCarlisle Yeah, I interpreted it the other way around :-)

@1010011010 so 9000 glyphs is 35 fonts if you pack them all, more if you arrange them in natural latin alphabet encodings leaving control character slots empty.

@DavidCarlisle Gets even more complicated since I have more than 250 ligatures... :D

9:49 AM

@1010011010 yes I was going to say, since you can not have automatic kerning/ligatures across different fonts that constrains how you chop up the font (you may have to lose some kerns and ligatures)

@DavidCarlisle kerns as well? why is that

@1010011010 just the way tex works. In the big font you may have kerns specified between a swash Q and a following lowercase normal letter, but if you expose the swash alphabet as a separate font to classic tex you can not have a font-specified kern between the swash letters and normal ones.

@1010011010 also of course you lose many hyphenation possibilities as hyphenation only applies to runs of characters in the same font.

10:08 AM

If I delegate the swash alfabet to be the math script font, I can just use small caps as a seperate font (using your U encoding to map the small caps letters in the regular abcdefg slots)... hmm

@1010011010 why math surely swash alphabets are only used in text?

@1010011010 if you make small caps a separate font why can't you use T1 encoding?

@DavidCarlisle Okay so I have one font which contains both regular glyphs abcdefg.... as well as small caps variants.

@1010011010 yes but just expose that to tex as two fonts, both T1 encoded

@DavidCarlisle How would I do that ? Doesn't the conversion to tex font metrics translate from glyph name to glyph slot?

@1010011010 depends what you use but in any case you need to map each glyph to a slot in the range 0-255 so it makes sense to send a and small caps a both to slot 97 in different fonts.

10:16 AM

@DavidCarlisle Yes, that's exactly what I meant. And I thought encoding was the way to do it... but I'm not sure now... :-)

As in, make my own ec file (same as t1) and just prefix all the alphabetic characters with small caps, like you said

10:38 AM

I'm still confused. So there's Type 1 encoding. How does that relate to the used .ec file? Is it just a matter of where TeX "expects" to find some specific glyphs, leaving the rest up to the user?

10:54 AM

@1010011010 type 1 isn't an encoding it is a file format?

@DavidCarlisle See.. I have no idea what I'm talking about.. :-)

I meant T1

Oh, that's called Cork encoding.

ooh e-mail from Bruno!

@PauloCereda yes

11:41 AM

Is the following a simple search for placing unnumbered chapters to the toc?

0

Q: How do I make the first Part and Section in my ToC appear without a number to their right?

In the preamble I have: \documentclass[11pt,a4paper]{book} \setcounter{secnumdepth}{4} \setcounter{tocdepth}{4} \setcounter{part}{-1} \setcounter{subsection}{0} Which works out great for the Part of the book, but not the sections which follow, which now appear as: Part 0.1 Chapter 0.2 Ch...

table-of-contents

11:54 AM

@TorbjørnT. I need your help on a polar plot, if you can.

@Alenanno You can always ask, though I'm no expert.

@TorbjørnT. In this answer, there is an artifact. I don't know how to remove it. If you divide the graph mentally with a cross in 4 sections, then look at the curves in the top-right section. There is a double line there. If you set the plot to have a thin line width, it's more visible.

I don't know why it's coming up

maybe the domain is too much?

Ah yes, it's the domain.

Sorry for bothering you. :D

@Alenanno No worries. I couldn't really see what you mean anyway.

@TorbjørnT. Well basically the plot is drawn using the same method you would use to write 8 with a singular line. If the domain is too big, you'll keep redoing the same path, thus a part of it will superimpose the other, so to speak.

@TorbjørnT. Sorry for the trouble though. :)

12:40 PM

Either duplicate of where to put own sty-files or just unclear

0

Q: How do I get MikTeX 2.9 to use a class file which I created and used successfully with MikTeX 2.1

The class was created some years ago. It worked correctly with MiKTeX until I upgraded to 2.9 I have created a TDS root and located my personal .sty files in this. They work successfully. I assumed that MiKTeX would similarly locate and use my personal class files. So I copied them into the TDS ...

packages miktex

Who knows

0

Q: Using greek letters give error on package installation

I need to use Greek letters in my documents. I've tried several things, but in each case TeX tries to install the file fonts\tfm\public\cbgreek\grmn1000.tfm When I select 'install', pdfTeX uses more than 50% of my CPU and then nothing happens.

pdftex packages greek

Solved by OP by using part instead of chapter

0

Q: Creating dedicated page for chapters while being present in TOC

I am currently working on the layout of my thesis and not completely certain how to implement this feature. There are two things i wish to achieve: A dedicated page for each chapter Chapter names being visible in TOC The TOC structure im looking for is: Chapter 1: Some chapter name 1 Som...

table-of-contents chapters

1:22 PM

@cfr Could it be beneficial to put up a chat? You seem to know more than can be contained in a few comments about this fonts business.

looks like Frank broke one of my packages:( (or at least I think it was Frank:-) latex-project.org/cgi-bin/ltxbugs2html?pr=tools/…

1:46 PM

@DavidCarlisle Are you sure you didn't break it yourself?

@egreg not totally, the I dug out the old cvs record but it seems to start just after that...

@DavidCarlisle I don't mind: using double rules in tables is even worse than \left-\right.

@egreg but you have to like the hhline/colortbl double coloured rules?

@DavidCarlisle The only double color I like is “black and white”.

@egreg so boring and conventional. Not even "green and white" ?

1:51 PM

@DavidCarlisle Hint: it's not Newcastle FC

@egreg Not Barcelona either?

@DavidCarlisle :P

2:04 PM

@1010011010 Sure. I'm going to be doing something else in a minute but I can come back later. I'm not an expert. I know a bit.

@cfr It seems to be fixed. I'm still confused how this happened. But I'm glad it did.:-)

I'll post an answer with a couple of things I did for the public to see, maybe it's useful to somebody someday.

2:31 PM

@cfr I just gave you the tick. The answer did not lead to fixing the issue directly, but the issue was resolved anyway and your links were actually quite useful :-)

2:42 PM

@egreg Green and white: Palmeiras. :)

@PauloCereda Which could be good as well.

@egreg <3

2:53 PM

@PauloCereda Today I did a new mountain road. Easy, but very nice, up in the forest where Venice took the trees for ships and buildings.

Cansiglio (Canséi or Canséjo in Venetian language) is a plateau in the northern-Italian Prealps, included in the provinces of Belluno, Treviso and Pordenone. Cansiglio is home to a very small Language island of Cimbrian. == Geography == The plateau rises immediately above the plain below, at more than 1,000 m above the sea level. It is in fact formed by a basin surrounded by rocky peaks, such as Monte Costa, the Cima Valsotta, Monte Millifret, Pizzoc and the Monte Cavallo; these mountains separate the Cansiglio from the short Val Lapisina valley and from Piancavallo. The plateau features several...

Wow, it looks awesome!

user image

@DavidCarlisle No stripes. :(

@egreg there were some in the version posted last night I covered them up:-)

3:13 PM

@StefanKottwitz Sorry didn't see your ping until now. Sure, any comment that you feel is too chatty and doesn't add anything, feel free to delete.

As long as the remaining ones still make sense. :D

3:45 PM

@DavidCarlisle GOAL

4:16 PM

@1010011010 Glad it worked. Note that you shouldn't be encoding æ Æ etc. as ligatures. I've put this in a comment, too. They need to be encoded as single letters.

Christian Hupfer

@cfr: Hello

4:29 PM

@DavidCarlisle Looking at the very top, the email address is hidden. In your response, the identity and email-address is revealed. Should be hidden there as well?

Christian Hupfer

user image

A nice white number on green background ;-)

@ChristianHupfer You're 111 short! ;-)

@cfr Yeah, I know. I just couldn't help myself but to add in more ligatures since the font contains so many.

Christian Hupfer

@1010011010: Well, optaining 111 rep will be hard ;-)

@ChristianHupfer just because the OP has not been around does not mean they don't care.

Christian Hupfer

@PaulGessler: Maybe, but he wasn't even registered at all

4:49 PM

@Johannes_B grr been caught out by that before, stupid system:-0 will fix, thanks

@Johannes_B fixed I think, thanks

@DavidCarlisle Just noticed it and thought i drop a note. :-)

what is tex?

@Yokhen a typesetting system, we must have an answer on site somewhere...

6

Q: What is TeX used for?

I'm a student right now and I'm wondering if it would be beneficial for me to learn LaTeX as of right now and if it would have any use of it to me. I do quite a lot of writing but they don't require a lot of formatting. For my Graphic Design work, I usually use Pages (Mac) for simple briefs for ...

@DavidCarlisle That's beautiful

5:16 PM

user image

@egreg -- is this the kind of heron you saw on your boat trip yesterday? this gangly adolescent will grow up to be a great blue heron. (he looks like he needs to be fed a fish.)

5:30 PM

@barbarabeeton My first thought when I saw this image was that it was surely posted by @PauloCereda as a creepy duck :-)

I just wanted to tell everybody who is interested: I've got tag search http://write-math.com/search/?search=%5Barrow%5D+circle :-)
The search still needs much improvement, but it's a start

@moose -- nope, but you can surely see how birds (at least some of them) are descended from dinosaurs.

@moose oy :)

5:45 PM

@moose did you say you thought that was @PauloCereda?

In relation to yesterdays event, does anybody understand the following question and can provide an answer?

0

Q: Numbering of head title on titlepage?

I want to combine several independent papers in a pdf-format. However, due to time constraints, I am looking for a "quick and dirty" solution. Therefore, I manually constructed a table of contents and adjusted the page numbers on my papers with the command \setcounter. In order to be consistent ...

horizontal-alignment numbering titles

@DavidCarlisle hey!

@barbarabeeton They were more elegant. ;-)

@PauloCereda so hard to tell you birds apart.

@DavidCarlisle quaaack. :)

5:53 PM

@DavidCarlisle :-) I needed Paulos "hey!" to understand it

@moose just assume any comments are insults, you'll be right often enough.

@DavidCarlisle :)

Christian Hupfer

@Johannes_B: Apparently it's a collection of papers, included as .pdf and \refstepcounter for it, then packaging it up to a LoP (list of papers) ... Shall I try an answer?

@ChristianHupfer Sure, better than nothing. I thought of something similar.

@barbarabeeton Thanks for having a look at the footnotemark stuff for amsart.

Christian Hupfer

@Johannes_B: Ok, I'll try later on

6:14 PM

Hello again. Some time ago I asked about why Knuth chose the spacing that he chose around \mathop{..}. And, after thinking a little bit more, I think I'm still not understandind. Sorry for being a little reincident.

I (could) understand that he didn't want to add any other type of atom, so he put \sin, \exp, etc. in the same place as \sum, \int, \etc. So I can kinda understand that after the \mathop{..} he automatically adds a thin space when there's no \mathopn{..} (which, IMO, doesn't look good when, e.g., \int a).

But why add a thin space before when, for instance, a unary minus is there?

For example, f(x) = -\sin x there's a thin space between the minus and \sin. Why a thin space there? Any relevant place where it's needed?

Extending that to the (may be mis)use of \mathop{..} also with \int, the fact that M = -\int a looks bad on both sides is what I don't understand (I could understand it looking bad only on the right). For me it “should” (without enough knowledge) be M = -\mathopen{}\int\mathopen{}a.

Anyone thinks something similar? Or has ever thought about this?

@Manuel I see no reason for omitting the thin space in case of -\sin x

@egreg Why? Any paralelism with other situations?

@ChristianHupfer Hello.

@cfr: Quaaaaack!

@1010011010 But these are NOT ligatures.

6:27 PM

user image

@PauloCereda Cwaaaaac! (Welsh-speaking hwyaden.)

(We don't have Q or K.)

@cfr yaaay! :)

6:45 PM

@Manuel -- when tex was created, memory was at a premium, so everything that could be packed into 8 bits (or a multiple) was so packed. although there are 13 atom types, only 8 enter into the spacing calculations (table, texbook, p.170). i couldn't find any examples of an operator with a unary minus in the texbook, so there isn't any evidence to go on whether this particular spacing combination was considered, and i don't have copies of the art of computer programming to check.

however, considering how careful i know knuth to be, i don't believe he simply ignored the situation.

@barbarabeeton The 8 atom types entering the calculations is an acceptable excuse for using \mathop for \int and such, yep :) The thing that I'm curious is the case of the unary minus (in the middle in the image). It's a choice (just say “Hey tex, between a unary \mathbin and a \mathop add a thin space”, so I believe it was a choice (not ignored), but I was just looking at a good example where it's necessary.

Ahm… or may be not. When a \mathbin is “unary” is it converted to a \mathord? In that case it might well be that \mathbin{-}\mathop{\rm sin} gets converted, since it's unary, to \mathord{-}\mathop{\rm sin} so the thin space is the automated space between an ordinary atom and a mathop.

In this last case (which I'm not sure if that's the behaviour: converting unary \mathbin to \mathord)… both spacings are due to 1900ish memory limits. Both, IMO, are wrong, although I'd have to admit I'm not Knuth nor experienced in this area. I hope, at least with LuaTeX, we get the power to change this kind of behaviours in a feasible way (without recompiling the tex program).

7:48 PM

Hello

@ℝaphink Ciao!

Come stai oggie, @egreg

?

Ciao

oggi

@ℝaphink Molto bene! Très bien! Very well! I made two very nice tours yesterday and today.

7:50 PM

nice :-)

I'm wondering what to call a quote I'd put on the side of a page

hmm, it seems I might want to use wrapfig

@ℝaphink And got rep capped nonetheless. :)

as per @Werner 's answer on tex.stackexchange.com/questions/52180/…

@egreg

@ℝaphink Oh! You're looking for troubles! :)

@egreg you don't need to post answers anymore, you'll get rep capped everyday

@egreg?

why?

I used @werner's code and ended up with something like this… i.imgur.com/D4B4OdT.png

@ℝaphink Just joking.

7:57 PM

hehe

hmmm, reading the doc definitely helps :-)

@cfr Always thinking about this:

Karl Pilkington: The Future [HD]

Martin Schröder

@PauloCereda regarding the blog entry: The EuroTeX was in 2005, not 2006. Please fix the image caption.

8:14 PM

@MartinSchröder Done, thank you. :)

@Manuel I think I prefer the lower setting for -sin x, for the integral it's debatable either way, but in that case it is very dependent on the font a more upright integral would probably want different spacing anyway,

8:53 PM

@Johannes_B Aaaaaaaaaarrrrrrrrgggggggggghhhhhhhh! We have more letters! ;)

9:04 PM

@cfr 1114111 of them:-)

9:19 PM

I've got a grant for my journey to Darmstadt! Yay!

9:44 PM

@DavidCarlisle You might prefer it, and even me (yeah, I'm undecided :P). But it's a lucky situation, I think it wasn't explicitly defined for the unary operators, it just end up like any other being a \mathord hence spaced by a thin space (even in \scriptstyle which, IMO, definitely shouldn't).

In case of the integral, well, the spacing should be a font thing, not an atom spacing.

I don't know myself, I'm in doubt, that's why I'm seeking for opinions. After all these years of TeX and yet no one thought about this? (I would expect someone to be worried about it, the same with \left( .. \right) inner spacing… which, for instance, motivated the creation of mleftright package.)

@Manuel yes but it follows from the general principle that a mathop binds closer to the argument on the right hence the space to its left. -sin x is -(sin x) not (-sin)x so having sin have a small space before

@Manuel the horizontal space from left right is a lot easier target for complaint it ii wrong more often than it is right:-)

@DavidCarlisle Someone could say that the -\int x is wrong more often than it's right. But yes, I don't know the floor I'm stepping in; I just wanted to activate your minds :)

@DavidCarlisle And about that, you could think of \sin x a shortcut of \sin(x) so with -\sin(x) the thin space looks unnecessary.

@Manuel There is no symmetry: the minus applies to all the expression \sin x or \sin(x).

10:00 PM

@egreg I don't understand. Does that imply that the thin space is better? I think, at least to me, that the no-thin-space version would be equally acceptable if there wasn't because of all these years with the thins pace there :)

@Manuel yes the thin space is better in that case.

@Manuel Yes, I'm implying that.

@cfr :-D

@Manuel not necessary but the space is an improvement as sin binds closer to the (x) than the -

It's probable that that's the right answer.

I just remembered from my first times with TeX… that people told that \int was always away from the contents (that's what the teacher told me the first time I met TeX). And just today, knowing a littlbe bit about how TeX spaces math (with atoms) I understood the reason behind that :)

10:48 PM

@DavidCarlisle ??

« first day (1680 days earlier) ← previous day next day → last day (3262 days later) »

Transcript for

Jun '157

$TeX - LaTeX$ TeX, LaTeX and Friends

\begin{chat}

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

$TeX - LaTeX$

site design / logo © 2024 Stack Exchange Inc; legal

mobile