TeX, LaTeX and Friends

cfr

01:48

@ApoorvPotnis (@DavidCarlisle) probably irrelevant, but I didn't read everything. do note that some of LM's .afms are themselves buggy in ways the .tfms are not. that is, LM's metrics are not even consistent with LM's metrics. (I don't even know what 'advanced widths' are, though.)

@ApoorvPotnis if you just want to see what it looks like in pdftex, you can use e.g. autoinst with default options to convert. you won't get fantastic results, but it will give you an idea. I wouldn't bother, personally, since you can see what the font looks like using luatex, but you could.

CroCo

02:27

Hello everyone, I'm looking for a modern solution to embed video in Beamer that doesn't depend on Flash. Your insights would be greatly appreciated!

Apoorv Potnis

06:49

@DavidCarlisle i don't know how to generate the tfms. I found out that there's a tool otftotfm, but don't really know how good the conversion is.

i.sstatic.net/pBP7LMfg.png i don't know if this means that the book weight is mechanically thickened. i will have to ask Prof. Tsolomitis

@cfr i see. i wasn't aware of this info.

@cfr i should give autoinst a try

i have one more idea in mind, but don't really know how practical it is. find the function which simulates the spreading of ink on paper, and apply that function to all the glyphs, using latex

mickep

@ApoorvPotnis I don't know if you like that, but it should be possible to "bolden" by also drawing the outlines in some width.

Apoorv Potnis

@mickep this looks good. but i found out that the result is highly dependent on the pdf reader

mickep

@ApoorvPotnis It indeed is.

cfr

07:29

@ApoorvPotnis you certainly cannot do this on the simple latex side. and I doubt there is a good way to do it in with pdftex. as far as tex itself is concerned (i.e. tex or etex as opposed to pdftex etc.) glyphs don't exist. so fake bold is something like printing a character twice, for example. which is why the results kind of suck. so the kind of fiddling @mickep is talking about has to happen on the rendering side, more-or-less. of course, things are different with luatex.

@ApoorvPotnis I wonder which metrics they used. hopefully not the afm ones.

Apoorv Potnis

@cfr indeed using modern engines.

@cfr if I understand correctly, then tfm or afm files were not used at all. Newcm is based on the OpenType version of Latin modern

Udi Fogiel

08:47

Hello. I'm getting different hyphenation patterns after a TeX Live update in a github action while I did not change anything related to it. In my local installation (which i probably did not update a couple of weeks) I don't see any difference. These are patterns for the Romansh language.

Does any one know from where the difference can come from?

Skillmon

@UdiFogiel there was a biggish update of language files earlier this year, most likely that's the reason.

Udi Fogiel

@Skillmon Oh Ok, Thanks! I found it now thanks to you: github.com/hyphenation/tex-hyphen/commit/…

Rmano

4

100 to go....

(and 300 bronze and 370 silver badges) 😜

cfr

@ApoorvPotnis I'm assuming the otf and type1 LM are all from the same source files. hopefully the afms are the odd ones out and not the tfms. (normally, you'd make the tfms from the afms, which is why it is a bit odd.)

@DavidCarlisle curious whether this is what you meant.

@ApoorvPotnis it depends how good your instructions are partly. or you can use fontforge.

David Carlisle

10:28

@cfr (@ApoorvPotnis) you can (with very mixed results) access the outline of a type1 font using postscript operators see for example the contour package so fake bold is possible with latex/dvips and type1 fonts (much easier with opentype eg FakeBold option in fontspec)

David Carlisle

10:44

@Rmano 400,100 more would be better still:-)

Rmano

12:13

@DavidCarlisle I'll go for that after the 370 silver badge!

Udi Fogiel

I worked on a port of ConTeXt's unicode bidi algorithm for LaTeX OpTeX and plain. Since the names bidi, luabidi and unicode-bidi are already taken, I thought about naming the package uax9, but CTAN does not really like that one.

Any suggestions?

David Carlisle

@cfr I just meant there are existing tools to get from opentype to some set of type1+tfm and I wondered what @ApoorvPotnis was using, as I'd expect that the tools would make tfms as part of the process.

@UdiFogiel what does it do differently? (coping with bidi package and babel is already stressful enough:-)

you could call it snahemalb.sty

Udi Fogiel

12:30

It support plain and OpTeX (which I use often). bidi does not work with LuaTeX, and babel's implementation, while working with complex structures in LaTeX, fails in simple cases which annoys me: https://github.com/latex3/babel/issues/285, Ive tried to port Babel's code to different formats but it was harder as the code is somewhat convoluted between the Lua and TeX parts.

Babel already has three modes for LuaTeX, bidi=basic, bidi=default and bidi=basic-r, having another one wouldn't hurt ;)

I will probably use this package for polyglossia as well.

David Carlisle

@UdiFogiel actually that's what I mean, another one does hurt. when trying to make tagging work (or even tables with the columns in an expected order) things would be much simpler if there were not competing implementations that you had to try to support.

Udi Fogiel

@DavidCarlisle What does that mean?

@DavidCarlisle This is almost a pure Lua module, no patching to anything is done, it won't affect tagging.

Skillmon

@Rmano I'm not sure whether you'll manage all those badges in just 100 rep...

Rmano

I can always start giving bounties...

It's more difficult t get the rep without any tick...

David Carlisle

@UdiFogiel are you sure? Does it (for example) cause table columns to get set right-to-left, that almost always affects tagging (or even non tagging) code. Or is this just about character level direction?

Rmano

12:37

(well, not literally, no)

David Carlisle

@UdiFogiel I mean longtable, multicol, ... get regular bug reports that things go wrong if rtl is implemented by bidi or babel or ... and we try to adjust so it does the right thing, but then documents come along with rtl implemented completely differently and we have to try to accomodate that without breaking the existing. The current state is "complicated"

mickep

@DavidCarlisle Does not sound fair to me.

David Carlisle

@UdiFogiel but despite that (as with luaotfload) pulling context code out to be usable in other formats is a good thing so thanks for that, I just hope that somehow we can get to a point that latex has a single language api and a single direction control

Udi Fogiel

This is just about character level direction.
The code is here: https://github.com/Udi-Fogiel/UAX9-LuaTeX you can experiment.
I did some experiments with tables but it is hard from the Lua side. https://github.com/latex3/babel/issues/221#issuecomment-2177442828

controlling the direction of the columns from the TeX end is easy, at least for boxed tabular, if the vertical box is right to left, the columns are right to left and vice versa.

David Carlisle

@mickep life isn't fair

mickep

12:43

@DavidCarlisle You can start xlatex, running context under the hood.

David Carlisle

@mickep we could just tell everyone to use context

mickep

@DavidCarlisle See, so many possibilities. That would also save @UdiFogiel some work.

Udi Fogiel

If you ask me, probably LaTeX should add some support for RTL, with luatex it is not that hard. It gets hard when external packages tries to patch the internal of LaTeX.

As for packages not maintained by the LaTeX team, their maintainer can add support if they are willing to do so.

mickep

@DavidCarlisle Is that how you got the rabbit into trying-out mode?

Udi Fogiel

@mickep Sorry, I'm an OpTeX kind of guy :) When I try to use ConTeXt I always feel like I started too late to keep up.

mickep

12:47

@UdiFogiel OK, so let us blame @DavidCarlisle for the bad idea :)

Udi Fogiel

@mickep Sorry, clicked on the wrong button...

David Carlisle

@UdiFogiel babel is latex/required/babel so basically by definition that is the "standard" rtl support but in practice bidi was there first and polyglossia does stuff and xstex and luatex have different mechanisms and stuff is confusing:-)

mickep

@UdiFogiel No problems. David is reading everything here, so we do not need to ping him.

@DavidCarlisle But now you need not to care too much about xstex or xetex or what it was called. ;)

Udi Fogiel

Polyglossia simply loads bidi, it does not patch anything itself...
Babel is a part of LaTeX, but yet it needs to patch tabulars if I'm not mistaken.

Try for example
```
\documentclass{article}
\makeatletter
\def\@array [#1]#2{\if #1t\vtop \else \if #1b\vbox \else \vcenter \fi \fi \bgroup \textdirection=1
\setbox\@arstrutbox \hbox {\vrule \@height \arraystretch \ht \strutbox \@depth \arraystretch
\dp \strutbox \@width \z@ }\@mkpream {#2}\edef \@preamble {\ialign \noexpand
\@halignto \bgroup \@arstrut \@preamble \tabskip \z@skip \cr }\let \@startpbox

Ulrike Fischer

@UdiFogiel well hbox is a primitive and it shouldn't appear in a LaTeX document directly. You should use \mbox and then one could discuss macro support. (such a \hbox is a problem for tagging too).

Udi Fogiel

13:01

@UlrikeFischer The example in there fail with an \mbox as well :(

In any case, Babel's approach is to use \pagedirection and \bodydirection which are not really documented, their source code is hard to understand, they are not included in the LuaMetaTeX engine, and after trying to use them myself for over a year in OpTeX I think they are kind of buggy (or at least I don't understand them).

Ulrike Fischer

@UdiFogiel I suspected that, that's why I said that one could then discuss macro support.

Udi Fogiel

@DavidCarlisle What is the meaning of that name?

mickep

@UdiFogiel read backwards...

David Carlisle

@UdiFogiel yes everyone patches everyone, but we hope to somehow break out of that.

Ulrike Fischer

@UdiFogiel Oh I'm quite sure that there is place for improvement in babel. Various patches from babel should move into the kernel and there should be more hooks and whatever. But with Javier more or less working alone on it and the rest of the team working on tagging, all that takes time. But I don't think that it helps if there is another system that doesn't target integratation into the kernel + babel. Btw: the @array definition looks different in array.sty.

Udi Fogiel

13:31

I mainly ported the code to be used with OpTeX, and the only requirement is support for a subset of luaotfload, which LaTeX has in the kernel, so I thought why not add a LaTeX package as well, it only took another couple of lines of code, and it works in some cases where babel fails (and vice versa...).

javier also said that he called the current algorithm "basic", as it is a basic implementation. Maybe this code will be an opportunity to revise the current code.

As for systems that do not integrate into the kernel, sometimes there are disagreements as in https://github.com/latex3/latex2e…

If you want, to avoid problems, I'll add to the readme that while the package can be used with LaTeX, it dos not have official support, and point to Babel.

David Carlisle

@UdiFogiel I was probably a bit harsh to be honest, just that the first reaction to "another ... package" is "oh no more support complications" not "oh good more features". But as you say first get it working, then we can see what's the best way to merge things.

Udi Fogiel

@DavidCarlisle well, I need a name for that :)

David Carlisle

13:46

@UdiFogiel But complications with column order are not so much with the physical re-arrangement, that is easy, the hard part is agreeing on the interface, eg does >{x}c<{y} insert x on the left or or at the start of the writing direction? similarly does a column spec of llc mean the first two columns on the left are left aligned, or the first two columns in the block direction are aligned to the start of the writing direction? there are various possible answers..

@UdiFogiel @mickep didn't like my suggested name

Udi Fogiel

@DavidCarlisle That's why adding a hook will be the best solution, isn't it? if a user define as one, it will be permanently from right to left, if zero than permanently from left to right, the definition can be dependent on the current language, or other parameters... As for "Does left should mean left or start of writing direction", I already said my opinion on that.

@DavidCarlisle I still think I'm missing something... did not find any reference to the name in older messages (how older should I look for?)

David Carlisle

@UdiFogiel sorry it is blame hans written backwards

Udi Fogiel

@DavidCarlisle Oh! that went over my head...

David Carlisle

@UdiFogiel :-)

yo'

Anyone feels like a dinosaur? My alma mater (not my department at least) is recommending a LaTeX book from 2002. In Czech, so probably still talking about cslatex.)

Apoorv Potnis

13:58

@cfr yeah, i suppose that both the otf and type1 lm are from the same sources. newcm-regular has the same glyphs as lm (otf), and similar metrics. newcm-book has the same metrics as well. that's why my claim that newcm-book would work with lm metrics (tfm ones. i wasn' aware that the afm ones have problems)

@DavidCarlisle i see. i think i had tried it very long ago, but didn't get satisfactory results

mickep

@DavidCarlisle It felt a bit unfair.

David Carlisle

@mickep by the standards of this chat room, suggesting a package name seems fair compared to suggesting that people should be eaten. Clearly divademalb would be an inappropriate name though.

mickep

@DavidCarlisle A bit difficult to read, indeed.

David Carlisle

@ApoorvPotnis similar to FakeBold in fontspec, the results are rather variable, the outlines are not intended to be stroked, just form the fill region so if you stroke them, even if the pdf renderer handles that well, you can lose definition in the corners depending on how the shapes work for each glyph. In the original metafont sources the model is essentially stroking with a pen, and so making fonts bolder with slightly wider strokes is a far more natural fit, but stroked fonts went away,

Apoorv Potnis

15:57

@DavidCarlisle I see.

For some reason, CM/LM looks okay on Macs/Apple devices, but not on Windows or Linux. I don't really know if there's any standard of how pdf readers are supposed to render text. The end result of how the eyes actually see the text seems to depend on the OS and the reader

David Carlisle

@ApoorvPotnis well it's always been that way. People generally were quite happy with pk bitmap tex fonts, they printed well and xdvi, dviview, and similar previewers rendered them well on screen anti-aliasing as needed. Then pdf arrived and in particular acrobat. It was clearly designed to highlight the benefits of scalable fonts the rendering of bitmap fonts was simply awful. A document that was perfectly readable in xdvi would convert to pdf that acrobat made more or less a grey fuzzy mess.

CroCo

16:33

Hello everyone,
I hope it's okay to repeat my comment again, but I'm really looking for some assistance. I'm searching for a modern way to embed video in Beamer without relying on Flash. I would truly appreciate any insights you might have!

I have a presentation this week, and I feel a bit stuck at the moment.

The solution in here is not working in Beamer.
https://tex.stackexchange.com/questions/516029/media9-is-becoming-obsolete-dec-2020-any-alternatives-for-embedding-video-audio/

David Carlisle

video in pdf is highly dependent on the pdf reader being used. It's much more portable to simply make the presentation link to an external video that opens up in its own player. But if you have any method that works on your system using the pdf reader you plan to use, you could in principle generate that in beamer, so beamer itself is not the issue

cfr

@DavidCarlisle sorry. apparently I only deleted half of that comment, so it failed to give any indication what it was originally about.

CroCo

@DavidCarlisle Thank you for your response. While I can open the video, switching from the presentation slides to the video feels awkward to me. I must inform the audience that I need to locate the video for this slide before open it. :<

even worse if I have many clips.

cfr

@CroCo can't you link to it? you don't need to locate it.

Ulrike Fischer

@CroCo add a link.

CroCo

16:46

Is this the only solution (i.e. adding the link) available for this problem?

David Carlisle

@CroCo yes

@CroCo or use a format for the presentation that includes video natively, such as html.

CroCo

How does the link work if the video is stored on my PC?

David Carlisle

@CroCo don't tell @JosephWright and @samcarter but there is life outside of beamer

CroCo

@DavidCarlisle I don't know html.

David Carlisle

@CroCo do you know pdf?

@CroCo that is, most people don't hand write either format.

cfr

16:49

@DavidCarlisle I'd rather hand write html than pdf, though.

Ulrike Fischer

@CroCo as David said, the problem is with the pdf viewer. You can embed videos but controlling if and how they are played is difficult. If there were a pdf that demonstrates something working well it would be easy to reproduces that with LaTeX but up-to-now I haven't seen one.

CroCo

@DavidCarlisle Unfortunately, I prepared the presentation but discovered the problem afterwards. Not sure if it is worth doing it again in PP (i.e., I believe you’re familiar with my attention) :).

cfr

@CroCo you can link to a file. I don't like doing this as it means the slides only work on my machine, but if that's not a concern.

CroCo

@DavidCarlisle at which level?

David Carlisle

@CroCo I mean do you need to know the internal structure of pdf to understand beamer? (given that you gave not knowing html as a reason for not using a html presentation system)

CroCo

16:53

@UlrikeFischer but LaTeX relies on the PDF viewer. If there are any issues with Adobe, it can affect the final outputs quite a bit.

cfr

@ApoorvPotnis how are you converting the fonts to type1? if you use the autoinst tools, you should be able to generate type1 and tfms. if you use something else, you may need separate processes. e.g. I usually do them separately if starting from otf.

Ulrike Fischer

@CroCo yes, as other problems with Adobe or other viewer can affect the handling of the output (e.g. screen reader have to rely on what the viewer passes on). Such is life.

David Carlisle

@CroCo this is the first time you have mentioned adobe. whether the video embedding is possible at all depends on the viewer, not on latex or beamer, so that's rather important to know which viewer you are using. (and if you are visiting somewhere and presenting on an unknown viewer don't even try to do this)

CroCo

Can you recommend a reliable alternative PDF viewer that enables easy video embedding?

David Carlisle

@CroCo no, I haven't embedded a video in pdf since the first demos of acrobat 1, which was er, a while ago:-)

CroCo

17:00

While I am familiar with how web browsers can view pdf files, it's frustrating that they frequently struggle to properly display slides in full-screen mode. This limitation can hinder presentations and reduce engagement.

@DavidCarlisle Adobe Reader is getting worse—it's heavy, slow, and keeps the PC's fan noisy.

I possess the professional version provided by my University. Not good at all.

David Carlisle

@CroCo If I put my browser in full screen mode this (for example) seems to work well enough, (there are various keystrokes, but just hit space to move through the presentation) w3.org/Talks/Tools/Slidy2

CroCo

I will give Chrome a chance and see if it performs better for my needs.

Thank you all for these valuable insights.

CroCo

17:26

@DavidCarlisle, @UlrikeFischer, @crf, thank you all. In Linxu, the Document Viewer enabled me to play a video through a link.

Far better than Adobe Reader.

In Chrome, the button is unclickable. I don't know why.

David Carlisle

@CroCo most pdf readers don't support video (including pdf readers in browsers that can display video)

egreg

@DavidCarlisle tex.stackexchange.com/q/735369/4427 🤪

David Carlisle

17:51

@egreg well I assumed when you posted it that was the idea:-)

samcarter

18:10

@CroCo You could compile your presentation as svg and use media4svg

David Carlisle

18:23

@samcarter I always found chalk and a blackboard worked

samcarter

@DavidCarlisle for mathy content, chalk and blackboard is definitely the best!

Skillmon

@samcarter it's essential that you use chalk and blackboard. How else should you show off your skill to draw dashed rules blazingly fast that you acquired through hours of hard training?

samcarter

@Skillmon :) much better than \draw[dashed] (0,0) -- (1,1); :P

mickep

18:45

@DavidCarlisle It is indeed the best.

Alan Munn

19:13

I hear that with the TikZ ban in the US, more people are moving to PSTricks. :)

2

David Carlisle

@AlanMunn fortunately the UK is free of such restrictions and can force everyone to use picture mode

mickep

20:18

@خالدحسني Hello!

D G

Any suggestion to make it more elegant is welcome:

\begin{align*}
\begin{multlined}[b]
(C-0):(K-273):(R-0):\\(F-32):(X-20)
\end{multlined}
&=
\begin{multlined}[t]
(100-0):(373-273):(80-0):\\(212-32):(220-20)
\end{multlined} \\
C:(K-273):R:(F-32):(X-20)
&= 100:100:80:180:200 \\
&= 5:5:4:9:10
\end{align*}

David Carlisle

@DG more elegant output or more elegant markup? for the output would need more details notably the text width as obviously it would look nicer without the line break but presumably it doesn't fit.

D G

more elegant output:

\documentclass[12pt]{article}
\usepackage[a5paper,hmargin=12mm,vmargin=12mm]{geometry}
\usepackage{mathtools}
\usepackage{fouriernc}
\begin{document}
\begin{align*}
\begin{multlined}[b]
(C-0):(K-273):(R-0):\\(F-32):(X-20)
\end{multlined}
&=
\begin{multlined}[t]
(100-0):(373-273):(80-0):\\(212-32):(220-20)
\end{multlined} \\
C:(K-273):R:(F-32):(X-20)
&= 100:100:80:180:200 \\
&= 5:5:4:9:10
\end{align*}
\end{document}

mickep

20:37

@DG How about something like that? ^

David Carlisle

@DG I don't know what it means, so hard to say but perhaps

\documentclass[12pt]{article}
\usepackage[a5paper,hmargin=12mm,vmargin=12mm]{geometry}
\usepackage{mathtools}
\usepackage{fouriernc}
\begin{document}
\begin{align*}
&(C-0):(K-273):(R-0):(F-32):(X-20)\\
&\quad{}=(100-0):(373-273):(80-0):(212-32):(220-20) \\
&C:(K-273):R:(F-32):(X-20)\\
&\quad{}= 100:100:80:180:200 \\
&\quad{}= 5:5:4:9:10
\end{align*}
\end{document}

D G

@DavidCarlisle I like this one. Thank you.

@mickep Thank you.

mickep

@DavidCarlisle I feel you got the tick...

David Carlisle

@mickep shame you are not @egreg, kind of cheating stealing ticks from a context user.

@mickep although our answers were broadly similar: "don't keep all the logical left hand side to the left of the ="

mickep

@DavidCarlisle Indeed. I'm a bit hesitant about aligning two equations on the = when there are more lines and nothing inbetween.

@DavidCarlisle Ha! Too easy.

خالد حسني

21:13

@mickep hi

mickep

@خالدحسني Any interesting news from the font world?

خالد حسني

@mickep Fonts is such a wide area, but I have been doing some work on math fonts lately (Noto Sans Math, IBM Plex Math, and Nagwa TK Math), thought that is probably old news.

I wrote also an article on math font development for Noto project, github.com/notofonts/math/blob/main/documentation/…

I hope it will help more people build more math fonts. It fiscusses some topic that is not discussed elsewhere AFAIK, like how to choose sensible value for the various math constants.

mickep

@خالدحسني Oh, so you were involved also in Plex. There were many nice improvements from the beta to the final version. It seems to me the font now works well.

@خالدحسني I did not hear about Nagwa TK Math before. Will look it up! It looks a bit "playful".

خالد حسني

@mickep Yes, I did the MATH table engineering for IBM Plex Math and some general math fonts consultation

mickep

21:28

@خالدحسني I saw that. Very good and clear pictures of what is there. Some of the parameters are a bit hard to understand. One is skewedFractionVerticalGap, where I have a feeling it is too big in many fonts. Given the interpretation you make (that I guess is the intended one).

خالد حسني

@mickep Nagwa TK is a companion for Adobe’s Tekton font, based on a Type1 TeX font that Nagwa acquired the rights for (I don't remember from which defunct TeX company), but the Arabic part is brand new.

@mickep The skewedFractionVerticalGap text was basically based on “reverse engineering” what MS implementation does, but probably not many fonts were tested against it.

mickep

@خالدحسني I downloaded it, will test. Thanks!

@خالدحسني I have a feeling the value should be negative. The upcoming figure is the text font fractions from stixtwo.

But in the stix two math font I see SkewedFractionVerticalGap set to 68.

خالد حسني

21:51

@mickep I’m away from my computer now, but I’ll check it later

mickep

@خالدحسني Sure, no hurries

Rmano

23:45

8

Yay

Transcript for

$TeX - LaTeX$ TeX, LaTeX and Friends