@NikeDattani both overleaf and arxiv uses real LaTeX, mathjax does not. It seems strange that that page does not render align correctly, but it is up to the mathjax configuration through https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/MathJax.js?config=TeX-AMS_HTML-full , I have a webapplication that uses the same setup but the TeX-MML-AM_CHTML config, there align works just fine. As others have mentioned, this is not a LaTeX issue, it is an issue with mathjax.
@UlrikeFischer @MarcelKrüger what is this mapping that only xetex can do? "We all know that Mapping=xxxxnumerals is only supported with XeLaTeX, thus I can't use polyglossia with LuaLaTeX." at gitlab.com/niranjanvikastambe/marathi/-/issues/2
@DavidCarlisle that uses a teckit file which contains instructions to replace chars, it is e.g. used to get the tex ligatures -- to endash, but it can also be used to implement transcriptions. It is quite xelatex specific, but you can naturally implement most things in luatex with lua-code (not quite sure how easy harfbuzz + transcriptions would be, @MarcelKrüger?).
@UlrikeFischer yes sure I assumed teckit was involved but for any specific teckit mapping it's usually just a simple search replace and doable in lua for the ones I have seen (just hard to do a single one off support for all mappings at once
I created a .map file for Cyrillic to Latin transliteration as in using xetex for automatic transliteration of Cyrillic letters
I placed the cyrillic-to-latin.tec file in the folder.
\documentclass{article}
\usepackage{fontspec}
\usepackage{ifluatex,ifxetex}
\setmainfont[Mapping=cyrillic-to-la...
@DavidCarlisle I don't really know when exactly the teckit mapping kicks in, e.g. regarding hyphenation. But even with xelatex mappings are difficult, as it can only use one. So you need to write "combined mapping" if you want more features.
@DavidCarlisle that's a plan. But it is a nuisance that language commands are not unified and that there are two systems. I have to do some shopping now if I want to be back in time for the meeting ...
@UlrikeFischer Teckit applies quite early (basically when XeTeX first builds it's native text nodes, that is before hyphenation). This e.g. means that teckit implemented ligatures can be broken using grace groups. For most practical purposes it would be compatible to do such mapping the beginning of the hyphenation callback, but in this devanagari file it doesn't matter as long as it is done before shaping.
@UlrikeFischer In general such mappings with HarfBuzz can be tricky, but in this case it are just one-to-one mappings, so they can be implemented without such problems. (They would work the same as the trep in luaotfload-szss.lua)
@DavidCarlisle The biggest issue I found for supporting all mappings at once happened to be normalization: Mappings can specify that they expect their input to be normalized, which in LuaTeX would require a normalization implementation which is fast and keeps track of node to character mappings.
@DavidCarlisle Yes, in this case that wouldn't be a problem, I was only referring to a potential more general solution. I could implement their replacements if they need it, but I think just generally switching to babel is the better solution.
@daleif One of the overleaf developers said in the TUG meeting last weekend that they explicitly pull the system fonts into the docker image to make more fonts available, so they should have access to them (or at least can make them available if asked nicely)
@Skillmonlikestopanswers.xyz The issue in that screenshot isn't really Acrobat's fault though. For Latin Modern, Adobe's problem mainly is that they actually implement Type 1 hinting correctly and Latin Modern looks best if you ignore hinting instructions completly. E.g. in this case the cap-height isn't aligned. Normally you are supposed to use BlueValues to indicate the range of your cap-height such that the hinting system can render them consistently.
In Latin Modern, some characters like "N", "C" and "R" have a height in this declared range (you can see that they are rendered consistently), while some characters like "E" and "F" are slightly short of reaching this zone, so they are rendered shorter. If the declared zone is adjusted downwards by just three units, this issue is avoided. Type 1 even has a feature to deal with such inaccuracies, but Latin Modern explicitly disables this by setting BlueFuzz to 0. 🤦
@Skillmonlikestopanswers.xyz I didn't mean to imply that Acrobat is particularly good or bad at rendering fonts (as a Linux user I barely use Acrobat and mostly don't even know how things look there), I only wanted to say that specifically for Latin Modern, any implementation which attempts hinting will have a hard time.
@AlanMunn -- I've been in touch with someone who might be able to do something "officially" about the font problem reported in tex.stackexchange.com/q/555446, but one necessary piece of information isn't clear: was the proposal submitted as a LaTeX file or pdf? Please ask the OP to add that information to the question; if it was a pdf file, I believe there's a strong case for pressure to be applied.
@AlanMunn -- I made that assumption, but the information must appear in the question so that the people who are looking at this have no "out" for ignoring it. (I'll be happy to tell you what I've done by email, but not in public.)
@barbarabeeton I'd be happy to hear it. I've taken the liberty of editing the question to make it clear that PDF is what's being uploaded.
@barbarabeeton I don't think that waiting for the OP to respond is necessarily going to be fruitful, and the system simply doesn't accept LaTeX source so I'm comfortable making the PDF assertion.
@PauloCereda Fedora is also a nice distro. I kind of remember using it at some point, but I can't remember where. My main system went from Debian (stable... don't do that) over Gentoo to Arch, so I never got any experience with the RPM based distros.
@MarcelKrüger ooh :) I've been using an RPM-based distro since Red Hat 8, then it became Fedora Core 1. And here we are with F32. :) I also used Slackware for a long time in my lab machine. People were puzzled to see a bloke with a huge monitor with a black screen and a tiny cursor. I was happy in init 3 and only a few times I had to startx. :)
@FaheemMitha Sounds reasonable. On my personal device I went ubuntu->fedora (+ a couple of different one for specific purposes as virtual machines) and at work scientific linux->ubuntu
@FaheemMitha Thanks for the link - interesting to read
I have a document on Overleaf that marks an error saying that a } is missing but it's not and then it says that I should type 2 and the error will go away. Where should I type 2 so the error goes away?
I already typed '2' before and after the line and it doesn't go away
@AlanMunn actually I doubt that it has to do with a missed }. The thing is that I have \draw[thick,->] and when I remove >, the error goes away; and when I include the >, the error of a "missing } is there again"
@América next time you should show the complete, exact error message ! Argument of \language@active@arg> has an extra } is much more understandable than your rewording.
I have values
1 2 2 3 3 4 4 4 4 5 5 5 6 6
and want that to plot as a line sharp plot histogram.
So, I set:
hist={
handler/.style={sharp plot},
intervals=false,
},
But he does not want to show the 2 times of the number 6...
Then there is a strangeness: If I do not set x filter/.expression={x==int...
@DavidCarlisle @UlrikeFischer Could you take a look at github.com/latex3/latex2e/pull/361? Especially: Do you think that I should add \changes? They seem completely useless, but we normally add them pretty consistently, so I'm not sure.
@PhelypeOleinik I'll leave it like this and change it if someone complains. Any situation where this would make a difference probably breaks a lot of other stuff anyway.
I created a .map file for Cyrillic to Latin transliteration as in using xetex for automatic transliteration of Cyrillic letters
I placed the cyrillic-to-latin.tec file in the folder.
\documentclass{article}
\usepackage{fontspec}
\usepackage{ifluatex,ifxetex}
\setmainfont[Mapping=cyrillic-to-la...
