@AlanMunn The reason 'TeX new site theme is live' isn't showing up on the main page is because "questions will be pushed off the front page of normal sites once they reach a score of -4, and only once they reach a score of -8 on meta sites." - from meta.stackexchange.com/questions/117119/…

@Milo For meta that's a crazy idea. So I guess people need to vote it back up to -7.

@samcarter You should've named tex.meta.stackexchange.com/questions/7787/… Make TeX.SX great again!

@Milo It shows up on the Questions view, but of course we're now all hiding the menu that makes that easily accessible...

@AlanMunn Yup, ideally I would like to get rid of that left-hand menu... But, that means losing access to the 'Questions' button. So I've kept the left-hand menu.

@Milo You may enable keyboard shortcuts and then use g→q. Then you could get rid of the left-hand menu.

@TeXnician Ooh, nice! Thanks for this tip.

@Milo You're welcome.

@Skillmon Partially implemented - I kept the "look nice" as "great" sounds like a bit too ambitious.

@samcarter and might be misunderstood. Good choice to keep it.

@William well that's interesting, how did you manage that:-) (seems to do same here. I'll trace....)

To use a "status-bydesign" tag to justify bad design seems questionable to me tex.meta.stackexchange.com/a/7796/117050

@AlanMunn can you remind me where the settings for that are?

@WillRobertson ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

@DavidCarlisle Go to your profile and Edit settings.

@AlanMunn wow that's too complicated :-)

! Interruption.
\__file_get_full_name_search:nN ..._full_name_str
                                                  \ior_if_eof:NF \g__file_in...
l.12     T
          he vorticity $ω$ is defined as $ω = ∇ × u$. $传$ 传
?
! Interruption.
\__file_get_full_name_search:nN ..._full_name_str
                                                  \ior_if_eof:NF \g__file_in...
l.12     T
          he vorticity $ω$ is defined as $ω = ∇ × u$. $传$ 传
?
! Interruption.
\__file_get_full_name_search:nN ..._full_name_str

@DavidCarlisle Yeah, I know. Who'd have thought it was done that way.

@DavidCarlisle BTW I think your ^^^^ and code don't line up very well...

@William @WillRobertson \__file_get_full_name_search:nN seems a little confused

@AlanMunn the ^^^ was intended for Will but the threads got intertwined :-)

yeah I got the message

@samcarter "Make America look nice again" would be better slogan too.

@DavidCarlisle how did you get this??

@William looks like a feature. It's not usually good style to use \fontspec multiple times 9or at all0 but it's not supposed to kill the document. I could trace it further but Will can probably fix it more quickly.

@UlrikeFischer @William's MWE a dozen or so comments above

@DavidCarlisle which one? The one with uccharclasses and all the fonts I don't have?

@William no I meant the ^^^ were for the other will (Robinson:-)

@UlrikeFischer yes I don't have them either but the document just hangs which isn't good

@DavidCarlisle doesn't hang for me. I even got a pdf which I find a bit surprising, I thought I would get font not found errors.

@UlrikeFischer Would be great if one could simply write user scripts to fix the problems there too!

@DavidCarlisle a sorry I have a local uccharclass.sty, will have to check without it.

@DavidCarlisle @William separating the font setup from the transition setting seems to work fine:

\newfontfamily\myfont{Arial}
\newfontfamily\myfontb{Times New Roman}
\newfontfamily\myfontc{Cambria}
\setDefaultTransitions{\myfont}{}
\setTransitionsForLatin{\myfontb}{}
\setTransitionsForGreek{\myfontc}{}

@William what's the actual use case here (I don't mean it's a bad example (actually it is clearly a bug:-) I just mean the use case seems very rare, normally when typeseting you want to know what font you are using and lay stuff out accordingly, a css-style fallback where each character individually falls back to some installed font that has the character is useful for getting arbitrary content read somehow on a browser but it is a long way from the use cases for which tex was designed

@UlrikeFischer interesting, as I said in my initial comment using \fontspec everywhere is bad style, interesting that it breaks so badly though, could you see where it was looping or did you just guess the simplification?

@DavidCarlisle I normally never use \fontspec but always set up fonts, so it was a natural try. Also \fontspec is a quite complicated command so I would never use it in "fragile" contexts.

Good evening to all active users.

@DavidCarlisle sorry laptop died had to plug it back in

Now I have seen the new site: i don't like much.

I have been just using mathjax and decided to transfer them from that to latex and well the font issues have been annoying

the use case in this example was me just using trial and error to try and get as many utf-8 as possible to disiplay

@DavidCarlisle, @UlrikeFischer @William halo'

in the browser everything pretty much displays fine

so the use case of the code above well isn't any except I figured I could just import all the languages I typically use

hello. I hope they will change the site again. that of now I don't like it very much. the fountain pen was really original as the green tick I liked. I hope that my eye can adapt as soon as possible. A cordial greeting to all of you.

Alan if there is a way to determine with 100% accuracy the correct font and imported package to display based of a specific character I don't mind adding it manually. The problem is at this point in time I am now not only learning latex for everything else(display) but also for inserting unicode characters

that I guess are slightly obscure

but googling them individually has been well not very successful

you guys have had to help me which I am certainly appreciative of

@Sebastiano nobody likes it:-)

@William The basic issue is that for math characters your choice of fonts is limited. So for math you need to use one of the available OpenType math fonts. For other languages/scripts the choices are much more varied, and any good OpenType font would probably be fine for you (e.g. the SIL Fonts are great fonts with good coverage). If you use LuaLaTeX or XeLaTeX, these fonts will work just fine (installed as system fonts) and you can enter all the relevant characters in the document source.

@William This is (roughly) because TeX treats math fonts and text fonts differently.

@William yes but in a browser a ransom note effect where each letter is in a different font is OK as you need to display arbitrary content preferably never issue an error message, but in a typesetting context that is so much a last resort scenario it's just not something that's set up in any of the normal document formats that people use.

@William note that you almost always have to mark up language transitions anyway as they can not be automatically determined, to tex language is much more than just the fonts to use, hyphenation, direction, spacing conventions are all language-specific, so it is natural to specify fonts to use for each language and automatically choosing a font for an arbitrary unicode character inserted with no markup into the text stream is almost never needed.

@DavidCarlisle @William Yes, this is a really important point.

Could it be that tex.meta.stackexchange.com/q/7781/36296 is the question with the most downvotes ever on tex.meta?

Massimo's comment is the best

@samcarter It is indeed.

@AlanMunn That's quite an achievement on a site with so few downvotes like ours!

@DavidCarlisle Ah. I thought I was one of the few. But is there a chance that it could improve the previous format?

@Sebastiano unlikely: the stackexchange staff like it (and this is not a democracy but a private dictatorship:-)

David your comment took the word outs of my mouth.

The last part specifically.

design disaster lol

@samcarter The site got much worse, I hope the same cannot be said about "my" new avatar ;-)

@DavidCarlisle You know how you said most characters shouldn't choke biber? It really doesn't like ä (Compiled fine after I put in \:a )

@Canageek what error did you get?

@UlrikeFischer Biber error: [841] Utils.pm:209> ERROR - Data file 'lit15.bib' cannot be read in encoding 'UTF-8': Can't decode ill-formed UTF-8 octet sequence <E4> at C:\Users\Matthew\AppData\Local\Temp\par-4d617474686577\cache-63e2f908d8b8c04291b‌​56fd8beb82a3890df0e98\inc\lib/File/Slurper.pm line 63.

It used to fail more gracefully, then biber used a newer version of some perl library

@Canageek sounds as if your file is not correctly encoded.

@UlrikeFischer Problem was fixed as soon as I removed the non-ASCII character

@Canageek well sure, if you remove the faulty byte then why should biber complain?

@UlrikeFischer Right. It used to fail gracefully, which made finding the right character to remove easier

@PauloCereda CONGRATS!

@Canageek but instead of removing the ä you should better reencode your bib-file in utf8. It is now obviously 8bit encoded and nowadays that is no longer sensible.

@marmot Finally one good thing happening to this site today!

@UlrikeFischer If I do that emacs will stop telling me when I've got certain characters in it though. This way it warns me when I've got curly quotes or stupid hyphens or such when I hit save.

@samcarter Why? Did @egreg draw a duck for @DavidCarlisle ? ;-)

@marmot You new avatar of course!

@UlrikeFischer That saves me a LOT of time hunting for - and curly quotes

@DavidCarlisle I do not want to believe that. But do you really say? You could improve the colors of the questions that remained pale pink of your favorite tags. The fountain pen had a certain professional effect. I realize that the more time goes by, the worse it gets instead of better. My sincerest greetings.

@UlrikeFischer (But even if I do, it doesn't print those characters correctly as I recall. It throws up another error saying it can't print them, and gives a (in the DOS prompt) garbbled image of them and lists the unicode number.

@Canageek then don't lie to biber and tell it that the file is utf8 encoding. Use the bibencoding key to inform it about the real encoding.

@UlrikeFischer Wait, aren't ascii files also always valid UTF8? I thought that was the whole point

@Canageek sure, but when you add an ä then it is no longer ascii.

@UlrikeFischer Or do you mean, don't lie and claim it is ASCII?

@marmot Did you see my new background image for the main site? i.stack.imgur.com/A99lA.png

@UlrikeFischer Oh, I see. Unicode sometimes (but is not required to) have a byte-order marker at the start. Is that what it is getting messed up by? Seems dangerous since it isn't "required or recommended" to have it

@samcarter OHH So cute! ;-)

@Sebastiano you don't want to believe that the network is run by a private company?

cana I doubt it but possible

@samcarter Just hope the bear is well-behaved and does not try to steal the honey liquor from the marmot. ;-)

@Canageek unicode is not an encoding. utf8 is, and ansinew, and latin1 and ... utf8 and ansinew encode ascii identically but ä differently.

@UlrikeFischer "The Unicode Standard permits the BOM in UTF-8,[3] but does not require or recommend its use.[4]"

@Canageek looks like user error, what encoding was your file in?

@DavidCarlisle I left it as ASCII, and this is a known bug I've seen before due to the perl library.

@marmot The duck in the middle keeps an eye on both to prevent any honey thefts.

Cana I believe Microsoft/Windows added the BOM but I'm not sure

@Canageek if it was ascii it could not have an accented character in it

@William Not when I save with emacs it doesn't, it seems

@William if you are replying to a comment please use the reply feature otherwise it is very hard to follow who is commenting on what

@DavidCarlisle obviously 8bit (I can reproduce the error with an ansinew encoded file).

Okay I thought it would get the closest match and ping them

hmm maybe only if multiple names with spaces

@DavidCarlisle Well, UTF-8 then, but it doesn't like the fact I leave emacs set to ASCII so that it tells me when I have certain characters that shouldn't be in ASCII in it.

@William You need an @ symbol for that

@William no you get no pings with the form you are using, You can ping someone with @ but better to reply to the specific comment, see the arrows at the left of this comment which links back to yours

@Canageek the bom is not relevant here. It is the way the ä is represented in the file.

@Canageek no it looks like you used an accented character in latin1 (iso-8859-1) so not accii and not utf-8

@DavidCarlisle Ah, that would make sense.

@Canageek naturally, I only talk sense:-)

Now to figure out how to get emacs to count this as an ASCII file and warn me about all the characters

@Canageek just save the buffer as utf8 ?

@DavidCarlisle Then it will have unicode characters I know don't work in it

@DavidCarlisle Ah, looks like updaing my files will fix it

@Canageek why won't they work?

@Canageek C-x <return> f will save the buffer in specified encoding if you say us-ascii at the prompt it will complain if there are any non-asciii characters

@marmot can this bear be so mean? chat.stackexchange.com/transcript/41?m=45884220#45884220

@DavidCarlisle Wait, when did it stop giving me error messages on curly quotes? Those used to break my document with an error

@Canageek I never use biber, so can't say:-) but what encoding were you using? you would get the error as above if you used a single byte curly quote but declared it to be utf-8

@DavidCarlisle Is there a way to see old versions a question?

@DavidCarlisle tex.stackexchange.com/questions/174723/… is when I asked about it

@Canageek yes click on the edit time and then you see the whole edit history

@DavidCarlisle Damn, isn't edited. Too bad, the top comment is using a bib file I had that didn't work. I don't have any old ones around, since I strip them down until they comiple.

can I do this in latex

other text <span style="font-family:arial">t</span>

essentially I was thinking I could find the characters that work for specific fonts and simply define them manually

@Canageek oh well I changed the inputenc message so it isn't the "gibberish" (actually important debugging information) but rather a unicode number such as U+4567 and also we have added a lot more characters with pre-defined utf8 mappings (as machines have more than 640K of memory these days)

@DavidCarlisle Oh, it shows the u+ number, but the actual character shows as gibberish

@William Okay, here's my attempt at a summary of the situation. PDFLaTeX: inherently built around various 7- and 8-bit encodings, but it’s been extended to support large parts of Unicode. In particular, UTF-8 has been the default input encoding in LaTeX since April. However, this is through the inputenc package (which you didn’t find in a package search because it’s now a basic part of the main kernel) and wouldn’t support Chinese.

@Canageek you can easily define it when you know the number.

XeLaTeX and LuaLaTeX both use Unicode internally and support packages such as fontspec and unicode-math. XeLaTeX has much better support for right-to-left languages than LuaLaTeX does. If you need multilingual text, I’d recommend you use XeLaTeX.

@William sure, you can locally switch to another font.

@Canageek not if your terminal is set up to show unicode, it is simply the unicode character fro which latesx doesn't have a definition, if your terminal can't show it either I can't do much about that (but at the time you asked the question i don't think it showed the U+... form at all)

Surely there is a way to specify font for a specific block of text

@William yes of course {\bfseries this is bold} for example, use any font command instead of \bfseries

@William Yes, you can change the font locally. Either declare a \newontfamily and switch to it, or write that as {\fontspec{Arial}t}.

fontspec perfect

To be able to display UTF-8 in the Windows terminal, type chcp 65001.

That's the UTF-8 code page.

@William it is much better to declare the font and then use the font command rather than use fontspec every time.

@William Excuse me, typo. \newfontfamily. Yes, I agree with #David.

@DavidCarlisle ;-)

@William So, what you probably want is \setsansfont{Arial} and then \textsf{t}.

@Canageek or simply look at the log file in something sensible, such as emacs.

is there an \endsansfont

because otherwise

@William no

I don't think that will work for my use case

@DavidCarlisle I've never figured out how to interact with the command line from within emacs

@Canageek i mean just load the log file into emacs after running tex

@William \textsf{t} puts only t in the sans font. You could also have written that {\sffamily t}, where the outer braces declare the scope of the font change.

@DavidCarlisle Oh, yeah, that would (hopefully) work.

@William it is just basic tex syntax not related to fonts particularly after {\foo xxx } the effect of \foo ends at the }

@William And there are more complicated options if you need to use more than one sans-serif font, but that's rarely been the case for me.

@William Anyway, back to your earlier question. Most math fonts probably don’t come with Chinese characters, but if you’ve set them up so they work in text, you can get them in math mode by writing $\text{传}$ or perhaps $\text{\textchinese{传}}$. You might also try \setmathfont[range=传, Scale=MatchUppercase]{Source Han Sans SC} or something like that, although I haven't always had success with it.

@WillRobertson By the way, the range= option of unicode-math fails silently if I give a range in the Private Use Area.

@Davislor thank you I am hopeful to figure out how to get Mathematica output to display properly

@William If you need them to be set as math variables or operator names, you can wrap them in \mathalpha{} or \mathop{}.

\documentclass{article}
\usepackage{unicode-math}

\begin{document}

    $$2 O_2  2 O_2$$

\end{document}

I used the Chinese symbol because well I had similar issues with those

not displaying

I don't think the character in the middle displayed properly in my browser?

exactly it is not* defined most places

@William And excuse me, what I typed got mangled by Markdown. There should be a single backtick, ` between the equal sign and the Chinese character.

Thank you I am using your examples trying to decide what to do with the Mathematica output

my current idea is to convert the Mathematica output to characters more common

↔

is basically what it looks like in Mathemtaica

@William well if you use chars from the private area of a font that you need exactly this font ..

hmm I am not sure if its private but definitely possible

probably. They convert everything else so well its well interesting the percularities

@William it is private. It is U+F3D4, en.wikipedia.org/wiki/Private_Use_Areas.

@William I did post one answer here that involved Chinese characters in math mode (tex.stackexchange.com/questions/438966/…), but it isn't the method I'd recommend.

@UlrikeFischer Is that private, or is one of the PUA codepoints that Chinese national standards use?

it shouldn't be a chinese character that one shouldn't

well I'm installing google's noto in hopes it solves my font problems

Well apparently the font standard wasn't designed to use all of unicode

@William no font uses all of unicode

@William it's a private use character so you shouldn't rely on it but if you want to rely on it use the newunicodechar package and use \newunicodechar{}{$\leftrightarrow$} or whatever other definition you want.

@William it is not just unused it is, by definition, only for use in private internal application files and has no meaning, and should not be used, in files that are exported externally.

@DavidCarlisle \setmainfont{UnicodeBMPFallback.ttf} comes quite near ;-).

So Mico is definitely pissed: tex.meta.stackexchange.com/a/7799/2693. I wonder if people are unhappy enough with the site design to move the site. Boycotting by itself seems a bit defeatist.

@AlanMunn there are always Stefan's sites but personally I think such moves can't really work and would just fragment the community.

@DavidCarlisle Yes, I worry about that too. We've had this discussion before, and I do tend to agree with you.

@William It isn't really even possible for the same font to cover Simplified and Traditional Chinese, Japanese and Korean at the same time.

@William But, here's how I got that example of yours to work.

\documentclass[varwidth, preview]{standalone}
\usepackage{amsmath} % for \text
\usepackage{ucharclasses} % for \setTransitonsForChinese
\usepackage{unicode-math}

\defaultfontfeatures{Scale=MatchUppercase, Ligatures=TeX}
\setmainfont{Palatino Linotype}[
  Scale=1.0,
  Ligatures={Common, Discretionary, TeX}]
% A good sans-serif font matching Palatino is Optima (or its free clone URW
% Classico).  A good monotype font matching it is Inconsolata.  Math fontseries
% matching it are Asana Math and TeX Gyre Pagella Math.

@William If you want any kind of special formatting for ideographs in math mode, you might want to define a command like \mathzh{} to provide it.

@Davislor thank you. Hmm

I like the webs display something approach

If it isn't 100% right

texture and font

compared to the others(on my system anyways) miktex is very nice

by the way

other kept exiting during install not sure why

my have been simple my old computer

@Davislor is that xelatex?

ahh I see I haven't installed Noto yet

1.45Gbs extracting

@William Well, TeX started out as a 7-bit encoding and has been gradually extending. @WillRobertson has done some great work in making unicode-math and fontspec, but right now, it doesn’t define any default fallback font. There are extensions of the default Computer Modern font that add support for Cyrillic, but not for Asian scripts so far.

@William Yes, that's with xelatex. It should also work in lualatex.

@Davislor you'd need to change the \usepackage{ucharclasses} % for \setTransitonsForChinese part for luatex, no?

I was wondering about this but assumed it would be closed(as it was) tex.stackexchange.com/questions/78215/…

oh well at least I got a good comparison answer

@DavidCarlisle Possibly. I didn't actually test it.

@DavidCarlisle Whoops, you’re right, that one only works for XeLaTeX.

@DavidCarlisle I know babel and polyglossia make you wrap a command around your Chinese characters, but that's probably a better approach anyway.

@AlanMunn @JosephWright should make it featured I think :)

@UlrikeFischer I don't know... isn't (s)he living with you? ;-)

@samcarter honey liquor ;-)

^^ I'm really actually pissed off just now.

:( of course I choose to lean latex right when a boycott occurs

perfect timing

good news is that the user base could drop to literally one person and SE has seemed to suggested the site will stay up

@William I'm sure the site itself won't disappear. And as with the massive exodus from GitHub (that didn't exactly happen), this might be a bit overblown too.

@William hey, don't worry too much about it. (1) There is other ways how to seek guidance, (2) the site is not going away, (3) what @AlanMunn just said :)

Anyone know how to get to the "USERS" page in the "Improved" site?

@PeterGrill Either through the lefthand menu (which you may have suppressed) or simply tex.stackexchange.com/users

@AlanMunn Yep that was the issue. I can't believe people think it is a good idea to take up that much screen space on the left-hand side. It would be much mode useful if it was back where it used to be!! Thanks for the info though.

@PeterGrill Click on the three horizontal lines at the top left. This will give you a dropdown with the usual links

@PeterGrill It's apparently being reserved for the "Next New Thing", customized question views.

@samcarter Oh thanks. I hadn't noticed that.

@PeterGrill There is also a user script to bring the old links back: stackapps.com/q/7938/54601 (the layout might need some improvement...)

@samcarter Ah yes, there it is. I swear I liked there earlier and it wasn't there! :-(

@samcarter Where are instructions on how to use those scripts? I am using Safari so don't think they will work but will consider switching to fix things if needed.

@AlanMunn people moaned about the black top bar when it was introduced but (a) they tweaked it a bit and (b) you get used to it so it worked out OK in the end. I suspect moaning now while they are rolling this out might have some effects, but I think the site will carry on pretty much as it was, whatever, comp.text.tex in a monspace font in emacs newsreader wasn't that pretty either...

@DavidCarlisle Yes, I think so too. But to the extent that some loud grumbling might make them rethink some stuff, it's moderately worthwhile I guess.

@PeterGrill For most browsers there are addons to run these scripts, for example greasemonkey for firefox, Tampermonkey for chrome etc. There are some browsers who support such scripts by naturally (Opera?). Unfortunately I don't really know about Safari.

@samcarter Ok, thanks for the info.

@samcarter How safe are these addons? (Yes, marmots sometimes do online banking when they need money for honey liquor. ;-)

@marmot safe enough, they can only do what javascript/css in the page can do anyway.

@DavidCarlisle Thanks!