TeX, LaTeX and Friends - 2015-07-29

« first day (1732 days earlier) ← previous day next day → last day (3208 days later) »

12:51 AM

@DavidCarlisle Physically, yes :) Mentally... not so much!

5 hours later…

5:43 AM

@JosephWright I uploaded a wrong pstricks.sty. Should be fixed by todays update, hopefully ...

6:21 AM

@Herbert Yes, it has been solved! Good!

1 hour later…

7:49 AM

Interesting bug of the day

0

Q: Svn-multi vs hyperref vs non-ascii svn info vs memoir headers

Here is the strange error of the day. Last weekend for "fun" I was playing around with Win 8 and TortoiseSVN. All of a sudden after a commit my LaTeX project no longer compiled. I'm down to the following MWE combo Master file: \documentclass[a4paper]{memoir} \usepackage[T1]{fontenc} \usepackag...

hyperref svn-multi tortoisesvn

2 hours later…

9:33 AM

@egreg all set to upgrade to windows 10?

9:59 AM

@DavidCarlisle Yes, on all my machines running Windoze.

@DavidCarlisle I'm sure it will break Cygwin. :P

@egreg I doubt it, I'm sure that that was priority number 1 during development of the update, as windows without cygwin isn't usable.

10:28 AM

@DavidCarlisle :D

10:42 AM

@DavidCarlisle LOL

@DavidCarlisle Yep Microsoft Linux. :)

I love English Language & Usage: 2.5k views on a "non-constructive" question with 2 votes :-)

@PauloCereda did you notice the power of google translate saves the day again: github.com/latex3/svn-mirror/issues/230#issuecomment-125915049

@DavidCarlisle ooh :)

Arthur Reutenauer

10:57 AM

@DavidCarlisle Yes, of course, XeTeX only loads the patterns in UTF-8, as does LuaTeX (on demand). I’m not sure what it could do with patterns in T1. Hyphenation patterns are set up to be loaded in two encodings for most languages, depending on the engine: T1 or some other 8-bit encoding for Knuth’s TeX and pdfTeX, and UTF-8 for XeTeX and LuaTeX. This is done in the (admittedly poorly named) package hyph-utf8, of which I’m also the maintainer with Mojca Miklavec.

@DavidCarlisle This happens at format generation time for all engines but LuaTeX, where you can load the patterns on demand (and then you get them in UTF-8).

@DavidCarlisle For some languages that have no legacy 8-bit encodings, the patterns are only available in UTF-8 for all engines, because there’s no reasonable 8-bit we could convert them too.

@DavidCarlisle Got the test version installed :-)

@ArthurReutenauer Indeed

@ArthurReutenauer @DavidCarlisle's suggestion was that XeTeX should read both sets of patterns in a systematic way. For documents coming for pdfTeX this would then allow selection of the T1 (or whatever) patterns if fontenc was loaded, or the UTF-8 patterns otherwise. For LuaTeX this should be doable dynamically, or perhaps better one can convert T1 (etc.) to Unicode at the input level.

@ArthurReutenauer The XeTeX part is somewhat speculative: may well be serious issues with the plan!

@ArthurReutenauer That's of course fine as there will not be 8-bit docs which are already using these languages (or at least using them with correct hyphenation)

Arthur Reutenauer

@JosephWright I’m interested to hear details :-)

@JosephWright Do I understand correctly that you mean 8-bit documents in XeTeX? Are there any example of such documents?

@ArthurReutenauer Any existing pdfTeX document where someone just goes 'I will replace pdfTeX with XeTeX'

Arthur Reutenauer

@JosephWright The patterns in hyph-utf8 are already converting from T1 (or whatever other 8-bit encoding) on the fly, so we can reuse that mechanism in LuaTeX.

@ArthurReutenauer The thinking was that at present if you take an existing (non-English) document you can't just switch engines, at least not without some adjustments

@ArthurReutenauer I thought that was all done at the TeX level?

Arthur Reutenauer

11:08 AM

@JosephWright Oh, right. Well, I wouldn’t expect it to work anyway, and I don’t think we should encourage users to do that.

@JosephWright Yes, at the TeX level. It’s a few lines of code: see l. 20-30 of tug.org/svn/texhyphen/trunk/hyph-utf8/tex/generic/hyph-utf8/… for example

@JosephWright The lines surrounding it are a workaround for pTeX, that uses some other crazy things.

@JosephWright Sorry, that’s the iniTeX level, in case there’s any confusion. Not at runtime - except for LuaTeX, as mentioned.

@ArthurReutenauer the trouble is, experts don't expect it to work, but the most natural thing for a user to try is just taking an existing document and running with xelatex or lualatex and see what happens. It's easy to inputenc work in xelatex (using its encoding support) but it's not a good idea to encourage that if hyphenation silently does the wrong thing. But apart from initial experiments I agree no one should be doing this, so you don't want to implement a vast infrastructure to make it work..

@JosephWright This is my case, for instance. @Arthur. I only want to use more than 16 math alphabets -- that's my motivation for switching to XeTeX. The problem is, I load fonts in the old way, because the new way gives troubles, for various reasons that nobody seems to be able to correct.

I almost ate my white USB stick thinking it was a white chocolate bar. :)

@PauloCereda you should tweet this message. :-)

Arthur Reutenauer

@DavidCarlisle Very much so. We should educate users about encodings and tell them to use UTF-8 if they want to use XeTeX and LuaTeX. Sticking to 8-bit encodings in that case is bound to fail in many bad ways and I don’t think we should support that.

@yo' Well, then convert your document to UTF-8, to start with. I understand you also have font problems, but the discussion I was having here was about legacy documents switching from pdfTeX and XeTeX.

11:19 AM

@ArthurReutenauer the problem is, this fails -- in UTF8, in both Lua and XeLaTeX:

Arthur Reutenauer

@yo' About your actual problem, ask a question on the main site, for Knuth’s sake. You can possibly expect this issue, which seems quite complex, to be solved just on the chat. This is not a support forum.

%!TeX program = xelatex

\documentclass{article}

\usepackage{tgpagella}
\usepackage{euler}
\usepackage[T1]{fontenc}

\begin{document}

Tomáš is not Tom\'a\v s

\[
t \neq T
\]

\end{document}

@ArthurReutenauer I know. These things had been asked before, and no satisfactory answer has been given

the actual problem is that Euler is not an OTF font

@yo' in maths but also there in text it has no chance of working as nothing is mapping the accented characters to the T1 encoding, but you are declaring that the fonts are T1 encoded

@yo' Did you try something like

\catcode`\š=\active
\defš{\char"B2\relax}

@DavidCarlisle that could be worked with since Pagella is somehow available in for fontspec

11:24 AM

for the š problem?

@JosephWright ok, this works, thanks a lot. The problem is: How many more like like this should I have? ...

@yo' yes it can all be fixed but you need to do it, best to get text sorted first as it's easier, but the suggestion above was to make the version that ypu would use for pdftex (with \usepackage[utf8]{inputenc} ) work with xetex/luatex.

@yo' I can provide photos of both USB stick and the chocolate. :)

@yo' 2^{21} :-)

Arthur Reutenauer

@yo' Well, then you know what the original problem. The next question is: why are you using Euler?

11:26 AM

@ArthurReutenauer because I like it?

Arthur Reutenauer

@yo' OK, that’s a valid reason. Going forward: what do you like more, that font or having your name typeset correctly?

@JosephWright \newunicodechar{š}{\symbol{"B2}} ;-)

@yo' Anything in T1 that doesn't directly map to UTF-8 will need handling: @DavidCarlisle probably knows better than I do how many cases there are

@egreg Yes, that of course also possible

@ArthurReutenauer I can have my name typeset correctly (with the help of \v s), it's more an issue of me smelling something's going bad :D

Arthur Reutenauer

@DavidCarlisle In practice closer to 2^7 ;-) We have the mappings in hyph-utf8.

11:27 AM

@ArthurReutenauer in practice only whatever is defined in T1enc, no?

@yo' Yes

Arthur Reutenauer

@yo' Do you have a problem with that smell?

@yo' It's just a question of looking over the upper half of the T1 range

@ArthurReutenauer not really that much (for now), I just hope I won't miss anything, since there's a lot of strange names of people throughout the thesis :)

now apologize me, I have a business meeting in 3 seconds :D Thank you all!

Arthur Reutenauer

@yo' So all you need is to check these names in your final document.

@yo' Bye. I would still recommend using some actual OpenType font, but whatever.

11:30 AM

@ArthurReutenauer yes, and be sure I didn't miss any (that's the tougher part)

@yo' you can just use \usepackage[utf8]{inputenc} with a version of the package you can find in my github, but currently I wouldn't recommend it as if you (or inputenc) make all the characters map correctly to T1 encoding it'll look like it's working but hyphenation is wrong unless you rebuild the format with T1 hyphenation patterns. Not using 16 math slots in a single expression is probably simpler:-)

Arthur Reutenauer

@yo' Of course it’s hard, that’s why it’s much more reasonable to just use another font. There’s also an attempt to port Euler to OpenType (github.com/khaledhosny/euler-otf), by Hans, Taco and Khaled. I haven’t tried it out myself, but since it’s very experimental it sounds like just another way to shoot yourself in the foot. Then again, you seem to like that.

Now back to the subject matter @DavidCarlisle and @JosephWright, do you really think that supporting 8-bit documents in UTF-8 engines is a reasonable use case? I’m in a position to improve the situation – from several angles ;-) – but I think it’s a really bad idea and we should strongly recommend against it. Not to mention I won’t have any time to work on that in the foreseeable future.

11:45 AM

@ArthurReutenauer some days I say yes other days no. It sort of doesn't make sense from a technical point of view, but from a marketing point of view, when a publisher such as @barbarabeeton asks "what will happen if I switch to xetex?" (or luatex) then if the answer is "all your documents will break unless you edit them" it's a much harder sell than if the answer was "mostly things will just work".

Arthur Reutenauer

@DavidCarlisle ;-)

@DavidCarlisle I have some vague sympathy for the marketing point of view, but the reality is that the correct answer is “mostly things will just not work at all and you should not do it”, and as much as I would like to sugarcoat it I can’t imagine any way to do it.

@ArthurReutenauer as I say some days I think that and would say "no" to trying to make it work, other days I think "how hard can it be...." basically if you put \XeTeXinputencoding "bytes" then you can use a pdftex document with inputenc and font enc etc, and it all works apart from hyphention. You've just about disabled every useful thing xetex does, but it's a way of gradually moving over. Perhaps, or perhaps its a really bad idea...

Arthur Reutenauer

@DavidCarlisle Hence I’d rather stick it to the technical point of view because there’s no good angle to market XeTeX (or even LuaTeX, in my opinion) as the successor to pdfTeX / Knuth’s TeX. Worse even, I’m firmly convinced that approaching the situation from that angle would have stifled real progress and we’d have the worst of both worlds: incomplete support in pdfTeX, and no UTF-8 engine at all.

People discussing TeX in a TeX chatroom... where is this world going to.

@ArthurReutenauer possibly.

@PauloCereda 日本

11:57 AM

@DavidCarlisle o.O

user image

@cfr: ^^ :)

Arthur Reutenauer

@DavidCarlisle I can try and convince Mojca to include 8-bit patterns for UTF-8 engines but I can’t promise anything. As a matter of fact, we’re already including 8-bit patterns for pTeX anyway, but that’s clearly a kludge - and of doubtful utility too.

@ArthurReutenauer So what do we use in place of pdfTeX then?

@ArthurReutenauer I'm really not sure it's a good idea, just observing that that would be needed if you wanted to make it work.

Arthur Reutenauer

@JosephWright Nothing. Just use pdfTeX. It’s going to be in distributions for the foreseeable future.

@DavidCarlisle Oh, sure.

@ArthurReutenauer Well, I completely forgot about the option of using fontspec for text fonts and good old euler for math fonts -- that should work, right?

12:04 PM

@yo' mostly (as long as you stick to ascii input in math)

@JosephWright @yo' This should be all:

\usepackage{newunicodechar}
\newunicodechar{Ă}{'200}
\newunicodechar{Ą}{'201}
\newunicodechar{Ć}{'202}
\newunicodechar{Č}{'203}
\newunicodechar{Ď}{'204}
\newunicodechar{Ě}{'205}
\newunicodechar{Ę}{'206}
\newunicodechar{Ğ}{'207}
\newunicodechar{Ĺ}{'210}
\newunicodechar{Ľ}{'211}
\newunicodechar{Ł}{'212}
\newunicodechar{Ń}{'213}
\newunicodechar{Ň}{'214}
\newunicodechar{Ŋ}{'215}
\newunicodechar{Ő}{'216}
\newunicodechar{Ŕ}{'217}
\newunicodechar{Ř}{'220}
\newunicodechar{Ś}{'221}
\newunicodechar{Š}{'222}

@DavidCarlisle I did. But why in the world does LuaLaTeX complain that

! LaTeX Error: Too many math alphabets used in version normal.

@yo' because you are using tl2014 and latex only got told that there were 256 math alphabets this year....

@DavidCarlisle ah ok. Time for another tl-install :-)

Arthur Reutenauer

@yo' I can’t say off the top of my head, but yes, you should definitely start with that. That’s why @DavidCarlisle was trying to tell you.

@yo' As we’ve heard repeatedly these past few days, unicode-math is ... not quite ready ;-)

12:09 PM

@ArthurReutenauer :-)

@JosephWright: where is Unicodeland, by the way? :)

@PauloCereda close to Nowhereland I suppose :-) But hopefully it's very far from Neverland :-)

@yo' Or Latinshire, probably. :)

damn 50GB for / directory starts to be small...

now I gotta go again. See you in the evening, and thanks for your help, @Arthur, @David and @Joseph !

> The Committee are very happy to announce that the 2015 UK-TUG Speaker Meeting will take place on the 31st of October at Trinity College, Oxford. Attendance at the meeting is free, and a light sandwich lunch is included in the day. To help planning for the day it would be welcome if people hoping to attend could let the us know. At the same time, talk titles are very welcome: the day is usually informal but it is useful if we have a rough outline of likely talks (including length).

light sandwich lunch is included in the day

wooooooooo

12:24 PM

@PauloCereda Is beer included?

@egreg ooh that's an important question! Let's ask @JosephWright. :)

12:40 PM

@PauloCereda This is being written from my iPhone.😎

@PauloCereda It's a pain, though, so I'm returning to the ordinary keyboard. ;-)

@egreg ooh a Unicode smiley! :)

@PauloCereda That's perhaps the only reason for using the phone.

1:01 PM

@PauloCereda ;)

How can I get a space if and only if there's a prenote?

0

A: How to change a \NewDocumentCommand-based \cite command to accept biblatex-style arguments?

Here's an attempt to blend xparse with Biblatex. Unfortunately, I can't get the space after the prenote to come out correctly. Either I get a space even with no note, or I don't get one even with. \documentclass{article} \usepackage[utf8]{inputenc} \begin{filecontents}{\jobname.bib} @mvreferen...

@yo' an alternative would be to get latexrelease.sty from somewhere and then add \RequirePackage[2015/01/01]{latexrelease} at the top of your document and upgrade to 2015 without actually installing it

1:40 PM

@JosephWright linkedin just mailed me to tell me that you know something about Chemistry. Should come in useful when typesetting Greek in Japanese texts!

2:14 PM

@egreg Another possibility is to download unicode.org/charts/PDF/U1F600.pdf and copy-paste characters from there 😸

@StephanLehmke This is even easier, thank you! :P

Whyever they have several kinds of camels but no ducks is a mystery to me 🐪🐫

@StephanLehmke According to a colleague of mine that was for some time in Somalia, they measure surfaces in camels: for instance my office could be six camels.

@egreg One-humped or two-humped?

@StephanLehmke I don't think it makes a difference for measuring. But, who knows?

2:22 PM

So you say the symbol for this unit is an actual camel?

@StephanLehmke Maybe that's the reason why Unicode added it. :) Perhaps there's some log about the inclusion.

What's happening? The third person mailing me this week asking for help with this error: tex.stackexchange.com/questions/245370 o.O

@clemens Check again. :)

@StephanLehmke ducks are in the pipeline for Unicode 9

@DavidCarlisle Yay. Long overdue I say.

2:38 PM

@StephanLehmke emojipedia.org/duck

@PauloCereda didn't fool me! :p

@DavidCarlisle Yea I noticed they have tons of different chickens but no ducks which is kinda racist if you think about it.

@clemens :)

@StephanLehmke Yay!

@StephanLehmke ooh

2:52 PM

Apart from it not being an answer, can somebody get some sense out of this?

0

A: Index (printindex) in appendix has different format than other chapters

I have completed a historical manuscript, and have added several appendices, should those men listed in the appendix, do those names listed in the appendices but not in the text need to be also added to the Index

@egreg I think he is indexing names of some people in his document and asks if he should index the occurrences of those names in the appendix, too.

(which would be an off-topic question)

3:06 PM

@egreg Hold on, I'll send you tfm and map files now.

@1010011010 Also mention what you did for producing them (with otf2tfm, I guess).

@1010011010 No pfb

@egreg Okay. Sent you the tfm files and an updated MWE.tex (one that contained pdfmapline instead of pdfmapfile to remove that dependency. The relevant command was otftotfm -e ec LeMondeLivrePro-Italic.otf fmljri7t.

Oops, that doesn't make sense. I think it was otftotfm -e 7t LeMondeLivrePro-Italic.otf fmljri7t

@egreg Hold on. Made a mistake. Sorry. I'll send you the real thing now. Otftotfm doesn't produce pfb files without automatic mode.

@1010011010 Font OT1/fmlj/m/n/10=fmljr7t at 10.0pt not loadable

@egreg All right. I remember what was up again. I'm not sure why, but there's something about how otf2tfm installs fonts that doesn't come natural to the fonttable package. So automatic mode is necessary to install the font correctly (somehow non-automatic mode doesn't produce the .mf files or something). Do you mind actually installing the font? I've now again reproduced the error going from the file I sent you earlier.

@1010011010 No, sorry. Without precise instructions about producing the fonts for pdftex I'm not going to experiment.

3:21 PM

@DavidCarlisle Do you think I should raise (u)pTeX support on LaTeX-L

@DavidCarlisle Also perhaps a warning about some deprecations?

I've done the following:

1. Put `MWE.tex`, `LeMondeLivrePro-Italic.otf` and `ot1fmlj.fd` in a working directory.

2. Open command prompt, use the command `otftotfm -a -e 7t --vendor somevendor LeMondeLivrePro-Italic.otf fmljri7t` with the otf file in the working directory. (For some reason, some files are only produced in automatic mode which are crucial to package `fonttable`.)

3. Run MWE.tex with pdfLaTeX. The error given is: `!pdfTeX error: pdflatex.exe (file c:/Users/Tobias/.texlive2014/texmf-var/fonts/

I've sent you an updated MWE.tex which I used in this process.

@JosephWright possibly or perhaps more usefully ask that someone raise it on whatever list/forum is appropriate for ptex so we get feedback from people who have used it, I guess not many of those on latex-l

@JosephWright ?

@DavidCarlisle \<engine>_if_engine:TF => (probably) \runtime_if_<engine>:TF (with addition of (u)pTeX to set), case-change stuff

@JosephWright ah that yes.

@DavidCarlisle Not many changes but will need people to adjust, not immediately but over the next say year

3:25 PM

@JosephWright or just after you delete the old names:-)

@DavidCarlisle Lesson learned: next time I do a removal I'll do a sweep over TL and warn everyone in advance

@1010011010 I get no somevendor.map file. I give up.

@egreg The command otftotfm -a -e 7t --vendor somevendor LeMondeLivrePro-Italic.otf fmljri7t should produce this map file?????????

@1010011010 It doesn't.

@egreg Strange. I'll send you that as well, then.

3:31 PM

@1010011010 NO

@1010011010 I don't want to pollute my TeX system.

Hello.

@JohnMolokach Hi

@DavidCarlisle The 'sort the line length' thing has revealed another LuaTeX log 'feature'

Is there a way to change the 'size' of \displaystyle while keeping the subscripts below symbols?

@JohnMolokach Didn't you just ask that on the site?

Yes.

But it seems like a chat will spring from it.

3:38 PM

@JosephWright what a surprise

Sorry. Did I break a rule or something?

@1010011010 With the automatic mode, updmap is run, which is something I don't want to do. Ever. If I remove it from the options and use the following MWE.tex, I get the image you can see.

\documentclass{article}
%\usepackage[english]{babel}
\usepackage{fonttable}
%\usepackage[OT1]{fontenc}
%\usepackage[utf8x]{inputenc}
%\usepackage{kantlipsum}
%\usepackage{geometry}
%\usepackage{microtype}
%\renewcommand\rmdefault{fmlj}
\pdfmapline{+fmljri7t--lcdfj LeMondeLivrePro-It "AutoEnc_os7w3liwzknjkjushu3ewpowag ReEncodeFont" <[a_os7w3l.enc <LeMondeLivrePro-It.pfb}
\pdfmapline{+fmljri7t--base LeMondeLivrePro-It "AutoEnc_os7w3liwzknjkjushu3ewpowag ReEncodeFont" <[a_os7w3l.enc <LeMondeLivrePro-It.pfb}

user image

@JohnMolokach No, just a comment :-)

@JohnMolokach Are you after \limits, perhaps?

OK sorry. I'll wait for a (hopefully) an answer.

yes

$\displaystyle\lim_{x\to c}\left(\frac fg\right)=\frac{\displaystyle\lim_{x\to c}f}{\displaystyle\lim_{x\to c}g},$ provided $\displaystyle\lim_{x\to c}g\ne0.$

\documentclass{article}
\begin{document}

$\lim\limits_{x\to c}\left(\frac{f}{g}\right)
= \frac{\lim\limits_{x\to c}f}{\lim\limits_{x\to c}g},$
  provided $\lim\limits_{x\to c}g\ne0.$

\end{document}

3:41 PM

I was writing some ideas for a paper when I found out someone else already wrote about it. The bastard was my past self. :)

The fraction with display style seems like it should have the numerator and denominator resized....

Ahh perfect! Thank you! Is there a similar thing for integrals and summations?

@JohnMolokach Still \limits :-)

@JohnMolokach I assume you know this isn't the default in running text for a reason

@PauloCereda how could he!

Perfect. Thanks. I'll look up the documentation on that command :)

@clemens Exactly!

3:43 PM

Bye! Thanks again!

@JohnMolokach but why are you putting such a large expression in inline math? the reason inline math uses a different layout for limits is to avoid opening up the paragraph so it all fits in 1 baseline space, but you are specifying inline math then fighting tex at every stage...

@ArthurReutenauer and @DavidCarlisle -- please keep talking, guys. there is a good angle to market xetex: the ability to use otf fonts. that's why the ams is interested. the stix fonts (v.2 in the works) will only be available as otf for some time, and if they are to be used soon, they will have to be used with xetex. and at ams this will have to be in a production environment, where if it's necessary to "correct" things that would work with pdftex, that will make things much more expensive.

@barbarabeeton ah but that's a rather different case you want to use OTF in math so you are going to need unicode-math package basically, the discussion above was whether it is worth trying to make a pdflatex document work unchanged and using the same 8bit fonts it was using, which is technically possible but one wonders if it is worth it compared to just staying with pdftex

@JosephWright I see why the accented characters are not normalised to ^^ab in the ptex log, it writes them in 8bit encoding not utf8....

@barbarabeeton So you are going to be after me every time I break fontspec or unicode-math by updating expl3 ;-)

@DavidCarlisle Ah

@DavidCarlisle We can try to normalise that I guess

@DavidCarlisle Possibly not worth it

@DavidCarlisle (u)pTeX support is fun, isn't it (almost as much as LuaTeX)

@barbarabeeton As @DavidCarlisle says, presumably such documents will be set up from the get-go to use UTF-8, so they will be OK

3:59 PM

@JosephWright I wouldn't try to do both at same time, you could perhaps have a top level flag that says whether the log you are normalising is utf8 or iso8859, so you could set that in the lua setup....

@barbarabeeton What's your take on the TeX Gyre math fonts being OTF-only? Some people regard this as sub-optimal, but if STIX is going the same way (more or less) there is a logic.

@DavidCarlisle Well we know the engine, so that can be decided by l3build

@JosephWright oh OK, yes

@DavidCarlisle First let me sort the latest update: have altered L3 but not L2e to normalise line length issues

@JosephWright i have a horrible feeling it's going to write Japanese in UTF8 and non-japanese in latin1 (even if it's greek) all to the same file....

@JosephWright OK:-)

@DavidCarlisle That was implied by the e-mail we got

4:02 PM

@JosephWright but I'm sure l3build will normalise it all beautifully

@DavidCarlisle Eventually

Can somebody pleas please pretty please size down the images of Manuüel Khner in the TeX-Showcase big list question? Messes up my whole system and i can't be the only one announced by images of Image Size: 4961x7016. I thought imgur would resize them automatically.

@Johannes_B language...

@JosephWright -- maybe. first, we have to get tex live 2015 installed and working on our current production. only then can we think about testing the new stix fonts. and yes, unicode-math will be a component of the latter. (the reason we won't even touch luatex is the fact that it's not sufficiently stable for a production environment. are you telling me that now xelatex isn't stable and won't be for a while?

@DavidCarlisle Indeed

@barbarabeeton No, that wasn't my point

4:07 PM

@egreg Did you produce this without automatic mode or by saying --no-updmap?

@DavidCarlisle Sorry, can't edit anymore. The answer and its pics crashed firefox. Maybe @Joseph can fix it, please.

I'm assuming those pdfmaplines were output from otf2tfm?

@Johannes_B Done

@JosephWright Thanks.

@DavidCarlisle -- i think we've got a 1.5-2 year backlog for most journals, which means that pdftex is in the picture for at least that long. but we were hoping to install tex live 2015 for production (we're still using 2012), but now i think i'm hearing that's not stable enough (or at leaast latex 2015) for us to do it safely?

4:09 PM

Because I've never seen that lcdfj line before.....

@barbarabeeton XeTeX is stable (though there are some questions about RTL support: probably not relevant to you!), it's just that occasionally there's a passing mismatch between expl3 and fontspec/unicode-math. We've had one recently, but hopefully I now know what to do to avoid it in the future.

@barbarabeeton don't see any issue about stability in the current discussion? You could wait a thousand years and still have issues at the point you change over

@barbarabeeton For a production system you are going to have one version of unicode-math, etc., so once it is working it will be fine

@barbarabeeton If you use the DVD version of TL'15 you'll be fine: we are always extra-careful about any updates during the DVD production phase

@barbarabeeton would your plan be to have separate author guidelines for xetex/stix submissions or to keep author guidelines as is and just use xetex/stix internally in production?

@barbarabeeton The team work pretty hard to make sure any code using expl3 gets fixed when an issue comes up, and Will is the maintainer of fontspec and unicode-math so any problems are normally fixed very fast.

4:14 PM

@JosephWright except when he's in a plane flying half way round the world to see @barbarabeeton

@JosephWright -- the stix fonts will be "enhanced" for pdflatex, with tfm and pfb files, but that will take a while, and there's pressure to start testing and some special production use before that. (other stipub members will need the pdftex option even if ams manages to switch.) regarding the tex gyre fonts, i have mixed feelings. i think it will force some individual users to switch to xetex, but i don't know any publishers who are using those fonts.

@DavidCarlisle I was using an itemized list, so no worries in the line spacing. Regarding that, I usually use \[2mm] at the end of a line containing display math when in a paragraph.

@JohnMolokach :(

@DavidCarlisle ?

@barbarabeeton Thanks: useful to have an insight from someone at the publishing end (for an 'end user', as I am when I use these things, they work nicely, but I don't need archival stability for my sources)

4:19 PM

@JohnMolokach any use of \\ within the document (outside of its use in tables and alignments) is usually a sign that something has gone wrong:)

@barbarabeeton Other than any changes @WillRobertson makes following the TUG2015 discussions, the interface of unicode-math is I think pretty stable, so the only issues will be transitory ones when we have updates out-of-sync. As I say, they tend to last only a few days, so for a system updated very rarely will not be an issue.

@DavidCarlisle -- it may be necessary to have separate author guidelines; to be determined. but we also have to accommodate a number of back-end databases; they only function in 8-bit at the moment, and getting those reprogrammed is nontrivial. (they're used for author name matching. and if there's anything we have to handle that's rife with accented letters, it's author names.) also, have to see what math reviews is doing. if they're still 8-bit-only, that's another stubbed toe.

@DavidCarlisle I think it's a matter of preference. I just dislike inline formatting with limits.

@barbarabeeton but you will also get people typing Greek math in Greek and using unicode for symbols...

@DavidCarlisle I'd never write it that way by hand so it seems awkward in text.

4:21 PM

@JosephWright -- i think it would be useful to have a discussion by emall once i get back to providence and have updated my group with what i've learned. with @DavidCarlisle too. okay?

@barbarabeeton Sure

@JohnMolokach but you are using a math formatter that goes to some effort to do that, so you are having to disable the behaviour all the time (and if you ever submitted to the AMS @barbarabeeton would take out all your \limits commands before publishing:-)

@barbarabeeton OK

@DavidCarlisle -- actually, that might be expected. the symbols part anyhow. and we'll have to deal with it. but for greek, i think we'll have to be explicit about using macros. what does mathml do? is a unicode symbol okay for, say, a greater-or-equal sign?

it will be a bit of a shock for the production editors ...

@barbarabeeton yes sure xml is unicode to its core. A MathML renderer never gets passed the name of the entity used or if an entity was used &rightarrow;, →, → are all indistinguishable by the time a mathml system sees the data, there is no information which was used.

@DavidCarlisle in that case I expect it is best practice to write text out in words and use the limit symbols in display math using double $$

4:28 PM

@JohnMolokach well not $$ in latex which does not support that (use \[...\]) but yes.

@1010011010 just without -a and adding the map entries manually.

@egreg The command prompt output both of those lines?

@JohnMolokach but actually the case of display math in an itemized list is a bit special currently using $\displaystyle....$ is probably the best way as latex doesn't really support any way of placing the item label sensibly by a displayed equation

@JosephWright -- back to the hyphenation question ... you (or someone else with inside knowledge) has hinted that hyphenation with xetex doesn't behave the same as with pdftex. does this mean that the patterns (for english) are different/give different results, or something else?

@barbarabeeton No, that's not the issue :-)

4:30 PM

@barbarabeeton luatex it can be different I think not the case with xetex. (unless I missed something)

@barbarabeeton The patterns read by XeTeX are (reasonably) in UTF-8 format and therefore are based around Unicode char positions

@1010011010 No, but it was easy to guess the missing one.

@barbarabeeton but for non-english languages the hyphenation is tied to font encoding so if you use a unicode font you need to use different patterns than if you are using an 8bit font

@barbarabeeton In English that makes no difference, but in languages where there are differences between Unicode and T1 (or ...) then some hyphenation will be wrong if you try to use a traditionalaly-encoded font with XeTeX

@barbarabeeton It's really a problem only for documents that don't start out expecting Unicode set ups

@JosephWright -- so if an english text is processed separately with pdflatex or xelatex with no special encoding, the results should be reliably the same?

4:33 PM

@barbarabeeton Yes

@barbarabeeton ASCII range is fine :-)

@JosephWright -- that's a comfort. so we really only need to watch out if there's a specified \inputenc?

@barbarabeeton For most users, they'll be loading fontspec so using a Unicode-encoded font set up and there is no issue

@barbarabeeton fontenc

@barbarabeeton or math....

@JosephWright -- not yet. that's where the (possible/probable) need for different user guidelines comes in.

@barbarabeeton Well again there is no issue in English as OT1/T1 has the ASCII slots just fine

4:36 PM

@JosephWright > :-)

@JosephWright -- but we get lot of authors with non-ascii names. the text should be fine, but references could go way haywire.

@DavidCarlisle Indeed: I think we didn't mess with the default encoding for XeTeX/LuaTeX on the grounds anyone using anything other than English is going to need fontspec anyway :-)

@barbarabeeton But they won't hyphenate well anyway (with English patterns), so presumably are done by hand

@barbarabeeton how do people mark that up \'{e} ?

@JosephWright -- and that's why we'll have to determine what math reviews/mathscinet is doing.

@DavidCarlisle -- \'{e} is what our author database name-matching now requires.

@barbarabeeton if using OT1 classic tex encoding then there is no hyphenation anyway for that.

4:40 PM

@JosephWright -- we make no assumptions about non-english hyphenation; there's a lot of discretionary hyphens in use.

@barbarabeeton You are fine then: the issue @DavidCarlisle raised only applies to non-English texts where they are covered by existing 8-bit patterns (so could have worked with pdfTeX) and where a XeTeX/LuaTeX document is created but attempts to use a traditional font encoding and expect hyphenation from patterns

@DavidCarlisle -- we actually try to not hyphenate names. but you're correct; when it's done, it's done by hand. (on the other hand, titles of non-english references could be handled much more nicely using a polyglossia approach.)

I see Tin Tin!

@barbarabeeton The issue @DavidCarlisle spotted hasn't been reported 'in the wild' that I know of: it's more that in some ways it would be nice to say to users 'LaTeX will take care of all of that' with no changes to documents when taking a pdfTeX source and using with XeTeX/LuaTeX. As discussed earlier with @ArthurReutenauer, that's perhaps not a sensible approach.

@JosephWright yes although people (real people) are very tolerant of hyphenation tables being wrong, so my fear is that lots of people's initial luatex/xetex documents are doing exactly this, unreported.

4:46 PM

@DavidCarlisle Certainly: I was thinking that the place it would be picked up would be publishers, but presumably if they are using XeTeX they are also adjusting the preamble

@DavidCarlisle The stock advice for a XeTeX/LuaTeX LaTeX document is 'Drop fontenc and inputenc and load fontspec' :-)

@JosephWright or sticking rigorously to 7bit coding conventions and English language.

@DavidCarlisle Yes, or that

@egreg Well, thanks for your help. confusing that there's a problem with the map file.

@JosephWright which is OK for a person but not so good if someone asks what will happen if they tried running their database of half a million documents on luatex because they want to move to that going forwards and don't want to maintain two systems....

home time..

@JosephWright This is duplicate of a question with 17 answers; the three that have been given here are almost certainly duplicate either. :(

2

Q: Line break within a cell of a table

I need a line break in one of the column headers. I tried using the suggestions from the earlier posts but since the column header is an equation, I couldn't use them. I need the line break in the column header of 4th column as given below :- The code i have written is : \begin {table}[h] \...

4:51 PM

@egreg Action required?

@JosephWright Merging? Or telling something to the answerer?

@egreg I can merge, but almost always we then get a complaint that in some way the question was subtly different!

5:04 PM

@JosephWright -- we'll be testing the stix fonts by reprocessing a lot of current articles that have already been set with pdflatex. i'll suggest that the production editors look carefully for hyphenation anomalies, and if we find anything really inexplicable or unusual, will report to you guys. i suppose we should also report if everything goes swimmingly, so you're not left just wondering.

@barbarabeeton Hyphenation should be fine: the things to look for are fontenc/inputenc/'classical' font packages (remove) and fontspec (add)

5:23 PM

@JosephWright I was very serious about my attempt (where I forgot something):

\usepackage{newunicodechar}
\newunicodechar{Ă}{\symbol{'200}}
\newunicodechar{Ą}{\symbol{'201}}
\newunicodechar{Ć}{\symbol{'202}}
\newunicodechar{Č}{\symbol{'203}}
\newunicodechar{Ď}{\symbol{'204}}
\newunicodechar{Ě}{\symbol{'205}}
\newunicodechar{Ę}{\symbol{'206}}
\newunicodechar{Ğ}{\symbol{'207}}
\newunicodechar{Ĺ}{\symbol{'210}}
\newunicodechar{Ľ}{\symbol{'211}}
\newunicodechar{Ł}{\symbol{'212}}
\newunicodechar{Ń}{\symbol{'213}}
\newunicodechar{Ň}{\symbol{'214}}
\newunicodechar{Ŋ}{\symbol{'215}}

@JosephWright But, of course, one needs that the T1 hyphenation patterns are loaded.

@JosephWright At least, the output will be correct (apart from hyphenation glitches), probably just missing hyphenation points.

@egreg Sure

@JosephWright -- "unicodeland" paper received. a few typos, but looks good. i see we need (at ams) to check the automatic uppercasing for running heads (author names on recto pages; ugh; we've had problems with icelandic and turkish, in particular, in the past).

@barbarabeeton Case changing is complex, certainly

@barbarabeeton I still need to finish off Greek properly: what I have at present is not quite right for modern Greek I think

@JosephWright: slides? :)

@PauloCereda Mainly live demo

@PauloCereda Will send to you

5:31 PM

@JosephWright <3

5:50 PM

@egreg isn't that more or less same as \input{utf8enc.dfu} with \DeclareUnicodeCharacter suitably defined?

@barbarabeeton also possible difference in image handling, hyperref, colour and any other driver-specific things

@DavidCarlisle -- thanks. any other pointers you can come up with will be gratefully accepted.

Christian Hupfer

Yeah:

ALICE COOPER - School's Out (1972 UK TV Top Of The Pops Performance) ~ HIGH QUALITY HQ ~

@DavidCarlisle Possibly

Christian Hupfer

Almost six weeks of vacation :D Teacher's life is good :D

@DavidCarlisle That's left as an exercise for Heiko.

@ChristianHupfer Well, I closed my office a couple of hours ago, but I must be there on September 2. :(

Christian Hupfer

6:02 PM

@egreg I'll just marked the date in my calendar. I'll appear then and teach you some TeX :-P

1

Q: Define a command with an odd number of braces

In my code, I change to '2 columns' mode typing: \twocolumn[{ I'm trying to simplify this code, by creating a new command: \renewcommand{\twocolumn}{\twocolumn[{} Is this possible to insert a bracket and a brace in this new command? I'm getting a lot of errors doing this.

A question for @egreg, @DavidCarlisle and @JosephWright (and all other TeX heros here, although I doubt the usefulness of such a command)

@ChristianHupfer I'm not answering without a full use case!

Christian Hupfer

@JosephWright I doubt there is such one, but I would be interested how it is done. Something with the D type of parameters in \NewDocumentCommand?

@ChristianHupfer Well \let\oldtwocolumn\twocolumn\RenewDocumentCommand\twocolumn{O{}}{\oldtwocolumn[{‌#1}]} would do it

@ChristianHupfer xparse does bracket matching :-)

Christian Hupfer

@JosephWright: I think I am misunderstanding the question somewhat. I thought of a completely other stuff.

@ChristianHupfer Well the question wants an unbalanced brace, which you can't do

@ChristianHupfer Or at least not in a way meaningful here

Christian Hupfer

6:17 PM

@JosephWright Yes, in that sense I understood it...

@JosephWright Yes, as I suspected... not meaningful/useful. I am aware about the {O{}} way of \NewDocumentCommand and its bracket matching feature (using it very often ;-))

6:36 PM

@DavidCarlisle Interesting discoveries with line wrapping the log. pdfTeX wraps at 79 bytes (so can split a UTF-8 sequence), LuaTeX wraps at (79 +1) bytes unless it will break a UTF-8 sequence, XeTeX wraps at 79 code points, upTeX wraps at 79 bytes unless it will break a UTF-8 sequence, pTeX goodness only knows (79 bytes perhaps, but the Japanese conversion can't be turned off so ...)

Christian Hupfer

@JosephWright: There's an answer now to the odd-brace number macro question:

0

A: Define a command with an odd number of braces

The specific answer to your question is `NO', it is not possible to define a command with an odd number of braces in the specimen code you have provided. However, if you can provide a MWE with real requirement, then somebody can definitely help.

@ChristianHupfer More a comment but ...

Christian Hupfer

@JosephWright: In my opinion, a comment rather than an 'answer'

@JosephWright my original plan of setting a very wide line length so nothing wraps has some appeal..

@DavidCarlisle It's not a big issue, really, as anything that has UTF-8 code points where this shows will probably be different between pdfTeX and the other engines in any case, so you need separate .tlg files anyway

Christian Hupfer

6:44 PM

Perhaps it's a dupe (I didn't check it, actually), but perhaps it should be undeleted:

tex.stackexchange.com/questions/257746/…

@DavidCarlisle What's a bit odd is that upTeX is described as 'not Unicode aware' outside of CJK in github.com/latex3/svn-mirror/issues/230#issuecomment-125930389 but it does know not to split the code point

7:36 PM

@DavidCarlisle Looks like we are now finding (e)pTeX bugs: tug.org/svn/texlive/trunk/Build/source/texk/web2c/eptexdir/…

@DavidCarlisle twitter.com/argami_/status/626309932435001344/photo/1

3

@JosephWright here we go again....

@Johannes_B I've never seen the one on the left

@DavidCarlisle :-)

Christian Hupfer

@Johannes_B @DavidCarlisle is that old he remembers even the time why such an error is called a bug :-P (bugs in ENIAC ;-)

@ChristianHupfer I think you must be confusing me with that old bloke who gave the last talk at Darmstadt

Christian Hupfer

7:49 PM

@DavidCarlisle Nope -- I rather think that you and that other 'old' bloke are the same person :-P

@ChristianHupfer @DavidCarlisle The one selfie you posted was indeed creepy ;-)

Christian Hupfer

@PaulGessler: Hello

8:04 PM

I need biscuits.

Maybe some Black Forest ones. :)

@PauloCereda I've had some Lotus earlier today, and I put Lotus spread (crunchy) on my bread for breakfast :)

@yo' oooooh

:)

btw, sound-mastering my first real thing on Saturday, cross your fingers for me :-)

Christian Hupfer

@PauloCereda Nice try, but the local duck ate all cookies here... :-P

@ChristianHupfer Oh no!

Das lokal Ente!

Better. :)

Actually, lokal is an adjective, so it is not capitalized. :)

8:19 PM

@David

@DavidCarlisle may I ask what is the difference between \[2mm] and \vskip 2mm in text?

sorry I mean "\[2mm]"

you know... there are two

actually anyone can answer, but @DavidCarlisle and I had a discussion about this earlier.

Christian Hupfer

@PauloCereda: Die lokale Ente :-P... and I did not capitalize something there...

@ChristianHupfer Oh no!

Christian Hupfer

Number three... The Stork @PauloCereda

@PauloCereda: Actually, my favourite bird... we've pretty much of them on the grounds here during spring and sommer, sometimes 25 of them at the same time

@ChristianHupfer LOL

Christian Hupfer

@PauloCereda and no, he is not pining for the Fjords @JosephWright ;-)

8:34 PM

@JohnMolokach quite a lot really, \vskip is a tex primitive if used mid line it will force the end of a paragraph, and add vertical space \vspace is the analogous latex command but its behaviour in a line is different, it adds space after the paragraph has been broken in to lines. \\ normally just forces a line break at that point without ending the paragraph, and then arranges that the extra space specified in [] is added after that line break.

8:54 PM

@barbarabeeton I hope to prepare the article over the weekend and send it to you on Monday morning.

@yo' My third one (the one with no real notes) is much the same: have to work out what to say!

@JosephWright that's an issue, too, here :)

@DavidCarlisle so adding two \ [2mm] at the end of a paragraph is the same as vskip 2mm on the line following a paragraph?

I mean the double \\\

@JohnMolokach never put \\ at the end of a paragraph, it screams underfull hbox badness 10000 to the log and makes horrible output. Never use \\\\

@JohnMolokach in an ideal situation you would never have any vertical space specified in a document as the default spacing specified by the document class will be correct, that ideal can't always be met but that should be the aim. there is an answer on site explaining what goes wrong if you put \\ at the end of a paragraph, i'll seel if I can find it..

@JohnMolokach tex.stackexchange.com/questions/82664/when-to-use-par-and-when (not the one I was looking for but probably near enough)

Christian Hupfer

user image

Yeah, craters named Spock and Skywalker ;-) But please don't use Kirk -- Pluto is not that bad to get a crater named J.T Kirk ;-)

9:12 PM

@ChristianHupfer Don't you like Kirk? But more importantly there's a Tardis Chasma :)

Christian Hupfer

@clemens Well, I don't like the Kirk Impersonator ... eh... William Ratner or something like that ;-) I definitely like Jean - Luc Picard or Captain Benjamin Sisko ...

2 hours later…

10:44 PM

@ChristianHupfer In Boston Legal Shatner was great.

Christian Hupfer

@clemens Never watched that series :-(

« first day (1732 days earlier) ← previous day next day → last day (3208 days later) »

Transcript for

Jul '1529

$TeX - LaTeX$ 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