@ChristianHupfer What interests me is only the text, the notes to the right or left margin, no special decorations. The characters that you you can review within the pdf files are times bold, italic, etc. written in LaTeX. Certainly 100% formulas are in LaTeX like everything else. It is not urgent at all.
@ChristianHupfer I'll try as well but unfortunately I'm very busy with my students at school. I can not even find free time for me. Thank you so much. If you can not let me know, you stay and you'll be a friend. Hello
@DavidCarlisle, Hi Torbjørn, David. Excuse me for my delay of the greetings. Gooddddd morning
This is LuaTeX, Version 1.0.4 (TeX Live 2016/W32TeX)
restricted system commands enabled.
(./test.tex
> \mathitalicsmode=\mathitalicsmode.
l.1 \show\mathitalicsmode
@DavidCarlisle There are no restrictions on the link because all have access. However I was interested in building a source with the symbols of the previous image and with this that I am attaching. I am very busy at school and my students make me angry. They tell me what are the bushes equations in life?
@DavidCarlisle Probably worth checking the L3 test suite: I get some failures that need looking at. For example, in m3box003 test five (splitting a box) gives different results again: I suspect Hans and the 'what does a \vsplit mean' business again!
@DavidCarlisle Indeed, looking more closely that's the one worrying change:
+ Underfull \vbox (badness 10000) detected at line 63
+ \vbox(28.45274+0.0)x469.75499, direction TLT
+ .\hbox(4.30554+0.0)x469.75499, glue set 444.75497fil, direction TLT
+ ..\localpar
+ ...\localinterlinepenalty=0
+ ...\localbrokenpenalty=0
+ ...\localleftbox=null
+ ...\localrightbox=null
+ ..\hbox(0.0+0.0)x20.0, direction TLT
+ ..\OT1/cmr/m/n/10 a
+ ..\penalty 10000
+ ..\glue 0.0 plus 1.0fil
+ ..\penalty -10000
+ ..\glue(\rightskip) 0.0
+ .\penalty 150
+ .\glue(\baselineskip) 5.05556
+ .\hbox(6.94444+0.0)x469.75499, glue set 464.1994fil, direction TLT
is there any lyx functionality to detect citation/cross-reference boxes that have text immediately adjacent, instead of a one space separation? I googled a bit, couldn't find anything.
@DavidCarlisle I think I will add (if it isn't already there) an issue to hyperref regarding links with unicode (non-ascii)-letters. I see more and more questions and there is no real solution. E.g. tex.stackexchange.com/a/355142/2388
@wilx If I get it finished I might put it on ctan or at least github, currently the combining characters that are supposed to control skin colours and map countries etc don't work
This is TeX, Version 3.14159265 (TeX Live 2016) (preloaded format=tex 2017.1.29) 22 FEB 2017 14:59
**\relax\halign{\valign#\cr\cr
! Emergency stop.
<template> \valign
<to be read again>
\cr
<*> \relax\halign{\valign#\cr\cr
(Sorry, Enrico, but DEK didn't have time to fix this
issue, so interwoven alignment preambles are not allowed)
No pages of output.
I have a horrible feeling @UlrikeFischer's going to push me into opening up the ho-tex code again. I wonder of a reasonable alternative would be to give her write access to the ho-tex github:-)
@DavidCarlisle Well with pdflatex you can't typeset the utf8 version in a sensible way. With xelatex/luatex this would be a sensible solution but one at first need to get the %-encoded (or decoded) version first. Do you know a good lua encoder? All version I can find %encode also the colon and the slashes.
@UlrikeFischer there should be utf8 inputenc support for cyrillic? writing a % encoder in lua (or tex) can't be that hard as it just % encodes each byte of the utf8 separately. Although to be honest I'm surprised it's needed these days most http systems will do the % encoding before passing the http over the wire. It's been a long while since browsers expected % encoded url in hyperlinks or the location bar.
@JosephWright I must admit the whole "rebase" notion never quite sticks in my head, what happens if you don't rebase that does not happen if you do?
@DavidCarlisle OK, the key here is that Git doesn't have a concept of 'revision number', rather we have a hash which is based on the changes made and their parent commit(s)
@DavidCarlisle If we rebase, we effectively move the commit to apply to the head of our history so it may get a new hash (new parent) but we keep a single order to the commits: matches with the way SVN works
@DavidCarlisle If we don't rebase, we can either deliberately make a merge (force one in every case) or can have a 'might be a merge' situation (depending on what has happened in the main line)
@DavidCarlisle As we need to mirror back to SVN, and it doesn't really work this way, it's best to rebase (there are other reasons people do that: philosophy of the history)
Thanks. it almost sounds reasonable (and isn't actually new information) but still usually just comes up when I'm following some suggested git sequence and I just shrug and do what's suggested. I'll feel I understand it if I ever think of doing it myself:-)
@DavidCarlisle Key for us is that the GitHub change makes it easier to take pull requests: we can stick to 'our' workflow of a single history without needing to do much by hand
@DavidCarlisle Sure there is utf8 support. But not inside a \url command. And there is no problem to paste https://ru.wikipedia.org/wiki/SLAM_(метод) in some browser. But hyperref can't handle it (with luatex): \href{https://ru.wikipedia.org/wiki/SLAM_(метод)}{\nolinkurl{https://ru.wikipedia.org/wiki/SLAM_(метод)}} prints the correct link, but the link itself is missing the non-ascii part and so doesn't work.
@DavidCarlisle If you want a 'single' history with no branches or merges, always rebase when dealing with these things
@DavidCarlisle As you can force an update to GitHub and 'alter' the history, it's possible to fix mess-ups in the 'logic': that's more tricky for the team set up as the SVN side would not like it!
@egreg features not bugs, eg now you can typeset in TU encoded variable width lm typewriter and lm dunhill. How have you managed without that previously?
@CarLaTeX Apparently, the Duck content does it make great. So I should add an image of a duck in many of my answers to get at least one upvote for it? ;-)
@DavidCarlisle Imho typesetting utf8-urls with pdflatex is the hardest bit -- even more if it should work with various encodings. This here doesn't look very promising (from url.sty):
% *Some* latin-1 or windows characters. Most will pass through safely to T1 enc
% mu inverted-? OE oe
% lots more to add, but I should later link into inputenc
\@namedef{Url@T1encSpecials}{\do\^^b5{\mu}\do\^^bf{\mathchar190 }%
\do\^^8c{\mathchar215 }\do\^^9c{\mathchar247 }\do\^^df{\mathchar255 }%
\do\^^ff{\mathchar184 }}
But I think I will look a bit at the \href{https://ru.wikipedia.org/wiki/SLAM_(метод)}{\nolinkurl{https://ru.wikipedia.org/wiki/SLAM_(метод)}} thing in luatex.
@PauloCereda Monty Python... I think the sketch with the superior human being, Harry Fink -- who can't come ;-) Eric Idle is playing the role of the sleazy master
@Sebastiano you can ask any question you like although it's always best to include a complete document showing how far you got, and don't ask people to crack password protected files
@DavidCarlisle I want to build notes for work of the foundations of modern physics. A course that I am attending college as a professional and cultural update. You can also see in my profile that I am a teacher.
@DavidCarlisle Where? utf8-url with pdflatex or the luatex problem? The link-problem with luatex is due to the use of \pdf@escapestring. I wonder if the implementation in pdftexcmds is faulty or missing something. What should \pdfescapestring do with unicode chars?
@egreg I think there is a bug in the luatex implementation. For "áéäüметод" I get with pdflatex with \pdf@escapestring \303\241\303\251\303\244\303\274\320\274\320\265\321\202\320\276\320\264` but for luatex only \341\351\344\374. (with and without \detokenize).
@UlrikeFischer if you are in the mode where it can use unicode in pdf, then just pass them through, if you are in the classic mode where it has to use the special pdf text encoding theer is not really anything useful you can do
@UlrikeFischer saw the comment, not had any time to look what it's doing yet
@JosephWright yes although for the particular case of % encoding url that actually simplifies things since you just want the hex values from bytes of utf8 encoding
@DavidCarlisle Probably not and certainly not in the main tests using TU: have to be added to the TU-special tests which I've yet to add to the 'main run'
Running checks on
tu-asciiquotes01
tu-ijcomposites01
tu-lmfam01
tu-lmtt01
tu-texligs01
tu-tlb4500
All checks passed
@DavidCarlisle yeah, emacs user would use M-x lamentation
@UlrikeFischer That WinEdt (at least the version people use at my department) spells out accents, for instance ů becomes \accent23u. Guess what? It breaks T1 encoding.
(sorry, I'm really getting mad of this, it's a recurrent problem that's appeared once more today after some longish discussion I didn't even want to be part of)
(another point is, I'm the LaTeX expert who shall help people, but the people are not willing to be helped)
@UlrikeFischer yeah. "Customable" is a forbidden word. Also, if this gets customized, I'm 100% sure someone will come and say "but you broke this my 163-year-old document".
I remember that a long time ago, when I was using Winedt (V5.0 I guess) it automatically replace some characters. For example, ç was replaced by \c{c} when I open the file.
Now, I need to do the same again. But I have Winedt 7.0 and I don't know how to translate all the accented characters autom...
@StefanKottwitz yeah, I know :-) I read some blogs occasionally, by one copywriter and one language corrector, I realized their misery is similar to mine...
@egreg yes i just noticed that, it's not a clear win to declare. If the font has a working composing character it doesn't help as the composite gets formed anyway. If the font has the composing character but not the composed versions declaring the composite will give missing glyphs in luatex, so it only helps if the font has the composite but not the composing character
One reason I'd never want to be a UX expert. (Actually, I play that role here to an extent. The answer by this very person who returned into the department was: And have people voted over it?)
@egreg @JosephWright luatex puts a dot more or less under in the first two cases, but not really in right place, xetex uses same glyph in all cases as far as I can see. I just looked down the list there are dozens of NFC pre-composed dot under composites, do I really need to declare them all :(
