@Alenanno Nevermind, the centernot package was apparently the solution I was looking for

@IrregularUser I was about to post that ahah

\documentclass[margin=10pt]{standalone}

\usepackage{amsmath}
\usepackage{centernot}

\begin{document}

$A \centernot\impliedby B$

\end{document}

@Alenanno Thanks for the help though and for taking the time to look :)

@IrregularUser No prob!

I got a bug report for the uspace package I have created. Wow!

Basically, the complaint is that as is THIN SPACE currently defined, it will introduce a space at end of line.

This is the current definition:

% thin space
\newunicodechar{ }{\leavevmode\,\linebreak[0]}

Maybe I should use a different definition? \hspace{0.16667em} instead?

@Alenanno \pdfnormaldeviate is a pdftex primitive that gives you an integer so you can scale that to whatever range you need

@wilx don't you want \linebreak[0]\, ?

@DavidCarlisle Doesn’t that just move the problem to the next line?

@wilx imho yes. It is not an unbreakable space, and so should disappear at line breaks. U+202F is the unbreakable small space.

@wilx no if it breaks at the penalty, the kern will be discarded

@DavidCarlisle I see, yeah, it seems to fix the issue.

OK, \newunicodechar{ }{\leavevmode\linebreak[0]\,} it is.

Does anyone have some informal presentation advice? imgur.com/a/qjd5Q < re this table... I'm trying to set this information out in word but having a really bad time with it... I'm just wondering if maybe there's a better approach to presenting this info? Sos it's kinda off topic :/

@baxx do you need any of the vertical lines?

oh no they're just there for visual, i would remove them ofc

i'm doing it horizontal instead at the mo

see if it's slightly less awful

do you think this is a more natural way to present? imgur.com/a/kk1K7

@baxx certainly better than the first

think that's what i'm settling with today :S thanks @DavidCarlisle

@JosephWright @barbarabeeton I think Barabara's \genfrac woes are more an issue with xetex than with the OpenType Font spec, mail sent to xetex list tug.org/pipermail/xetex/2016-October/026837.html

I got another issue (same GitHub issue) with PUNCTUATION SPACE because \hphantom creates unbreakable space. I have searched and this is what I came up with:

\RequirePackage{calc}
\newlength{\uspacePunctSpaceWidth}
\setlength{\uspacePunctSpaceWidth}{0pt}
\newunicodechar{ }{%
  \leavevmode%
  \setlength{\uspacePunctSpaceWidth}{\widthof{.}}%
  \hspace{\uspacePunctSpaceWidth}}

It seems to solve the issue. What do you guys think?

@wilx looks very odd:-) hang on...

@wilx should be , not . I'd use \hspace{\fontcharwd \font `,}

@DavidCarlisle Width of comma instead of full stop?

@wilx yes and not using calc or measuring a box

@DavidCarlisle Thank you.

lol, I think I have bitten more than I can chew: ``\\allowbreak`` expands faster than ``\\linebreak[0]`` (but is not equivalent) · Issue #3 · wilx/project-uspace

@wilx I was going to say before I'd just use \penalty\z@ not \linebreak[0] but didn't as i should try not to stop people using official latex syntax:-) but I'd probably use \allowbreak here, to the end user it looks like a simple space character it shouldn't be trying to do too much conditional clever re-arrangements

@DavidCarlisle Seen that

@JosephWright yes thought you would have, wasn't sure if barbara was on the list though:-)

@DavidCarlisle NOt sure what to do with that... :P

Hello by the way.

@Alenanno pgf and fp packages have wrappers to provide random integers in a given range I think (@JosephWright probably knows the details:-)

@DavidCarlisle Ah yes, but they repeat values as well.

So if I did \pgfmathparse{int(random(1,10))} \pgfmathresult in a loop, I might get duplicates.

@Alenanno easy to avoid repeats (I have an answer on site somewhere I think) just keep doing it until you get a new one

@DavidCarlisle Yes I think I saw your answer, was it the one with "\prunelist"?

@Alenanno no idea:-)

@DavidCarlisle :D ahah well, it was an answer to a similar question.

@Alenanno basically if each time you get a \pgfmathresult you do \@namedef{foo\pgfmathresult}{} then when you do another one if \csname foo\pgfmathresult\endcsname is defined then you have seen it before so call rand() again

@DavidCarlisle How do you check if it's defined?

@Alenanno \@ifundefined :-)

@DavidCarlisle Ah duh, of course.

Thanks.

> Brexit hikes price of all Apple’s Macs in the UK by up to 25%

:(

@PauloCereda what a surprise... :-/

@DavidCarlisle -- thanks for pointer. barbara was not on the list, but has joined.

@PauloCereda The line 'they have to' is amusing: I'm sure they were still well in profit at the old prices!

@JosephWright Indeed.

@PauloCereda you could always buy a windows box instead

@DavidCarlisle and install Linux.

@PauloCereda you'd miss ms paint

@DavidCarlisle in 3D? :)

@PauloCereda 3D paintbrush? :)

@yo' theverge.com/2016/10/26/13380524/…

although even with classic paint one can obtain stunning 3D effects:

@DavidCarlisle holy 3d cow!

@David you need to call insurance on that box. it has been damaged on delivery... :D

hi!!

@AaronHall Aloha!

🙇

@PauloCereda Shakira didn't help :-( en.wikibooks.org/wiki/Talk:LaTeX/…

@Johannes_B Oh no!

@Johannes_B the discussion about h! just above that is completely wrong:-) and the discussion of morefloats on the actual book page is outdated/wrong

Hey guys

@Alenanno Hola!

@PauloCereda So... I apparently reached the Main Memory Size limit of 5000000. XD *looks into externalization*

@Alenanno Ducks don't have memory problems. :)

@PauloCereda Like elephants?

@Alenanno Oh quite the opposite! Elephants do remember, we ducks remember nothing. By the way, what were we talking about? :)

@PauloCereda That you owed me 5000000. :D

@Alenanno I don't remember that. :)

@PauloCereda Of course not. :P

@Alenanno or luatex (or look for infinite loops:-)

@DavidCarlisle Uhm, never used it. But would switching from Xelatex to it imply a lot of changes?

@Alenanno possibly, possibly not. But it does dynamic memory allocation....

@DavidCarlisle Thanks, I'll have a look then.

Just returned from a nice meeting, about 20000 people celebrated the Republic Day on the Old Town Square, Prague.

@barbarabeeton what is from the view of unicode the logical font for the operator names in math? unicode-math uses the textfont (\mathrm} but it looks wrong to me, I would use the normal textblock of the math font.

\documentclass{article}
\RequirePackage{unicode-math,xcolor}
\setmainfont{Arial}[Color=red]
\setmathfont{Stix Math}
\begin{document}\pagestyle{empty}

$\lim x = 0$ \quad   $\mathop{\symrm{lim}} x = 0 $
\end{document}

(textfont and color is a bit dramatic to show my point ...)

@UlrikeFischer -- i agree that the operator names should be the same as the main text font; i can't vouch for the unicode committee members, but suspect they would think the same. (i'll check.) i personally wouldn't set an article in sans serif (nor can i think of any math journals that follow that convention), but it's certainly the usual practice for slides. so i think \symrm isn't the answer here. (where did you see this?)

@barbarabeeton the sans serif (and the color) is only to make the difference more pronounced. And it sounds as if we are differing. I wouldn't use the surrounding text font but the chars from the math font. At first as operator names are so obviously someting "mathematical" and so should be set with a math font and at second as the upright chars from the math font will fit better.

@UlrikeFischer -- the only manual i have readily available only distinguishes between "roman" and "italic". since sans serif in math was until relatively recently vanishingly rare, i interpret that to mean essentially "upright" vs. "sloped". i do happen to have in my possession a physics book from 1979 in which the figures are set in sans; there, "sin", etc., are also in sans, whereas they are in regular roman in the text. a matter of consistency. (i've sent the question to my unicode contacts.)

@UlrikeFischer that's probably a side effect of mathrm having changed from its previous uniocde-math meaning which is what is now symrm

@UlrikeFischer Note that there are math fonts that do not have rm characters (think Euler)

@barbarabeeton I'm really sorry but I by using a sans serif text font I seem to have set you on a completly wrong track. Please assume that the "lim" in red is in latin modern:

\documentclass{article}
\RequirePackage{unicode-math,xcolor}
\setmainfont{Latin Modern Roman}[Color=red]
\setmathfont{Stix Math}
\begin{document}\pagestyle{empty}
the roman limes blblb alblb

$\lim x = 0$ \quad   $\mathop{\symrm{lim}} x = 0 $
\end{document}

@DavidCarlisle I don't want to know the technical implementation but if the logical decision is the right one. Which chars would you use for lim e.g. in a math section in word?

@yo' also unicode/open type math fonts?

@UlrikeFischer word? no idea but I think unicode-math, having introduced symrm should chaneg \operatorname and pre-defined operators like \lim to use that not \mathrm

@UlrikeFischer well, if Euler were a unicode-math font, what should be it's rm version?

@yo' you'd have to ask Euler

@DavidCarlisle LOL

@UlrikeFischer -- the understanding when putting together the unicode math alphanumerics was that the "roman" would be the same as the text font. there are some instances where sans serif has a particular meaning, but that's usually for single letters. so i think what i said before was consistent, and it doesn't matter whether the text is sans serif or latin modern. functions should match the text.

@barbarabeeton i.e., as I would put it, these two shall look the same: sin\,$x$ is $\sin x$.

@yo' Well if there were a math font it should also have glyphs at the a-Z position. There is an abandoned project: github.com/khaledhosny/euler-otf which made this rm:

The x looks a bit odd compared to the rest of the symbols.

@yo' -- yup. gotta go -- have an appointment. maybe back later.

@UlrikeFischer because you use Euler as if it were a text font, but it's a math font, and lim written in it will always look like l times i times m

@barbarabeeton @yo' @DavidCarlisle: sound as if there are contradicting opinions ;-). If I have a full math font my first instinct would be to do all math with this math font. If the functions follows the text font, then why not also the numbers, "=" and "+" etc?

@UlrikeFischer the entire math alphabet mechanism in latex is built on the idea that math alphabets do not follow the text font. (you may of course set the math roman font to be the same as the main text font but that's a different issue)

@UlrikeFischer note that eulervm gives you both options, by default it uses the text figures, you can choose Euler figures by a package option

@yo' No I did load the open type euler as math font, and I didn't complain about the lim but about the x which imho is to light and should be a bit higher (but the project has been abandoned, so there is nothing to gain to discuss the faults in the font).

@DavidCarlisle Yes, that's why I think the operator font should follow the math and that unicode-math made either the wrong decision or forgot to adapt the font setup in this place too. But @barbarabeeton seems not to agree ;-(

@UlrikeFischer I suspect Will forgot (we could ask him:-)

@UlrikeFischer I think Barbara was answering a slightly different question as to whether you should make the operator font the same as the main text font (eg sans serif in sans serif documents) which is a question of style not of tex setup

@DavidCarlisle Well she said "and it doesn't matter whether the text is sans serif or latin modern. functions should match the text." And I do understand the style problem, but imho the default starting point should be a consistent math setup.

@UlrikeFischer then open type euler is wrong. The x is a CM/LM x, not an Euler one, I'm not sure how it belongs there....

@UlrikeFischer I wouldn't like to disagree with @barbarabeeton so choose to take an optimistic interpretation of what she wrote.

@DavidCarlisle ;-)

@barbarabeeton,@DavidCarlisle, @UlrikeFischer Isn't the fundamental issue that 'math fonts' are really a sticking plaster. In handwriting there is just the hand the work is written in, and as fonts are idealised handwriting the same should follow. But we have few fonts that offer a range of math glyphs so we end up with the idea of 'math fonts'

Of course, that doesn't help with the practical matter at hand ...

@JosephWright no I don't think so, not in maths, in maths "fonts" are really "alphabets" and roman v italic has same status as latin v greek

@DavidCarlisle Yes but written by hand they'd still match: you wouldn't have two separate scribes writing the text and the maths

@DavidCarlisle What I mean is that we allow mixing one math mode font with a non-identical ('designed together') text font only because most fonts don't have the math glyphs. In an ideal world this would not be the case.

@JosephWright in an ideal world you give the choice. I chose, for instance, Palatino for text, but not for math (I could, it exists!), I chose Euler there because I love the combination. Also note that my personal script is significantly different for text and math!

@yo' On the last point, you are not I suspect a mediaeval scribe :)

@yo' On the former, I guess I'm saying I don't actually think you should have that choice (easily): it's essentially the same as mixing glyphs from arbitrary different fonts between say different words in prose.

@JosephWright no, I'm not

@yo' :)

@JosephWright no, it's the same as choosing completely different font for text and for titles

@yo' We are going to disagree here: to me, maths as symbols is simply maths as words shortened down :)

@JosephWright ummmmmm no :-P

But maybe, yes, however, then there should be both \setmainandmathfont{Pagella} and \setmainandmathfont{PagellaAndEuler}

@yo' Like I said, I did start with 'in an ideal world'. As we all know, it's hard enough getting anything reasonable sanserif in math mode ...

@JosephWright well, the only case that really works is "CM/LM everywhere"...

@JosephWright Yes, this a part of the problem. But naturally people can also choose to use a different style for the math -- as they use a different font for code or quotes. In both cases we will have to think about the border cases. Should a number (an SI unit) follow text or math etc.

@UlrikeFischer SI unit follows the same convention as an operator of course: text font.

@yo' Yes, but siunitx also sets the numbers in text font. Should math do this too?

@UlrikeFischer yes

@yo' Didn't you just wrote that your math script is different from you text script ? ;-) I would agree with you if the text font has a matching math font. But if they are different I would in most cases (at least in displaymath) value the consistency of the math style more and not exchange a number of symbols only because they match the text.

Talk for tomorrow finished. :)

@UlrikeFischer sorry I wasnt quite precise, its the variables that i write usong the math script where o have a vwry distinguished shape of all the letters. For things like "lim" and "ln" i use my text handwritimg

(sorry for the typos, im walking back home to go to bed)

@DavidCarlisle Contributors very welcome.

@Johannes_B sorry that's not a slippery slope i plan to stand on, have picked up far more tex stuff than I can cope with without editing wikibook as well:-)

@DavidCarlisle That is very true, and we all are thankful for that.

@DavidCarlisle Which part exactly is completely wrong? It needs rewriting.

@Johannes_B oh the morefloats thing is wrong as latex has 52 floats not 18 to start with and \extrafloats lets you have as many as you want (and so morefloats isn't needed)

@DavidCarlisle I meant the h! stuff.

@Johannes_B ah that was just on the talk page I think wasn't it? (it was all wrong as I recall:-)

@DavidCarlisle I think i am going to create a new page on floats, move what is needed from other pages and remove that stuff from that pages.

@DavidCarlisle I removed the center environments in the code examples. I didn't think about checking the text.

@DavidCarlisle Can i ask your permission to completely remove the wrapfig part? Nobody will argue with DPC :-)

@Johannes_B well you don't need my permission:-) I don't really know the wikibook, wrapfig is a pretty good package and should be somewhere although it doesn't necessarily need to be mixed in with float descriptions

@DavidCarlisle That page is about figures, not figures. The page on tables introduces table and the whole placement stuff again.

@Johannes_B as I say I really don't want to be tricked into getting an overview of the whole book:-)

@DavidCarlisle Alright. I stop bugging you about it. You have other stuff to worry about.

@DavidCarlisle Is longtable incompatible with floatrow at the moment?

@cfr shouldn't be, why?

@Johannes_B I have a tikz-pgf badge to get for example!

@DavidCarlisle It seems to be. I tried answering a question about problems with threeparttablex and floatrow, but I can reproduce the problem with just floatrow and longtable (and showframe).

@DavidCarlisle Paint will get you there ;-)

@Johannes_B sounds a good plan

@cfr mwe?

\documentclass{article}
\usepackage{longtable,showframe}
\usepackage{floatrow}
\begin{document}

\begin{longtable}[l]{lccc}
  \caption{Caption which must extend to quite a long thing if we're going to see the problem clearly.} \\
  competence 1 & \textbf{0.75} & -0.14 & 0.04 \\
  competence 2 & \textbf{0.84} & -0.06 & -0.07 \\
\end{longtable}
\begin{longtable}[c]{lccc}
  \caption{Caption which must extend to quite a long thing if we're going to see the problem clearly.} \\
  competence 1 & \textbf{0.75} & -0.14 & 0.04 \\

Without the alignment option, it looks like the first case more-or-less. Using c makes it appear as I'd expect the default. But using l or r doesn't work well. If I create a dummy fr-longtable.sty, I get the same problem.

@DavidCarlisle ^^^

@cfr it's same without floatrow isn't it?

@DavidCarlisle Hmmm. Yes, it seems so. But removing the [l], [c], [r] works in that case, whereas that's not really an option of floatrow is loaded.

@cfr looks like the caption is aligned on the page not the text block, does LT really do that??

@DavidCarlisle Not by default if floatrow isn't loaded. Well, it may be. I think it is just masked by the centring still being centre.

@cfr no I mean if [l] is used (and no other package)

@DavidCarlisle Yes. Or if r is used. At least, it seems to?

@cfr reading longtable.sty now, who wrote this stuff...

@DavidCarlisle ;)

@DavidCarlisle You might want to investigate Parfit's views on personal identity.

oh it's tex:(

spot the difference

\documentclass{article}
\usepackage{longtable,showframe}
%\usepackage{floatrow}
\begin{document}

\begin{longtable}[l]{lccc}
  \caption{Caption which must extend to quite a long thing if we're going to see the problem clearly.} \\
  competence 1 & \textbf{0.75} & -0.14 & 0.04 \\
  competence 2 & \textbf{0.84} & -0.06 & -0.07 \\
\end{longtable}


xxx

\begin{longtable}[l]{lccc}
  \caption{Caption which must extend to quite a long thing if we're going to see the problem clearly.} \\
  competence 1 & \textbf{0.75} & -0.14 & 0.04 \\

@cfr I need to work out exactly when ^^^ happens probably not tonight

@DavidCarlisle Thanks. I thought for a minute you were saying that all uses of longtable should include a 4in rule .... And it isn't even metric!

@cfr that's the default caption width....

@DavidCarlisle Oh.

@cfr Oh I see...

@cfr basically longtable isn't too happy if (a) you use l or r and (b) the caption is wider than the natural width of the table without the caption. It's probably been that way since about 1990 and no one has ever complained before now:-)

@DavidCarlisle Oh, I see ...

@cfr fixable but usual concerns about a few decades of legacy stuff and too late to think about that tonight.

@DavidCarlisle Fair enough. It's not my document you've broken :-).

@DavidCarlisle Anyway, the OP has a workaround since I suggested using c. Since they wanted centred longtables anyway, that should work OK.