@Skillmon -- I'm directing this to you, since you posted the only answer (so far) to the question Spacing after period at end of inline math. The OP and I have gotten into a rather niggly discussion in comments about why it's impossible to avoid manual intervention. Is it worth fashioning this information into an answer? If you say it is, I will.

@UlrikeFischer not yet, but base passes and I'll build up graphics on that over the weekend, that's the plan at least

quack

Dinner!

oh no

@DavidCarlisle at this time?

@DavidCarlisle if there is something to test on windows (which you don't have) and miktex say so ...

@UlrikeFischer he is mean. :)

@barbarabeeton yes you might, after all you have much more insight in this.

@barbarabeeton and thanks for stepping in, he didn't believe me at all when I wrote him that there is no way in TeX to do this automatically.

How dare people doubt mr. rabbit?

<3

@PauloCereda Contrary to common believe, killer rabbits are fallible.

@Skillmon oh no

@PauloCereda I'm good at programming useless stuff in this language called TeX, but my knowledge of the inner workings is still limited, I'm afraid

@Skillmon <3

@PauloCereda and the last couple of years (since I got so involved with TeX), all my other programming skills started to decay.

@barbarabeeton @Skillmon see tex.stackexchange.com/a/91373/2388 for an extended discussion.

@UlrikeFischer that \noic implementation is interesting, thanks for that link. Also this gives good spacing for the problem at hand. Care to post a comment/answer at tex.stackexchange.com/questions/514326 ?

@UlrikeFischer these work so far

\documentclass{article}
\usepackage{graphicx}

\begin{document}


1 \includegraphics{a.b.png}

%2 \includegraphics{a.b}

3 \includegraphics{foo bar.png}

4 \includegraphics{foo bar}

{
\graphicspath{{nospace/}}

5 \includegraphics{a.png}
}

{
\graphicspath{{space here/}}

6 \includegraphics{a.png}
}

\end{document}

@DavidCarlisle will \includegraphics{{a.b}.png} work, too? (sorry if I'm annoying)

@Skillmon currently no, I'm wondering if it should, or if it should work anyway (that is try adding the default extensions before reporting a missing file) or try adding default extensions if the extension it found (.b here) is not known.

I wonder if @egreg is watching the rugby

@DavidCarlisle just because it used to work and I used it in some of my old documents as a crude hack...

@DavidCarlisle @egreg has to look women hockey in the afternoon: 2/3 November: Germany – Italy, Mönchengladbach

@Skillmon you used to need {} to hide the first . so for example {a.b}.png would be seen as .png but now it handles braces there by throwing them away so {a.b}.png works like a.b.png (which works) but that leaves the default extension case... I'm leaning towards always trying the default extensions if the file not found just in case you have a.eps.png and use \includegraphics{a.eps} ....

@DavidCarlisle imho in the long run all this braces tricks are a problem if file names are moved around, so it is better if it works without them.

@UlrikeFischer currently a.b (if a default extension needs to be added) is quite consistent, it doesn't work with or without braces (you could as always use the graphicx ext= key but if you are doing that you may as well write the filename as a.b.png so I think trying the extensions before giving the error is the thing to do....

@UlrikeFischer can I have a local space here/a.png in an l3build test? support only copies files doesn't it? (I'll check later, but thought you might know:-)

@DavidCarlisle what if there is a file a.eps and a file a.eps.png and I specify a.eps because I want to include that file and then a.eps.png gets included?

@DavidCarlisle or what if I'm extraordinary stupid, and have a file which really is called {a.eps}.png?

@Skillmon no only planning to try extensions before giving the file not found error, so if the filename as given is there then it will work, but there are edge cases with scripts tha make files on demand

@Skillmon you get what you deserve

@Skillmon but only the first {} is discarded and only for compatibility with existing uses so {}{a.eps}.png or {{a.eps}}.png would both work (note this is in the format and applies equally to \input it is not in the graphics handling

Machine user :)

@PauloCereda the latex one?

@DavidCarlisle I think it copies only files, but you could probably use the texmfdir variable. Or perhaps use a path like ../../support/space here/a.png.

@DavidCarlisle Clinic for young basketball referees, this morning. :-)

@DavidCarlisle my husband sent me this link for you: 78.media.tumblr.com/ac0711610bf94147e06741a1ff5834f9/…

@UlrikeFischer yes I thought about those too but if I set up texmfdir it (should) work anyway so i can't test \graphicspath, and "knowing" where l3build builds stuff so setting a path back to the source tree seems immoral:-)

@UlrikeFischer it was lucky I didn't watch the match

@DavidCarlisle hm. so either support already copies a folder or @JosephWright will get a new feature request ;-)

@JosephWright any ideas on best way to l3vuild check \graphicspath{{space here}} ?

@UlrikeFischer I must say I thought we had done \include{foo bar} when we did `\input, just slipped past, it seems.

tugboat just dropped through the door

@DavidCarlisle spaces in files names are so far away for me, that I never really test if they work or not.

@UlrikeFischer you are so old fashioned. Next you will be saying that you run latex by typing latex file.tex a commandline not pressing some control-shift-altgr-F5 key shortcut.

@UlrikeFischer My colleagues not only put spaces in file names but also -, ° and any other special char Windows accepts in file names!

I'm very modern here: I use ctrl+alt+p for lualatex, ctrl+alt+ü for luatex, ctrl+alt+ö for luahblatex, shift+crtl+p for pdflatex, shift+ctrl+a for pdftex, ctrl+alt+x for xelatex, ctrl+alt+l for latex. I hope they don't invent a new engine ...

@CarLaTeX Windows has way more restricting file names than any other OS typically used today, just saying...

@UlrikeFischer I wonder if it's possible for me to type ctrl+alt+ü from this keyboard? At least I can type z

@DavidCarlisle Found \begin{filecontents*}{\jobname.tex} in an overleaf document. Could even work ...

@UlrikeFischer you should try testing all possible outcomes of the iftex tests....

@UlrikeFischer eek what does that do if the main file is jobname.tex?

@DavidCarlisle I would, but texlive is letting me down: no update yet.

@UlrikeFischer git pull :-) (but you'll need a big keyboard to make all the shortcuts )

@DavidCarlisle well I think as overleaf sets jobname to texput it will create a texput.tex.

@DavidCarlisle ctrl+alt+z seems to be free, so I could extend a bit.

@UlrikeFischer you'd never find it

@UlrikeFischer @JosephWright I think probably I should use runtest_tasks -- A hook to allow additional tasks to run for the tests as then I can copy a test png file to loads of silly space names and space named folders in the Lua without having to have multiple files in the repository.

oh no that runs after the test boo.

@Skillmon Nevertheless, the creativity in file names has no limits :):):)

@DavidCarlisle you could unpack a zip with a folder structure. And imho if there a more than one check runs it should work in the second run. There was just yesterday the discussion with moewe about this.

@UlrikeFischer yes but I could also add another hook before it runs tex so it works on the first run too....

@UlrikeFischer or use this one before all the checks might be best: checkinit_hook = checkinit_hook or function() return 0 end

@CarLaTeX I prefer simple filenames: [a-z][A-Za-z_\-\d.]* for regular files and [A-Z][A-Za-z_\-\d.]* for folders.

@UlrikeFischer checkint is where it does the existing copying from support so adding a folder creation in the checkint_hook would work I think

@Skillmon people don't even think in terms of file names any more, they just click something and end up with Copy of chapter 3 (1) .tex and expect that it just works.

@DavidCarlisle best naming scheme I encountered so far was using Pokemon names and the final version was the last evolution of the initial name (so Charmander was the initial version, Charmeleon the edited one, and Charizard the one which was handed in)

@DavidCarlisle way better than copy of the file-final version -- really final this time - i'll hand it in like this for sure - accepted version.docx (disclaimer: the docx at the end of the file name does not imply that I think every MS Wort user would do this)

@DavidCarlisle There are a few bugs probably about copying files with spaces in ... I've tracked some but not all down cross-platform

@JosephWright I'll add some tests and find out but I have to get the tex code working first so that I think the test should pass...

@Skillmon Me too, complicate names are only source of problems! But not everyone understand it :(

@Skillmon ooh Pokémon

@PauloCereda could you translate ? ;-)

@UlrikeFischer :)

This grid typesetting thingy looks like it could be handy. I.e.

And it's not currently an option in LaTeX? So if you want to do it, you have to use ConTeXt?

@FaheemMitha Grid setting is hard and the LaTeX OR isn't ideal: Frank has some unreleased code that works - ask him about Alice (@DavidCarlisle: we should remind him to actually release it)

@JosephWright LaTeX OR? Alice?

I don't suppose you mean Alice in Wonderland.

Does anyone have a precise definition of grid typesetting they could point me to?

@FaheemMitha OR = Output routine. And well, Alice in Wonderland is a theme in L3 publications…

@FaheemMitha yes

@JosephWright Frank's Alice paper is public somewhere isnt it?

@DavidCarlisle Yes

@FaheemMitha It's hard to have a strict definition. But the gist of it is covered by the introductory section on the ConTeXt wiki.

@FaheemMitha well yes and no. It's also a matter of how you describe things.

@FaheemMitha basically it means that the baseline of every line of text is positioned an integer multiple of the baseline spacing from the top of the page.

@TeXnician Oh. Thank you.

@FaheemMitha which is fundamentally the opposite of what TeX's box and glue model of typesetting tries to achieve....

@DavidCarlisle Oh. So the lines line up across pages?

So, for example the 5th line on every page would be the same distance from the top of the page?

@FaheemMitha exactly, as if you were writing on squared graph paper, "the grid"

@DavidCarlisle And is the layout of the grid user-definable?

And how about the positioning of a particular line on the grid?

@DavidCarlisle TRON

@FaheemMitha where, in context? I assume so, it would be silly to allow a user to be able to specify an 14pt font without being able to specify the grid spacing it fits on. The grid layout is trivial it's just specifying some lengths the hard part is not doing what tex normally does if you make say a display math equation which is just insert whatever space is needed and start typesetting after that without keeping track of the absolute position.

@PauloCereda non

@DavidCarlisle Sure, in ConTeXt. I was thinking generally, but I guess if the implementation doesn't exist, it doesn't make sense to ask about its features.

@FaheemMitha Even with ConTeXt grid typesetting (with math and images and...) did not work perfectly last time I tested (around a year ago).

@mickep Oh. What problems were there?

@FaheemMitha as I say it's not about things existing or not existing, it is about describing a glass as half full or half empty

@DavidCarlisle I don't follow.

@FaheemMitha Text easily go out from the grid, it is difficult to get good spacing around section headings and formulas. It could have been me not knowing how to do it properly, though.

@mickep So there were problems even with ordinary text?

@FaheemMitha it's hard to do and chances are it doesn't always work. You could describe that as "having the feature" and if it doesn't work logging a bug somewhere, or you could describe that as "not having the feature" or "having a partial implementation" It's words not technical details

@DavidCarlisle Ok.

@FaheemMitha Ordinary text after formulas/figures/heads and so on did sometimes jump out from the grid. But, as I said, it might be me not knowing how to do. (I could perhaps try my file again.)

@mickep So the file itself wasn't just plain text?

@DavidCarlisle oh

@FaheemMitha if you just have a 20 page paragraph with no headings and no figures then grid setting is automatic whetever you do

@FaheemMitha The input file was plain text, but the output included images and formulas. If you only have a text document I'm sure there will not be big problems.

@mickep Yes, I meant whether what you were trying to typeset was plain text or not. TeX input is always plain text. :-)

@FaheemMitha is zzzzz $X^{2^2}$ zzzzzzz plain text? (and the difficulty is what to do if that doesn't fit, tex normally just opens up the line by the minimum required to avoid over-printing but to maintain a grid you need to not do that (and over-print the line above) or really open up the line by a full extra linespace, or to change the text and use a more cramped style so it fits or...

So, the important question for today: when do I decide to release an alpha of siunitx v3

@JosephWright you could decide this now, just don't decide at which point in time you do release it.

@Skillmon :)

@JosephWright NOW

@PauloCereda Hmm, tempted: I'm really just wondering whether I can get some more functionality working ... it's certainly improved on alpha-1

@JosephWright I already twitted about... too late. :)

@PauloCereda OK, I'll perhaps do that: needs a blog post anyway

@JosephWright woo

we should have been back on imperial units by now

@DavidCarlisle ooh

@DavidCarlisle is it supported by siunitx v3? @JosephWright

@DavidCarlisle tell me, how many fluid scruple is one cubic inch?

@Skillmon 0.08 cubic inch to the scruple kylesconverter.com/volume/fluid-scruples-(us)-to-cubic-inches

@DavidCarlisle and how much pressure in inches of mercury is inflicted by a pound on a quarter circular inch?

@DavidCarlisle your converter says 13.3, not 0.08...

@Skillmon I was very disappointed that my new(ish) car that I got this year has user settable units for speeds but doesn't offer furlongs per fortnight

@Skillmon I should have realised a reciprocal was a bit much for you to follow, sorry

@DavidCarlisle to put it in the words of my constructions professor: "Even though it's just a minor error, the bridge would've crumbled anyway" (not a great translation, sorry)

@Skillmon Nicht jeder kann ein großartiger Linguist sein

I expect my German is as good as my scruple conversions (given the common source)

@DavidCarlisle you got Kyle's converter to translate English to German?

@DavidCarlisle that sentence was perfect, btw.

@Skillmon no but in both cases I just asked google to outsource the question

Have to do the groceries, have a great day!

@Skillmon that's the thing about using an infallible oracle, you tend to get perfect answers.

@DavidCarlisle No. By plain text, I meant no math. Even inline.

@FaheemMitha section headings, are they plain text?, or paragraphs with vertical space or ...

@DavidCarlisle Not sure.

@FaheemMitha it's your definition so if you are not sure, then no one is:-)

@DavidCarlisle I really had not thought about it that much.

@PauloCereda Blog post coming up

I finally did some testing : siunitx with expl3 is faster than doing things by hand: texdev.net/2019/11/02/siunitx-v3-alpha-2

@JosephWright woo

ooh alpha 2

@bp2017 no, it would be an odd thing to ask and they would say no:-)

@bp2017 hopefully soon the browsers will have access to OpenType Math table based math rendering so will be able to do math typesetting. Adding a latex-like user interface over that wouldn't be something for the html wg or the browser makers it would be programmable in javascript like mathjax

@bp2017 adding classic tex math rendering with 7bit fonts in a custom font format with custom font encodings would be a total non starter for the browser code base, which is Unicode throughout.

Now, will I get any testers

@UlrikeFischer I'm hoping to tempt you to try siunitx v3 ... I'd like thoughts on the font stuff

Probably should bother Frank too

@JosephWright ooh

@JosephWright I did install it. But I have again forgotten how I install also the docu ;-)

@UlrikeFischer l3build install --full

@JosephWright I was just checking the l3build docu :-)

@UlrikeFischer Why bother ... just ping me ;)

@JosephWright Is it intended that the install target of l3build has a description containing "textmf" instead of "texmf" (see GitHub)?

@TeXnician No

@TeXnician Fixed

@JosephWright I wonder if \__siunitx_print_text:n shouldn't use \fontfamily{\familydefault}\selectfont, \fontfamily{\seriesdefault}\selectfont etc instead of \rmfamily and \mdseries. This would imho solve tex.stackexchange.com/questions/378442/….

ooh fixes

@CarLaTeX @egreg ^^^ today in Mönchengladbach

@UlrikeFischer ooh

@UlrikeFischer Oooooooooooooooh

Just got in from fireworks very cold and very wet

@DavidCarlisle You celebrate some anniversary of this old xml package somebody was asking for earlier today?

@mickep no we celebrate not blowing up parliament, in this case using spare money from the Mercedes F1 team

Oh, yes, you are one of the Brexiteers.

@mickep er no, not in my name

@DavidCarlisle wasn’t that the fifth of november? remember, remember, …? or have you not blown up parliament more than once?

@HaraldHanche-Olsen yes but quite often public events move to the nearest weekend

@DavidCarlisle ah, but then it should have been moved to a monday so people can have an extra day off. like they do across the pond, it’s really one of their better inventions.

@HaraldHanche-Olsen we have not blown up parliament lots of times, perhaps we should have fireworks every day.

@DavidCarlisle what remarkable restraint!

@UlrikeFischer v3?

@JosephWright yes, at least if I didn't mess it up. I tried something like this:

\documentclass{article}

\normalfont
\DeclareFontShape{OT1}{cmr}{lt}{n}{<->phvr7t}{}
\renewcommand\seriesdefault{lt}
\providecommand\lseries{\fontseries{lt}\selectfont}

\usepackage{siunitx}
\sisetup{mode=match}
\ExplSyntaxOn
\cs_set_protected:Npn \__siunitx_print_text:n #1
 {
  \text
    {
    \bool_if:NT \l__siunitx_print_text_family_bool
    { \fontfamily{\familydefault}\selectfont }
    \bool_if:NT \l__siunitx_print_text_series_bool
    { \fontseries{\seriesdefault}\selectfont }
    \bool_if:NT \l__siunitx_print_text_shape_bool

@UlrikeFischer I do wonder if all of those currently-fixed font things should be abstracted as keys, so you could do text-family-reset = \fontfamily{\familydefault}\selectfont. But perhaps you are right: would be clear and avoid even more keys

@UlrikeFischer I'll give a little thought and perhaps adjust tomorrow

@UlrikeFischer Does have a nice right to it ...

@JosephWright I don't know if more keys are needed, but imho the main point is that \mdseries or \rmfamily is not always the default series or family.

@UlrikeFischer Fair point: I'll look to adjust

@UlrikeFischer Alpha testing already useful

@UlrikeFischer Take it you like the new font plan

@JosephWright I did only read the code documentation of the print module and tried the two questions, so I don't have a very thorough understanding, but it looks quite logical. Btw: I think e.g. the description of reset-text-series is wrong. It speak of "math series" instead of "text series".

@UlrikeFischer Could be: lots of copy-paste