@UlrikeFischer yes, I saw @DavidCarlisle's and your messages.

@barbarabeeton And Mike Spivak wanted to be paid for LAMS-TeX.

@DavidCarlisle Looking forward to your thoughts on xbox :)

> Japanese Vending Machines Are Selling Bugs You Can Eat

@JosephWright will try to look later

@PauloCereda better than selling poisonous ones

@DavidCarlisle ooh

Hi

I really need help with images in latex

if I create a figure with two horizontal or vertical subfigures and I decrease their size to fit the page or whatever, their resolution will be poor

I want the resolution to be high (i.e. I want the letters to be very well visible as if the image was big)

I am using this code

@nbro if it has text in it, it's usually best to avoid bitmap image formats (but if you are scaling the image down, then resolution _increases so why is it "poor"

why png ?

I don't know. Latex doesn't support svg out of the box

damned latex

what format do you suggest?

@nbro export SVG as PDF, problem solved.

@nbro it does if you use the includesvg package but even without that, convert the svg to PDF so it's still scalable

I once tried that includesvg and I was not able to make it work cuz I was having problems so that's why I gave up on that

damn, I need to write code now to produce svg images... damn!

I really don't understand why latex doesn't accept svg

what a limitation!!!

It should accept them out of the box, no?

@DavidCarlisle because if it contains text, png is much better than jpg.

@nbro latex really doesn't care, it just needs to know the size, but if you choose to input svg and output pdf, then something has to convert svg to pdf, it does not happen by magic.

@Skillmonlikestopanswers.xyz er yes but PDF or EPS would keep it as text which is better still

@DavidCarlisle ooh magic

why input and output? I just want to use an image

latex should just accept svgs in the includegraphis command

@Skillmonlikestopanswers.xyz 🥕

that's what I mean

@DavidCarlisle well, if the choice is limitted to bitmap...

@PauloCereda <3

@Skillmonlikestopanswers.xyz <3

@nbro your image is in SVG your final document is PDF, how do you think the image gets from one to the other?

@DavidCarlisle ooh by converting SVG to PDF?

@PauloCereda good plan.

@PauloCereda but only via converting it to PS first, then to EMF, then back to SVG, then to PDF?

well, latex should be able to convert svg to pdf, in the same way that png is used in a pdf

@Skillmonlikestopanswers.xyz ooh a chain

@nbro PNG isn't used in a PDF. In fact, LaTeX has the hardest time with PNG of all the formats it supports.

@nbro latex accepts any form understood the system you are using, pdflatex understands svg if you use inclusesvg, if you are using dvisvg then it inputs svg

inkscape mygraph.svg --export-pdf=mygraph.pdf

luckily, I saved the data of the experiments and I don't have to repeat the experiments

@nbro which software do you use to get SVG plots?

My question is: if I convert a svg to a pdf, will I get high resolution even if I scale the images?

I need to spend some time writing code to read the experiments' data, so I would like to know if it's worth it

@DavidCarlisle Hope you like my not-\vcenter plan (That one might actually be useful, accessibility-wise)

@nbro as long as the SVG contains vector objects and not pixel objects, the PDF will also contian vector objects

@nbro thing is vector-based, so resolution is not significant, actually...

@nbro that isn't how software works. saying something should happen isn't the same as making it happen, and why should it happen, why svg, not tiff or giff or windows metafile (all of which some latex systems support and some do not) you happen to want svg today but that doesn't mean that the latex executable should incorporate an SVG rendering system that compiles on every platform that supports tex.

Unless you have raster elements in your SVG, then things get tricky...

@nbro svg and pdf are scalable (unless they are just holding a bitmap) so normally there is no reslution involved

@DavidCarlisle ooh Apache Batik

well, I just want my images to look nice and the letters (e.g. the legends, the labels on the axes) to be high resolution and not blurred

@nbro you are blurring them by generating bitmaps, latex is not involved. Do you have to make SVG, can't you save to PDF from the original application?

btw, is it possible to disable the sound notification when someone tags me here? I get a loud irritable sound when someone tags me cuz usually I am listening to music

@DavidCarlisle Yes, I think so. I think I can save to pdf directly

I am using matplotlib

@nbro Depending on your browser you can mute tabs.

Any suggestions to make matplotlib produce images that are meant for research papers and thesis?

@nbro depends on the browser, firefox has a "mute tab" menu item on each tab.

@nbro matplotlib supports saving to PDF, even while using LaTeX to render your labels. And if you use the correct preamble-settings in matplotlib your figures will look just as if they were created only for your document.

@nbro I think I have an answer somewhere about the way I use it, let me dig.

@nbro well generate pdf then, as if you generate svg you just need to convert it in to pdf later if you want it to be in a PDF document. Again that is nothing to do with latex, just a fact of life SVG can't be inlined into a PDF you need to convert it to PDF to include it.

@JosephWright it is odd if a internal command is called vcenter that explicitly tries to avoid to use \vcenter ;-)

@nbro ^^ there's a volume icon in the upper right corner

oh, thanks god

I mean it's nice to have visual notifications, but this sound isn't good

maybe the volume can be decreased

@nbro take a look for example here or here.

@UlrikeFischer :)

@UlrikeFischer I've used hcenter and vcenter as 'long' forms of hc/vc in names generally, so for xbox it means the 'right' thing

@UlrikeFischer Glad you spotted it!

@UlrikeFischer Next is I guess \frame: that really will round out the box stuff

@JosephWright is there really a \frame command? Doesn't that clash with beamer?

@JosephWright funny in a way that the original motivation for coffins was wanting \parbox[t] and \parbox[b] at the same time

% \begin{macro}{\frame}
%
% The following definition of |\frame| was written by Pavel Curtis
% (Extra space removed 14 Jan 88)
% RmS 92/08/24: Replaced occurrence of |\@halfwidth| by |\@wholewidth|
%    \begin{macrocode}
\DeclareRobustCommand\frame[1]{%
  \leavevmode
  \hbox{%
    \hskip-\@wholewidth
    \vbox{%
      \vskip-\@wholewidth
      \hrule \@height\@wholewidth
      \hbox{%
        \vrule\@width\@wholewidth
        #1%
        \vrule\@width\@wholewidth}%
      \hrule\@height\@wholewidth

@UlrikeFischer ^^^

@UlrikeFischer yes it's like \fbox with no fboxsep, but shares no code with it

@UlrikeFischer beamer does all sorts of gymnastics to restore the 'original' \frame, and to deal with environment versus command

@DavidCarlisle It's .. different

@JosephWright a bit more picture mode oriented

@DavidCarlisle Framing is one thing I've heavily changed revised; that would have to be agreed on (basically, I think \fboxrule should not overlap with text)

@DavidCarlisle Yup

@JosephWright ? oh you mean if fboxsep is 0?

@DavidCarlisle I did a version using coffins, but there are two issues: the s inner pos argument, and the fact that \vcenter aligns to math axis not current reference point

@DavidCarlisle Yes: see the comments in \hbox_frame:nnn: I've added the width of the rule to the sep, then used that to work out where the 'content' should be

% \begin{function}{\hbox_frame:nnn, \hbox_frame:nVV}
%   \begin{syntax}
%     \cs{hbox_frame:nnn} \Arg{content} \Arg{thickness} \Arg{border}
%   \end{syntax}
%   Inserts the \meta{content} in a horizontal box which will bear a frame of
%   \meta{thickness}. A \meta{border} is left between the \meta{content} and
%   the frame. The frame will overprint any part of the \meta{content} which
%   lies outside of the bounding box at natural size (\emph{i.e.}~the frame is
%   printed \enquote{after} the \meta{content}). The apparent size of the

@JosephWright can't look much now will look later but the rules don't overlap the content in the original fbox do they?

One needs to submit a package to CTAN named broccoli.

@PauloCereda insert But why? meme here.

@JosephWright oh you mean over rather than under print the rules,

@Skillmonlikestopanswers.xyz dunno :)

@DavidCarlisle Yes

@DavidCarlisle Hmm, I think I still have something wrong ... just tried

and the results were wildly different

@JosephWright second opt arg is alignment not a length?

Enola Holmes

> While reimaginings of Sherlock Holmes are a dime a dozen, the Conan Doyle Estate had a particular issue about this version. The estate filed a lawsuit against Netflix, saying that because the film depicts Sherlock Holmes as having emotions — yes, you read that right — it violates copyright, since they claim that the character only has emotions in the stories published between 1927-1928, which still belong to the estate.

@PauloCereda they should update it so he has emoji instead of emotions

@DavidCarlisle I once saw Shakespeare's Romeo and Juliet book rewritten as a set of WhatsApp messages. It was... unnerving...

@DavidCarlisle Oops

@DavidCarlisle Ah, that was it:

\documentclass{article}
\usepackage{xbox}
\begin{document}
\setlength\fboxrule{10pt}
\setlength\fboxsep{0pt}
\framebox[2cm][l]{Hello}
\end{document}

@DavidCarlisle With the 'classical' implementation, the rule overlaps the start of the text; at present, the xbox one moves the text along so (here) there is no overlap

@DavidCarlisle I've interpreted l as 'Starting at the left side of the rule + sep combo': classical approach only accounts for the sep

@JosephWright in \frame yes not \fbox or \framebox

@DavidCarlisle I'll check my maths again :)

@JosephWright hmm I get

\documentclass{article}
%\usepackage{xbox}
\begin{document}
\setlength\fboxrule{10pt}
\setlength\fboxsep{0pt}
\framebox[2cm][l]{Hello}
\end{document}

which isn't how I remembered it I was sure an fbox was 2fboxrule+2fboxsep wider than the content...

@DavidCarlisle Well yes, that's my point

@DavidCarlisle That's what I've implemented (I hope!) - it seems more natural to me

@JosephWright it's a bug.

@DavidCarlisle Which is: current behaviour?

@JosephWright yes watch:

@DavidCarlisle Teach me to read the docs

\documentclass{article}
%\usepackage{xbox}
\usepackage{color}
\begin{document}
\setlength\fboxrule{10pt}
\setlength\fboxsep{0pt}
\framebox[2cm][l]{Hello}

\fbox{hello}

\fcolorbox{red}{blue}{hello}


\end{document}

\fbox is wider by 2rules and 2 seps, it's framebox that isn't which seems wrong

@DavidCarlisle Drat, I'd forgotten about colour boxes

@DavidCarlisle Ah, right: I guess I'm at the advantage that 'faster' is relative: I'm using the same internals for both \fbox and \framebox

@JosephWright I'll have to look what we say I suppose its OK if the width argument is specified as being the outer total width, but it seems a bit useless, and it means the default is not \width:

\documentclass{article}
%\usepackage{xbox}
\usepackage{color}
\begin{document}
\setlength\fboxrule{10pt}
\setlength\fboxsep{0pt}

\framebox[2cm]{Hello}

\framebox{Hello}

\framebox[\width]{Hello}


\fbox{hello}

\fcolorbox{red}{blue}{hello}


\end{document}

@DavidCarlisle dinoswords.gg

@DavidCarlisle Hmm, I wasn't planning to find a bug: xbox is really mainly about testing xparse (or at least that was the original idea)

@JosephWright there are always "interesting behaviours" to be found....

@PauloCereda That's nuts

@DavidCarlisle @JosephWright not sure if it is relevant: but there is a patch for framebox in xcolor, and there is an old open bug report: latex-project.org/cgi-bin/ltxbugs2html?pr=latex%2F3655

@DavidCarlisle > Report: New Switch model with better graphics coming in 2021

@UlrikeFischer Oh, I did that :) The version I have draws the frame after the content is placed

@DavidCarlisle Ah, so I need to adjust: cool

@DavidCarlisle That's OK, that's what I do, it's about how you align the 'content' with the 'rules' - I'll read the implementations again and see if I can track down the differences

\setbox\@tempboxa\hb@xt@\@tempdima
     {\kern\fboxsep\csname bm@#2\endcsname\kern\fboxsep}%

@DavidCarlisle ^^^ Point there is that this sets the content in kerns of length 2 * <sep>, whereas I went for 2 * (<sep> + <rule>), which felt more natural

@JosephWright yes I will dig out a 2.09 version later see what it did, for framebox the 2e extensions should just have been \height and friends, and color safety but it's possible we dropped something there....

@DavidCarlisle In the vertical direction, I agree with 2e (we both include the rule width)

@JosephWright but you can't specify the vertical extent in the same way so that's a bit different anyway (more like \fbox)

@DavidCarlisle Sure

@DavidCarlisle Looks the same in 2.09 (just downloaded from CTAN)

@JosephWright good so any quirks in the design are Leslie not me:-)

@DavidCarlisle :)

@DavidCarlisle It's interesting seeing how the comments go all the way back to 2.09

@DavidCarlisle not your fault?

@JosephWright big update at learnlatex:-)

@DavidCarlisle Any idea what \@pboxsw is for? It's set in the kernel, but not actually tested

@DavidCarlisle Ah, forget that: I see, it's about \vcenter

@DavidCarlisle @JosephWright hm, should dvilualatex loads the dvips drivers? Probably nothing else makes sense?

@UlrikeFischer Yes, it should: isnt that already the case?

@UlrikeFischer You remind me I was planning to get a review done on my split-the-backend ideas

@JosephWright no it does. I was only wondering if it the right choice. But as it is more or less a dead end now it probably doesn't matter.

@UlrikeFischer Don't people need it for tex4ht and the like?

@UlrikeFischer dvisvgm from LuaTeX ...

@JosephWright yes, but they probably don't care about most specials.

True

@UlrikeFischer or get that extended dvips into texlive (if it works?)

@DavidCarlisle Good plan

@DavidCarlisle Spotted another 'easy win' to avoid math mode: \underline (re-implemented as a one-rule \framebox, basically)

thanks guys, the pdfs are definitely better than the pngs

I guess I should always use pdfs for research

@nbro In general, if you can go for vector graphics, go for vector graphics...

@egreg -- Well, AMS-TeX was contracted, i.e., paid for, and so was the conversion of AMS-TeX to amsmath. Both took substantial time and effort, and developing LAMS-TeX was about the same order of magnitude. Remember that TeX development was supported by grants; only after its release to the world was it declared free to use. And the *TeXbook" is still not free; see the text at the top of texbook.tex.

@DavidCarlisle I was looking at the dimensions of the pagecolor stuff, currently the rule uses \pdfpagewidth, and when I did set this to \paperwidth in dvips.def dvilualatex errored ;-(.

@UlrikeFischer where have you defined the rule? in dvips.def or in graphics? can't it use \paperwidth directly?

@DavidCarlisle nowhere yet. I was wondering which length to use at best. I mean can I rely on \paperwidth being equal to \pdfpagewidth e.g. with pdflatex? I often don't change \paperwidth but only the pdf commands if I want to change the mediabox.

@UlrikeFischer but otherwise pdftex.def sets the primitive to \paperwidth or \stockwidth (for memoir) question is for memoir if the stock is bigger than the logical paper size and you use \pagecolor, which size gets coloured..

@UlrikeFischer but if dvips.def is being used for dvilualtex it should probably have that whole setting block from luatex.def in addition to its traditional \special{papersize=\the\paperwidth,\the\paperheight}%

@DavidCarlisle well I think one should try to color the "maximum", so the MediaBox, page color should color the page. If people wants some smaller area, they can add custom background pictures (including ducks). So I'm trying to decide how to get this size at best.

I almost had to ask a question ... TeXbook Appendix G to the rescue

@JosephWright G and D the best bits:-)

@raf -- This is the first time I've looked at MathB.in. Instructions are generated with, I presume, something like Markdown. It doubles backslashes, which in a TeX context is lethal. I've encountered this before, and it was the cause of many wasted hours. Unfortunately, I don't know any fix to that, or any alternative websites. (I'll try to post a bug report, but may not be successful.)

@barbarabeeton doubles? I just tried $\frac{1}{2}$ and it made a half?

@DavidCarlisle -- I didn't say that code you might enter is wrong. What I mean is that, if you look at the tutorial, the instructions shown as "text" rather than "input" show double backslashes. Hardly good fodder for newbies.

@cis why would you need 3D if data is clearly 2D?

@barbarabeeton it didn't occur to me to look at the instructions!

@DavidCarlisle -- Well, since much of my existence has been engaged i writing documentation, my experience differs from yours.

@PauloCereda Legitimate question. Regardless of the fact that this representation is foolish: Because it looks beautiful (and because a user asked for this representation)

@cis suit yourself, it looks confusing to me. :)

@DavidCarlisle :)

arara is great!

@AlanMunn oh no

@AlanMunn <3

@PauloCereda I've extended the yaml you wrote for me to make 4 different documents since the Zoom versions of the course require individual links.

@AlanMunn woo that's great!

@PauloCereda And the ability to do shell commands as well is really useful so I can copy the final versions automatically to the TAs' folder.

@AlanMunn very happy to hear! :)

@AlanMunn busy semester, I believe :)

@PauloCereda Yes, starting next week. There's a lot more management for putting everything online.

@PauloCereda And we have students in different time zones and other fun things.

@AlanMunn Oh my, that sounds challenging!

@yo' welcome back! Successful webinar. :)

> The British royal family has a strict rule against playing Monopoly because "it gets too vicious."

@PauloCereda If only they could make that apply in the real world too...

@PauloCereda Monarchs aren't what they used to be.

@PauloCereda they have a monopoly on real land ownership why bother with a silly board game

@JosephWright I guess one can blame zap@space but I mention it anyway:

\begin{filecontents}[overwrite]{test-pgf.sty}
\RequirePackage{pgfopts}
\pgfkeys{/testpgf/.cd,testkey/.store in=\myshowkey}
\ProcessPgfPackageOptions{/testpgf}
\endinput
\end{filecontents}
\documentclass{article}
%\usepackage[testkey ={blub,quack}]{test-pgf} %works
\usepackage[testkey = {blub,quack}]{test-pgf} %fails
\show\myshowkey

\begin{document}
abc
\end{document}

@UlrikeFischer I really do wonder if I should sneak the @rawoptionlist code back in

@DavidCarlisle :)

@UlrikeFischer Oh great

@JosephWright I reverted the checkin really as the unused option list code is tricky and can break things (and it's not clear what used means in a kv context) but if we literally just saved the lists then packages could revert their option list to the original version before processing options, so it would only affect that package

@DavidCarlisle Would be good I think

@DavidCarlisle Meanwhile, I'm pretty sure I've reimplement boxes :) A great victory for expl3 ;)

@JosephWright github.com/latex3/latex2e/commit/…

@DavidCarlisle Yeah, the risk was all the other stuff: there's no downside to saving the raw data

@JosephWright I'll mail team list

Arrgh, I have to remember PSTricks to revise an old document.

@DavidCarlisle imho we should do it, it would give packages which wants to handle keyval a fighting chance ... (my example was from a real question: someone complained that the schule package didn't work).

@UlrikeFischer I'll make a PR so people can review

@AlanMunn :)

@PauloCereda tell me that :(