Building 'pretext build latex-novideo' creates extraneous '\end{minipage}'

[APEXCalculus]

I am building a .tex file on Codespaces ("ubiquitous-spoon") and use the command pretext build latex-novideo.

In the resulting apex.tex file, line 61256 is \end{minipage}, but there is no \begin{minipage} ever used in the document.

When that line is commented out (and all figures are generated/placed in the right folders), the file compiles without error.

The .tex code is coming from sec_vector_fields.ptx, roughly line 650.

What is causing that extraneous command to appear?

[sean-fitzpatrick]

Hmm, that is weird. Is your codespace up to date?

I think no: I updated the build targets to better reflect that the main edition the one without videos. (Also, I wanted file names to match the names on my server, so I don't have to always rename everything when I update the book!)

You can build the PDF directly with pretext build apex-pdf.

Where this is coming from: the equation at line 650 of the vector fields section used to be a single line.
But that line was wider than the text width. I had a hack in place to put that equation in a minipage and shift it into the margin, to prevent it from running off the page.

But I decided this was silly: I should just split the equation across two lines!

I'm not sure how you ended up with some intermediate stage where the minipage was ended but not created. That's weird.

Suggestion: clear the contents of output/, do a git pull to update main, and try doing pretext-build apex-pdf.

[sean-fitzpatrick]

Also: @oscarlevin has updated the codespace container to use a new version of Asymptote. But I have to submit a change to the pretext script to take advantage of this. I'm hoping we'll have the asymptote PDF builds fixed soon.

[APEXCalculus]

I saw that you had changed the build commands; I merged the update to main but recognized that it didn't seem to make a difference in the Codespace. Turns out I had to update that, too, which I've done now.

So I'm building apex-pdf now and will report back to see how things go.

[sean-fitzpatrick]

Ok. Make sure you change the asymptote method back to 'server' or you'll hit errors there.
I'm hoping that won't be necessary by the end of the week

[APEXCalculus]

Building gave me a LaTeX error:
! LaTeX Error: File xcoffins.sty' not found.`
It then doesn't create a pdf, nor does it provide (as best as I can tell) the .tex file.
I'll see if I can figure out how to update the LaTeX installation.

[sean-fitzpatrick]

Ok. That is probably another case of the Docker configuration not having every LaTeX package. The xcoffins package is what we use to put stuff in the margins.

[APEXCalculus]

Got it to compile; needed to install the l3experimental package.

Now prextext build apex-tex produces apex.tex that I can download. It does not produce a pdf; it seems there is an issue with the font. I'm fine with that at the moment, for the .tex code compiles on my machine without error (though I already have the .asy files saved - I haven't worked on getting them generated into the assets folder yet.

[sean-fitzpatrick]

There's a new container for the codespace that has xcoffins, but I need to update the container for that to work.

There isn't a target named apex-tex so I'm surprised that built anything.

I'm still not getting asymptote to build correctly in the codespace. Some images work, and some fail.

Oscar said he built asymptote with the new Vulkan libraries that are required, but running asy --version in the codespace says otherwise.
The result is that the images aren't rendered properly. They look like cartoon versions of the correct images.

[APEXCalculus]

Oops - typo. I meant pretext build apex-pdf, not apex-tex.

When I manually compiled asymptote, some failed because some characters got replaced by html code. In particular, > was replaced by >, and likewise for <. These symbols were sometimes used in if statements, causing Asymptote to fail; other times, they were in labels causing the subsequent LaTeX calls to fail. The boolean and, using &&, also caused issues.

I certainly didn't code the original .asy file that way. I wasn't sure if these symbols appeared because of something PreTeXty, or if was a result of my script that pulled the code from the .ptx file.

That certainly won't explain some rendering in a cartoonish way.

[sean-fitzpatrick]

The symbols <, >, & are the three reserved characters in XML. In order to keep the Asymptote code in the PreTeXt source files, they need to be escaped. But when the PreTeXt script does its thing, it rips out the asymptote code and puts it into an .asy file, with the necessary preamble. Any escaped characters should be converted back at this point. It then builds the image, and once all images are built, copies them all over to the generated assets folder.