Image leavevmode

[Alex-Jordan]

Two pictures are worth 2000 words. So, before in LaTeX:

And after:

This PR applies \leavevmode before atomic images that you would want to be in line with a number off to the left (from a list, exercise project). Then it backs up by an appropriate amount of space so the image is "top aligned" with the number. The "backing up" is either by 1.5\baselineskip or 0.5\baselineskip depending on the situation. These values are explained in the comments.

[rbeezer]

Two pictures are worth 2000 words.

Hah!

I can't test right now, so I'll ask. What is the behavior for HTML?

[Alex-Jordan]

Current HTML behavior could be better. But I think some minor CSS changes could handle these.

I didn't take a screenshot, but the new task and exercisegroup behavior is like the list behavior from the earlier screenshots. Sepaarated thiis ini two commits so you can see the new examples before and after the real commit here.

[rbeezer]

Thanks (I'm at 35000 feet right now.) Will study more carefully once at a desk. A bit hard to judge with only the limited context in the diff. So take as first reactions:

• Almost certainly, the removal for stage should go on a new commit.
• What about a new argument to the image environment, which if not blank, does the leavevmode and uses the argument as the backup. I want to avoid writing fiddly bits into the body repeatedly.
• Once this PR is in good shape, then let's confirm the CSS is agreeable on the HTML side.

[Alex-Jordan]

I can do first bullet and look into second.

Not sure I understand 3rd. Maybe it is not clear but status quo is that both HTML and LaTeX have some undesirable behavior, but they are different. LaTeX puts these images too high. HTML/CSS puts them too low. So it's not a situation where "at least they are in sync". Seems like we can make one better and independently make the other better and they will be in sync then.

[Alex-Jordan]

Frustratingly, writing \leavevmode\nopagebreak\vskip{-1.5\baselineskip} before \begin{image} does not produce exactly the same results as putting begin=\leavevmode\nopagebreak\vskip{-1.5\baselineskip} as an argument to the tcolorbox for an image environment.

So I have your idea for an additional argument to the image environment "working" but the spacing is not quite as good. And it is not consistent across all the types (list items, exercises, tasks, exercisegroup exercises), and doesn't seem to match my understanding of where -1.5\baselineskip and -0.5\baselineskip should be the exact values to use. Here is one pic to compare with the second pic from my original post:

Maybe hard to see but the pics are now a bit too high. And I don't have screenshots to show but on the exercisegroup exercises they now are a bit too low. I've even been playing with combinations of placing %
 here and there too, and this is the best one.

If it has to be this way, it seems there is some internal thing about how tcolorbox begin=... works that I don't understand.

[rbeezer]

Frustratingly

Welcome to my world.

Have you seen the tcolorbox key: /tcb/box align? I see it described on p. 91 of the current manual. Maybe the base alignment would work here. Perhaps the \leavevmode might need to be independewnt of that? Or in before or ...

Bullet 3: I just meant to get this pretty much finalized before we get the HTML stuff wound up.

[Alex-Jordan]

I resorted to using a \NewDocumentEnvironment around the existing \NewTColorBox for images. In the former, I can include the leavevmode etc. without the TCB interference I described in my last post.

[Alex-Jordan]

Didn't see your last note. I'll look into box align.

[rbeezer]

And to further add to the mix, from p. 16 tcolorbox documentation:

Basically, \NewTColorBox operates like \NewDocumentEnvironment.

Just use a case-sensitive search. ;-)

[Alex-Jordan]

Basically, \NewTColorBox operates like \NewDocumentEnvironment.

I saw that and the current version of this PR is based on that. If you have nothing better to do (heh) take a look aat the current version.

\NewTColorBox does not offer the before-code and after-code argments that \NewDocumentEnvironment offers. Instead, it assumes all you want to do is begin and end a tcolorbox. And it gives you a place to pass options to that tcolorbox.

[Alex-Jordan]

I tried using box align=top, which puts the top of the image on the baseline of the line. So far so good.

The next step would be to "back it up" by one baseline skip. Everything I have tried with tcb options (before, before skip, and other things either does not work, or does not work as hoped for. For example, before skip=\notblank{#4}{-1\baselineskip}{} somehow affects the positioning of the list markers too, as you can see here:

I have to give up on this. Let me know if you think the NewDocumentEnvironment approach is not too clunky or if you otherwise foresee issues with it.

[rbeezer]

\NewTColorBox does not offer the before-code and after-code argments that \NewDocumentEnvironment offers.

I sorta expected that would be the case.

I have to give up on this.

You aren't trying hard enough. ;-) ;-) ;-)

An extra depth of nesting structures in the preamble won't bother me. And the body is still just as semantic (nearly!) as before.

Latest commit looks good visually. Might do a round of merging (PRs, little stuff while traveling) later this afternnon. Since I have "have nothing better to do". ;-)

Thanks for working through this one.

[rbeezer]

Mysteries I'll never understand. Compare current PDF of sample article at website (or build yourself from current master), and the PDF generated with this PR.

• Many pages are in-sync, until Figure 25.28 moves to a new page.
• Right before Figure 28.2, the paragraph word-wraps differently.

Not a show-stopper and I think I'll merge this very soon.

[Alex-Jordan]

I have many versions of the PDF built:

• tip of this branch. Fig 25.28 is at top of page 150.
• one commit prior (with the sample article additions but without the new placement for those images). Fig 25.28 still at top of page 150.
• yet another commit prior (no additions to sample article). Fig 25.28 still at top of page 150.
• yet another commit prior (9a534cb, 5 commits behind current origin/master). So now also without the cutting of the stage template. Fig 25.28 still at top of page 150.

In other words among all those, I get no changes in pagination.

Among the 5 commits are your LaTeX copyright commits. Are they in play for explaining that inconsistency? If you put this PR on top of current master, do you get the same inconsistency?

[rbeezer]

Well, I'd not added this PR onto current master. But no difference when I do.

copyright is only for book. ;-)

Not a problem, and I bet we have subtle differences in packages somewhere. Just a curiosity.

Thanks for all the checks.

[Alex-Jordan]

I just built the PDF from e938a31 (current master without this PR) and I still get that figure on page 150. But yeah, the public one has it on 149. So I guess it is some weird package difference.

Should we be concerned as developers that we don't get the same output? Could make future confusion if you and I are off by half a page in the sample article PDF.

[oscarlevin]

Regarding this oddity, this is an argument for requiring developers to use a fixed developer environment with common versions of latex and python etc. Codespaces can do that, which can also be set up locally as a docker image </bearpoking>

[rbeezer]

Merged, and now rebuilding website.

I really like how this ended up - the code and the feature.

Usual editorial work on the history. Sample images were great. Now that they have served their purpose, I moved them to follow the feature implementation.

Thanks!

[rbeezer]

<bearpoking> No kidding! ;-)

LaTeX is a strange beast. And sometimes counter-intuitive. "Not a problem" - and I'll go furhter, "I won't lose any sleep over this one."

[davidfarmer]

Did I miss the need for some corresponding CSS change, or is the HTML okay?