Henri Menke

This looks like a very nice addition, thanks! However, we need tests for the old and the new behavior. It would also be nice if you could write commit messages...

> If this is considered a backwards compatibility break I will think about this and come back to you. > It doesn't look like I can do this retroactively. You...

The CI failure is a bug in l3backend.

Could you please write a little more succinct commit messages? Commit messages can have more than one line, so something like this would be more appropriate: ``` manual: add warning...

You can run `git rebase --signoff -i origin/master` and replace all `pick` by `reword` to rewrite the commit messages and automatically signoff.

Signed-off-by is not really important for small fixes. It's more important for new features. I don't want to get trolled again by someone who doesn't understand their own licensing terms.

Unfortunately, we can't do that. What we can do, however, is change the default value for `shape aspect` to unset and use it if the user explicitly sets a value.

The culprit lies in the fact that the definition of `\pgfmathifthenelse` is ```latex \pgfmathifthenelse: macro:#1#2#3->\pgfmathparse {#1}\let \pgfmath@argument@1 =\pgfmathresult \pgfmathparse {#2}\let \pgfmath@argument@2 =\pgfmathresult \pgfmathparse {#3}\let \pgfmath@argument@3 =\pgfmathresult \pgfmathifthenelse@ {\pgfmath@argument@1 }{\pgfmath@argument@2 }{\pgfmath@argument@3...

That looks very tricky, however, note that the two step `first` + `second` is not equivalent to the single step `first`. You can easily check this by adding another circle...

The manual bounding box worked for me (but you have to take the linewidth of drawing the circle into account) ```latex \documentclass[dvisvgm]{article} \pagestyle{empty} \usepackage{tikz} \usetikzlibrary{animations} \begin{document} \begin{tikzpicture}[very thick, animate =...