I updated to the 5.0 development branch just now, and I found that
the following link in my tree broke:

> [Zorn’s lemma](https://en.wikipedia.org/wiki/Zorn\%27s_lemma)

Removing the escaped \% fixes it. The error message is

    ^ Expected text but got: (Eval.V.Content
                                (Types.Content
                                   [(Types.Text
                                       "https://en.wikipedia.org/wiki/Zorn");
                                     (Types.TeX_cs %);
                                     (Types.Text "27s_lemma")]))

> Is it documented somewhere? If not, can it be added?

This is documented at [1], and you can use arbitrary HTML tags according to [2].


[1] https://www.jonmsterling.com/jms-00D0.xml
[2] https://www.jonmsterling.com/jms-00RH.xml

Sure, but perhaps bikeshedding a little more on the actual syntax.

> The scope of such a function is a bit unclear... How bad would it be if we *at least* just rendered the KaTeX code as text (but without attempting to transform the code into plain text mathematics)?

This would be OK if we're taking the *forester* code, not KaTeX code, because
otherwise the title will be overwhelmed with complicated macros. (If the macros
were not complicated, I wouldn't define a shorthand for it in the first place.)
But would this be a bit hacky since it doesn't respect beta-equality anymore?

> Ultimately I would like to have a version of the command palette that can have KaTeX nodes inside it!

This would be very helpful, but it is probably not realistic to expect that
everything using the title will support this. Wikipedia for example resorted
to not use any special symbols due to "technical limitations". A forester
directive like \alttext{...forester code...}{plain text} is perhaps the most
universal solution.

I tried it out and nothing broke. Although I did wrap my macros in
#{...}, because the docs say it is idempotent, and this is also the
canonical way to do it in Typst (where we define functions, not macros,
so we must be in math mode to actually scope check the math symbols).
So after this change I need to not do that because this is not idempotent
anymore: x {+} y is not the same as x + y. So I'm fully in favor of this
change, as long as the docs are updated accordingly.

I'll try out the commit, but incidentally I'm stopped by OCaml,
roughly because of a tangle of problems that I finally diagnosed to be
 https://discuss.ocaml.org/t/unable-to-install-some-packages-with-opam-on-macos/14434/6
which may be a datapoint regarding the discussion on build ecosystems.

When writing trees related to delta functors, forester drops the
formula in titles shown in `query all`, and the Ctrl-K pop up.
I didn't remember exactly but I think this is not the previous
behavior. What would be a nice way to make it show at least
something? A tree titled "-functor" is really unhelpful here.

Perhaps we can set up a small ad-hoc procedure to approximate the
formula, covering KaTeX built-ins such as greek letters and \hom,
\lim? This has surely been done before. More generally, is there a
way to tell forester how to render plain text from contents?

> I would like to keep it simple *and* language/locale-independent. For that reason, I would probably not like to include a pluralisation filter.

I think Wikipedia simply looks at characters directly following the link, not
specifically looking for plural endings. Of course, this is still language
dependent since some languages don't want spaces after links. Maybe a
more explicit syntax for adding pre/postfixes into the link is desirable?
For example, [[link|s]] might be a solution (I don't know if | is allowed in
links or not, but we can always bike-shed later).

I suppose a few features documented in (for example)
  https://en.wikipedia.org/wiki/Help:Link
can be useful. For instance, [[link]]s will tuck the -s in the link,
very useful if you just want to add a simple plural. A way to
control the initial case could also be useful, but perhaps it is too
complicated.

Wow, thanks for the quick response!