Toggle further math environments

[clason]

(This is a follow-up to #1858.)

There are further pairs of math environments that would be useful to be able to toggle easily between them. (Some of them are already covered by cs$, but a toggle would be more convenient.)

• $$ ... $$ -> \[ ... \] (one-way, since since $$ should not be used in LaTeX! note that ts$ currently errors in this situation)
• \[ ... \] <-> \begin{equation*} ... \end{equation*} (here the direction -> is already possible via cs$, but <- is not).

It's not clear to me whether this would be better included as a three-way toggle, or a second toggle (tsm, for "toggle surrounding math"?).

[lervag]

I believe I have fixed the issue with cs$ - it should now work with math envs. The list of environments is hard coded and may need to be expanded. And it will currently not work with nested environments. This may be improved upon, of course.

[lervag]

I believe the current code should be easy to understand:

vimtex/autoload/vimtex/env.vim

Lines 299 to 311 in 7afb9de

	function! vimtex#env#toggle_math() abort " {{{1
	let [l:open, l:close] = vimtex#env#get_surrounding('math')
	if empty(l:open) | return | endif
	let l:map = {
	\ '$': '\[',
	\ '\[': '$',
	\ '$$': '\[',
	\ '\(': '$',
	\}
	call vimtex#env#change(l:open, l:close, get(l:map, l:open.match, '$'))
	endfunction

Currently, the only toggle loop is between $ and \[. Do we want to make this more advanced, or is this the most "expected" behaviour? Remember, cs$ should now work as expected and allows to make an adjusted version here.

I'm thinking that it could be interesting to have the loop be three-way, i.e.:

$...$  ->  \[ ... \]  ->  \begin{equation} ... \end{equation}  ->  $ ... $

[clason]

I think that's a very good compromise. (Although the \( -> $ transformation is a bit opinionated; the "official upgrade" would be the other way around ;))

The three-way loop is also attractive (and I now agree it should be without *), although I understand if people found it annoying to skip the unwanted alternative. If it allowed code simplification, however, I'd be all for it!

[lervag]

Ok. How do you think the updated behaviour is? Does it resolve your issues?

Although the \( -> $ transformation is a bit opinionated; the "official upgrade" would be the other way around

Good point. I personally have never become used to the \(...\) syntax and I more or less never see it anywhere. But you are probably right that it is officially considered an upgrade. Still, let's be pragmatic. If anyone should complain I'll be glad to reconsider!

[clason]

I think it's quite nice, yes!

(And personally I agree on the syntax; it's just too painful to type. I just mentioned this for the sake of being pedantic :))

[vneiger]

The changes look good on my side as well. Thanks for the work on both issues!!

Just an off-topic comment: I don't think the syntax \(...\) is more painful than another. I discovered snippets thanks to the vimtex documentation, and for example using Ultisnips with something like

snippet $ "math \( ... \)" i
\(${1:${VISUAL}}\)$0
endsnippet

makes it as convenient to use \(...\) as the more classical $...$.

[lervag]

Yes, that's a good point, @vneiger. I believe my "problem" with \(...\) is that I don't need it. That is, there's nothing wrong with $...$ for me - it works as expected and is easy to type.

[xsrvmy]

On the topic of \(\) and $...$, the first syntax is actually better in some ways because it makes syntax highlighting and conceal changes more contained if you type a backslash at the end of math mode.
Is it possible to expose that list to configuration by any chance?

[lervag]

Is it possible to expose that list to configuration by any chance?

Sure. I've pushed an update that exposes it through :help g:vimtex_env_toggle_math_map.