Word/symbol expansion

[mclearc]

This may be due to my ignorance of the finer details of meow, but I can't seem to do the following:

take a word/symbol pair like @kitcher2018, which is pretty standard for of citation that I use. If I place the point on @ and call meow-next-word I'll get kitcher2018 highlighted. Now let's suppose that I also want the @ highlighted. How can I easily do that? Calling either meow-mark-word or meow-mark-symbol won't work. And meow-block doesn't do what I would expect here either, since it just jumps the point to a bracketed expression rather than expanding the block to the smallest next larger increment (in this case @kitcher2018). And meow-bounds/inner-of-thing doesn't help either. Any suggestions on what to do here? Could meow-block be modified?

[velppa]

For such cases I use meow-mark-word meow-reverse meow-left-expand, which are w;H
for me. Not sure whether it's the most effective approach.

Related, given the text with | denoting cursor:

(format |"%s %d" font size)

meow-next-word would select %s, but if it's a single-quote instead of
double-quote, it would select it as well '%s. Is it expected?

(format |'%s %d' font size)

[DogLooksGood]

Emmm, in elisp, strings can't be quoted by single quote.
But it should still select %s, because word selection only include characters those recognized as -_w in syntax table(whitespaces/symbol constituent/word constituent).
A simple skip-syntax-forward/backward is used here.

In your case, '%s is selected, probably because ' is added as one of -_w in syntax table. By default, it belongs to prefix, you can check the syntax table with C-h s.

[velppa]

Thanks for sharing, didn't know about syntax tables. Surely, in Elisp strings aren't single quoted, the example was purely illustrative as I bumped into it in some other mode.

[eshrh]

Sorry, maybe I don't understand...

If you have (| is the cursor) @|kitcher2018 and call meow-next word (e), you get kitcher2018| highlighted (identical behavior to meow-mark-word here). Then calling meow-mark-symbol (W) seems to make @kitcher2018| highlighted on my computer. Is there something wrong about this?

[DogLooksGood]

@mclearc You are right, there are some details not mentioned by the document. Syntax table controls what characters are included in word selection. It is also the way builtin forward-word/backward-word works. Meow respects the builtins whenever it's possible.

In your case, modifying syntax table is the best solution. Syntax table is the perfect solution for people who want to tweak the underlying behavior. You will like it once you get used to it.

[mclearc]

I think it is totally reasonable if you want the user to make their own modifications here, especially since meow doesn't intend to make any changes to the underlying emacs functioning. But it might be useful to flag this clearly in the README info, since this won't be at all obvious to people switching from other modal bindings, especially from evil.

[DogLooksGood]

Yes, I agree. But it's not easy. When I switched from evil to vanilla, I felt like I have to relearn everything. Since evil made a context which do not need its users to know much about Emacs itself. And Emacs itself is complex too, by said Emacs is customizable, as long as you have new perceptions, you can have new customizations. It will be too overcrowding to list all potentially useful information in Meow's documentation, It's also hard for me to figure out what is worth to be listing. If we want a good documentation for evil users, that must be a migration guide, from the perspective of evil. Of course, it will take a lot of time. At the moment, I am not ready for that, there is still room for other improvements. Currently, I'm focusing on making Meow a good modal editing package at least for people who is familiar with vanilla Emacs.

[eshrh]

@DogLooksGood

I think it's still valuable to offer an option to use the same syntax table across modes. Here's what I'm thinking:

Offer customization variable, meow-syntax-table, initially set to nil. Every call to bounds-of-thing-at-point and other thingatpt.el functions is wrapped with (with-syntax-table). If the variable is nil, then it uses the syntax table of the current mode, default behavior. Otherwise, it uses the given syntax table.

[DogLooksGood]

@eshrh It doesn't feel right to me to introduce another variable. modify-syntax-entry can either set globally or set for a specific syntax table of a major-mode. I think it provides everything needed for a same syntax table cross modes. And of course it works with all commands. Our meow-syntax-table can only be used in meow, will not work with builtin commands.

Why there's no such single global syntax table? Just because it's not a good idea.
For examples,

• a-b-c in Lisp / C
• ~a in Elisp / Clojure
• 'a' in Lisp / Python

There are just so many differences.