Fixes syntax highlighting

[appascoe]

This resolves #67 .

It appears that using thing-at-point, when looking at the returned properties of the thing would include that the thing was already fontified. This effectively would turn off any syntax highlighting, including the naive highlighting before sclang-class-list is loaded. Switching to another function, current-word, appears to have the correct behavior: when typing out a class, it is not highlighted until the word matches a class in sclang-class-list. For example, starting to type SinOs results in plain text, but upon finishing with SinOsc, it becomes highlighted.

[smoge]

thing-at-point → current-word swap is probably fine, but I still have questions about what the problem was

Good thing to test: reproduce or backtrace to show where thing-at-point returned nil/wrong value

Also, you mixed space and tabs, and the code isn't alined. Remember to check that. The original code uses spaces.

[appascoe]

Ah, sorry, just relied on the automatic tabification. Don't know what was up with my setup as I had to manually go re-align it and untabify; should be fixed now.

I did go through and trace what thing-at-point returned. It returns a whole substring with a set of properties (thus, the original code calls (substring-no-properties thing) to strip those properties out to then search the sclang-class-list). I wouldn't classify this as returning the wrong value, and it definitely wasn't nil. However, one of the properties indicated that the substring was already fontified, which I suspect was preventing the final, correct fontification from happening.

I last did any real lisp stuff 15 years ago and I think to figure out what is really going on would be a deep-dive into the underlying emacs syntax mechanism. I think I've gone about as far as I intend to with limited expertise in the language, the lack of unit tests, etc.

[smoge]

I last did any real lisp stuff 15 years ago and I think to figure out what is really going on would be a deep-dive into the underlying emacs syntax mechanism. I think I've gone about as far as I intend to with limited expertise in the language, the lack of unit tests, etc.

As I said, this is probably fine. If it works, it works. All I'm saying is that it would help to know the underlying problem. :-)