Don't highlight symbols that appear in strings about symbol-overlay HOT 4 CLOSED

The expression (nth 3 (syntax-ppss)) returns non-nil when point is inside a string, so this would be easy enough to do. I tend to agree when viewing this simplistically, but there's a compelling counterpoint: inside strings in Ruby and other languages, you can interpolate code, e.g. "hello #{name}". And when point is inside the interpolated code, e.g. on name, then (nth 3 (syntax-ppss)) still returns non-nil. It's pretty clear to me that a user would still want name to be highlighted here.

from symbol-overlay.

@purcell thanks for the quick response! You're totally right about languages like Python and Ruby with that style of format string. What about a toggle (with the default being the current behavior)? One could then enable or disable it in a major mode hook or similar.

from symbol-overlay.

Please, if you decide to address it, either leave it opt-in, or make such behavior very explicit in readme.md. People expect this mode would highlight anything that is under the caret, and if it doesn't, this looks like a bug. In fact, there's an actual bug where the mode doesn't highlight words sometimes.

from symbol-overlay.

My preference overall is to not add any new logic relating to this. It would be hard to implement in a sensible way. e.g. what happens if "highlighting in strings" is disabled, but a user actively tries to highlight a specific word in a string. Should that be a no-op?

In theory it looks like you could already easily get this effect on a mode-by-mode basis by adding your own function into symbol-overlay-ignore-functions: see here.

from symbol-overlay.