So I'm working on something I think will be cool, but I've run up against a road block and was hoping someone here might be able to point me in the right direction before I invoke higher powers (e.g. emacs-devel).

I'm trying to interface with a library that needs to be notified about changes in the document. This is what after-change-functions is for. Such a function receives the endpoints of the changed region (after the change), and the length of the region before the change, in terms of characters. This is good, but I need it in terms of bytes.

The endpoints can be converted to bytes with position-bytes, but I can't convert the length before the change happened because characters may have been deleted, and I need to know which characters those were to compute their length in bytes.

So the only workaround I've figured out so far is this:

• In before-change-functions, record the beginning of the region and the actual contents of the prior region as a string.
• In after-change-functions, use the stored information to compute the actual number of bytes in the region that was changed.

This seems to work but it looks brittle. One call to before-change-functions may be followed by several calls to after-change-functions, so the latter needs to update the recorded region info every time it runs. This is rife with opportunities for off-by-one errors, and maybe expensive: we're looking at something like two calls to buffer-substring-no-properties, three calls to substring and one call to concat per change. That feels like overkill.

Also, the manual warns against using both before-change-functions and after-change-functions in tandem, presumably for this very reason.

Are there better ways to do this?

My current code, for reference: https://gist.github.com/TheBB/3cb185b75d826a67c31d09983e12eedc