just published parinfer 1.4.0 to npm

2x speed improvements

~20ms to process 2800 lines

mode functions now return changedLines if you don’t want to overwrite the whole buffer everytime

and they also return an error object if you want to report the errors in the editor somewhere

code-wise, I did a lot of rewriting so it’s easier to compartmentalize what it’s doing, and it’s about 200 lines shorter

and I’m finally working on documenting the algorithm here: https://github.com/shaunlebron/parinfer/blob/master/lib/parinfer.js.md

apologies to the plugin authors, I expect it to stabilize after I anneal it a bit while writing docs on it

the api is of course stable, but for anyone trying to port the code, I recommend waiting on that until I finish the docs

@shaunlebron: Thanks for the update, I’ll wait a bit then

Is it possible to partially process files rather than having to do the whole thing?

i.e. from a particular change, can you work out how much of the file you need to process? I guess that would be different for indent and paren modes.

Indent should only require updating the top-level forms containing the changed lines I think, but paren mode changes might extend further.

Marijn Haverberke of CodeMirror fame came up with the concept of "interruptable, resumable parser”, maybe something which could be applicable? read “A parser” section, http://codemirror.net/1/story.html#parser

http://codemirror.net/doc/internals.html