Jonathan Hayase
Jonathan Hayase
Or maybe something as simple as `(memq evil-next-state '(motion normal))`.
@nonsequitur when this is merged, it's going to be awesome!
Out of curiosity, what does `smex` do that `counsel-M-x` doesn't?
I see two ways of doing this: 1. Cache the result in a text property attached to the candidate. 2. Find a root helm function and `let` bind the cache...
When Emacs 25 is released, I'll revisit this and probably fix this along with #1 and #5.
Here: ``` (defun try-expand-flx (old) "Try to complete word using flx matching." (if (not old) (progn (he-init-string (he-lisp-symbol-beg) (point)) (if (not (he-string-member he-search-string he-tried-table)) (setq he-tried-table (cons he-search-string he-tried-table))) (setq...
I definitely feel like a recency weight would be helpful. Recency could be thrown into `helm-adaptive`, `company-statistics`, minibuffer histories, etc. Right, now I'm thinking of `constant/log(n)` as the weight, but...
I'm starting to lean on summing `constant/sqrt(n)` for each candidate. I think it yields a good balance between recency and favoring often-used candidates, even if they're less recent.
Sorry if this wasn't clear, but I believe `flx` already implements a O(nm) matching algorithm. The implementation is a bit non-traditional, but the asymptotic complexity should be fine (as far...
> The problem with \w is that in most implementation it is not unicode friendly. > You'll split on accented letters like é, û, à, and probably any non latin...