wcheck-mode (was: Finding packages to enable by default)

[Teemu Likonen <tlikonen@iki.fi>]

[-- Attachment #1: Type: text/plain, Size: 1997 bytes --]

Stefan Monnier [2014-06-20 21:28:08 -04:00] wrote:

> I just had a look at wcheck-mode and it looks very interesting. Would
> you be interesting in contributing the code (we could at a minimum
> include it in GNU ELPA, tho I think it might be worthwhile in Emacs
> itself).

Either way is fine. The papers have been signed so I guess you can just
pull the code. Potentially Emacs developers can improve it. I'm pretty
much of type "I just scratch my own itches" and to me wcheck is about
ready. So I'm not interested in new heavy maintenance burden. But I'm
here and can help.

> I see it is "display and timer driven". The "display driven" part is
> interesting since it saves you from having to "touch" text before it
> gets syntax-checked. And the "timer driven" part makes it less
> intrusive performance-wise. Also it operates on larger chunks of text
> at a time, which is more efficient.

I remember trying a couple of different things years ago and they all
were based on the idea of reading the buffers' window content kind of
visually. Invisible text is skipped and text outside window areas is not
read nor painted.

> I have a question: why does it use window-scroll-functions and such
> hooks instead of using jit-lock?

Maybe just because I'm not familiar with jit-lock. Wcheck uses some of
its features, though. Anyway, the idea is to trigger wcheck read event
for after any change in window configuration, buffer content, outline
visibility etc. It's perfectly fine to use other methods if they offer
similar trigger. After a quick look, maybe through jit-lock-register?

> One downside for the user is that ispell.el's M-TAB might not work as
> well since by the time wcheck's timer runs to discover your typo, the
> user might "too far".

Wcheck's "action" feature uses synchronous process to get information
From its back-end program. Maybe wcheck's version of M-TAB feature can
be implemented similar way. I've never used M-TAB, though.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 835 bytes --]