Re: Emacs lily mode

[Nicolas Sceaux]

Le 15 juil. 2010 à 08:28, David Kastrup a écrit :

> Well, it does not look like this is going to be a fast thing.
> 
> a) Nicolas has decided against using CEDET/Semantic for the parsing of
>   Lilypond because of performance reasons.  That is a no-go for my
>   tastes because of being Emacs fanboy: if Semantic is not good enough,
>   it needs to be improved.  Fortunately, its performance problems are
>   claimed to be mostly fixed.
> b) I was banking on CEDET being an integral part of the Emacs
>   development version.  That is not the case with regard to the
>   development tools of CEDET allowing to _write_ new mode support
>   rather than use existing ones (like c-mode support).
> c) Nicolas' Scheme/Lisp/Elisp coding style is a world of its own heavily
>   depending on whatever is available to make for some object oriented
>   Common Lisp programming style, to a degree where the code is utterly
>   unfathomable to people not familiar with the libraries providing the
>   respective syntax macros.  Since those facilities are not "built-in"
>   but heavily rely on macros and support functions, a lot of which are
>   _not_ to be loaded at runtime for "standard" Emacs modes since they
>   change Emacs' operation, one has to touch a lot of areas in order to
>   convert the code into something that can at least _run_ without
>   loading cl and stuff.
> 
> So this does not look like leading to a common project anytime soon.  As
> a start, I am working on b), namely pestering CEDET developers to try
> getting a version of CEDET useful (and documented) for writing new mode
> support into Emacs.
> 
> a) is not going to fly for me: too much code to maintain separately from
> Emacs main.

David,

All your points are valid.  Being more used to Common Lisp, my coding style
tends to be much influenced by it.  Moreover, I took inspiration on a Common
Lisp library, heavily relying on generic functions, multiple dispatch, etc.
So indeed my elisp code is far from being idiomatic.  Besides, I am not too
happy about having reinvented the LR parsing stuff, which is mainly broken
anyway.  That's mainly why I have not advertised this emacs mode.  It works
for me, but I don't think it's good enough to make a publishable software.

I am keen on seeing your progress.  Having a useful lilypond mode in Emacs
would be really great.

Nicolas