[slime-devel] ruby Was: Daily ChangeLog diff

[Andras Simon]

On Thu, 19 Feb 2009, Helmut Eller wrote:

> * Andras Simon [2009-02-19 14:01+0100] writes:
>
> > I'm not sure where (or how far) you want to take it, but this looks like very
> > good news to me! Is it ready for alpha-testing?
>
> If you want to use it, you have to fill in many missing parts.

I wish I could...

> This was primarily a exercise to learn a bit of Ruby.  The file
> implements only very little functionality.  The only command that works
> is slime-eval-region.  The debugger isn't a real debugger it just
> displays the backtrace (without local variables) from the exception
> object.
>
> Certain commands, like slime-eval-last-expression, would need some elisp
> code which understands Ruby's syntax a bit better.  Or perhaps it would
> be better to have different commands like eval-current-line because "the
> last expression" is not a very useful concept in a infix syntax.

Syntax-related stuff could perhaps be lifted from ruby-mode. Although
I'm not sure how good that is; ruby-forward-sexp for example doesn't
quite do what I think it should. (But it's possible that the problem
lies with my expectations.)

>
> > I.e., are you ready for
> > taking questions and bug reports from clueless users?
>
> I can answer questions, but I don't have a good idea how a decent
> Slime-for-Ruby should look like.  E.g. tab-completion for method-names
> seems to be hard(er) to implement in Ruby because most names are
> interpreted relative to some module/class while in Lisp most
> function-names are global symbols.

I wish I knew how to go about it. Maybe a combination of slime and jde (very
fancy java mode for emacs) would be needed. (It's been a while, but I think it
could do completion.) BTW it seems that ruby-mode can already do some
symbol-completion based on ri:

http://groups.google.com/group/comp.lang.ruby/browse_thread/thread/50a36818c7f9f41/bda34f18a5cd0e04?lnk=gst&q=emacs+ruby-mode#bda34f18a5cd0e04

Of course it's not the real thing.

> My impression was that Emacs's ruby-mode works quite well with irb in a
> comint buffer and hence, something like Slime is less necessary.  I was
> also pleasantly surprised that Matz seems to be one of the authors of
> ruby-mode.

I'm afraid I'm spoiled by slime. ruby-mode is much better than nothing, but
much less capable than ilisp was. But it's possible that I'm missing
something.

Thanks!
Andras