[ruby-core:20223] Re: Proposal: deferred blocks

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Well,

once you want as string representation of a lambda, you are already =20
close to getting the rest. Hitting language constructs is a reason, =20
but maybe i should elaborate a bit on my opinion: they are language =20
constructs, because they have an unchangeable meaning. I see this as a =20=

cornerstone of Ruby (otherwise, i would use lisp ;) ). So if you want =20=

something with a similar meaning, Ruby (especially 1.9) gives you a =20
lot of rope to play with such tricks as I do. It doesn't go the whole =20=

way because it was never the real intention of the language. I also =20
voiced the advantage of my approach: it works outside MRI 1.8.x. =20
(Actually: it works improves on 1.9, especially instance_exec gives =20
you a lot)
All share a common problem: following whats happening is truly =20
beginning to get hard for non-zen programmers.

But to get down to the detail: i don't really see a true necessity. =20
Sure, its fancy, but fancy doesn't get my work done better. All you =20
want to do is to side-step the interpreters interpretation of what =20
Ruby is and give it your own. Also, i don't see the advantage when it =20=

comes to error handling (fancier error messages?).

Also: Ruby 1.9.1 is in freeze and this is a feature I don't see worth =20=

thinking about before the next big release. Let's play a bit of the =20
users perspective: we should stop adding features, i want to start =20
using this.

Regards,
Florian Gilcher

On Dec 2, 2008, at 8:06 PM, Yehuda Katz wrote:

> My example was only a real-use that already exists.
>
> You articulated the problems with your approach (namely, it falls =20
> apart when you hit language constructs). People have been playing =20
> with this for a while, but a big blocker is the fact that it's known =20=

> not to work outside of 1.8.x MRI.
>
> As I clarified earlier, I'm not asking for ParseTree functionality =20
> in Ruby. I'm asking for procs to optionally remember their string =20
> form so that they can be processed by other utilities (like =20
> RubyParser). It's not as intensive as requiring that Ruby have a =20
> standard AST across implementations, and might potentially have =20
> other benefits (error handling, for instance).
>
> -- Yehuda

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.8 (Darwin)

iEYEARECAAYFAkk1kHwACgkQyLKU2FMxSOKQAQCfT8AVzQXyn0Wt1QB3zU/BWxtV
e7QAn1/GW6EhIYlAcnOsRvE8yMnBEwgi
=3DVLk2
-----END PGP SIGNATURE-----