On Mon, Nov 28, 2011 at 2:29 AM, Yukihiro Matsumoto <matz / ruby-lang.org> wr=
ote:
> Hi,
>
> Define "well-designed C API" first.

I meant it is not clear if a function is public or not.
include/ruby/intern.h contains too many functions for historical
reasons, and I would like to make a whitelist of public C APIs.

> Although I agree with some design flaws in CRuby's C API, I don't
> think it can be fixed by merely removing C functions. =A0I am against
> this proposal, since it will be more damage than benefit, by causing
> serious compatibility problems, at least temporarily.
>
> I don't think threatening community is good way to encourage
> evolution. =A0It is like taking C functions as ransom for
> documentation.

I agree. We need to keep compatibility.

I thought it is a good opportunity to make the whitelist because we
have some time until the release of Ruby 2.0.
But certainly it can be done without removing functions from the
header. We also have enough time to do a moderate process.
I'll propose another process later.

>
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =
=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0matz.
>
> In message "Re: [ruby-core:41321] [Proposal] C API arrangement"
> =A0 =A0on Sun, 27 Nov 2011 15:10:43 +0900, Yugui <yugui / yugui.jp> writes:
> |
> |Hi, mame-san
> |
> |As we discussed offline, I propose arrangement of C APIs.
> |Ruby's C APIs are not well-designed. Some of them have been
> |thoughtlessly exposed.
>
>



--=20
Yuki Sonoda (Yugui)
yugui / yugui.jp
http://yugui.jp