From: Vladimir Sizikov <vsizikov@...>
Date: 2008-06-05T18:59:48+09:00
Subject: [ruby-core:17118] Re: [PATCH] Iconv#iconv(str, start, length) doesn't really convert str[start, length]

Hi,

On Thu, Jun 5, 2008 at 1:57 AM, Vincent <vincentlu@gmail.com> wrote:
> I Agree. That patch should have been created for 1.9 or later version
> (if the core team
> decided to have this change.)
>
> I was going to create a documentation patch for ruby_1_8, and another patch for
> trunk with incompatible changes, but the latest revision 16809 just
> shows up today.

OK, let's recap the current situation:

1. ruby_1_8_6 branch: old behavior, out of sync with docs
2. ruby_1_8_7 branch: old behavior, out of sync with docs
3. ruby_1_8 branch: newer, backward incompatible and broken behavior:

a) it is clearly broken:
ruby -riconv -ve "p Iconv.open('ascii', 'ascii').iconv('abcde')"
ruby 1.8.7 (2008-06-04 patchlevel 5000) [i686-linux]
""

and

ruby -riconv -ve "p Iconv.open('ascii', 'ascii').iconv('abcde', 2, 7)"
ruby 1.8.7 (2008-06-04 patchlevel 5000) [i686-linux]
"cde\000so\000"

b) it is still not fully in sync with the docs:
"length:	conversion length; nil means whole the string from start", but
/opt/ruby1.8/bin/ruby -riconv -ve "p Iconv.open('ascii',
'ascii').iconv('abcde', 0, nil)"
ruby 1.8.7 (2008-06-04 patchlevel 5000) [i686-linux]
""

c) Sooner or later it will be propagated to 1_8_7 branch and released
as a patch level update
to 1.8.7, right? So we'll end up with rather different behaviors even
for the same version (1.8.7)

4. trunk: newer, backward incompatible, behavior, and:

a) It differs from both 1_8_6 and 1_8 behaviors, at least it correctly
handles the common case:
ruby -riconv -ve "p Iconv.open('ascii', 'ascii').iconv('abcde')"
ruby 1.9.0 (2008-06-05 revision 0) [i686-linux]
"abcde"

b) It is still out of sync with the docs, w.r.t. length argument, see 3b)

c) It doesn't handle the out of bounds length correctly, see 3a)

Oh, my! :) So, it seems that anybody who'd like to write cross-version
compatible scripts or
even cross-patchlevel compatible scripts, should not use this method at all.

Thanks,
  --Vladimir