From: Luis Lavena <luislavena@...>
Date: 2009-01-11T04:49:49+09:00
Subject: Re: Pure Ruby Zlib::GzipWriter

On Jan 10, 4:21 pm, Michal Suchanek <hramr...@gmail.com> wrote:
> On 10/01/2009, Luis Lavena <luislav...@gmail.com> wrote:
>
> > On Jan 10, 2:47 pm, Heesob Park <pha...@gmail.com> wrote:
> >  > Hi,
>
> > > 2009/1/10 Luis Lavena <luislav...@gmail.com>:
>
> >  > > On Jan 10, 8:15 am, Heesob Park <pha...@gmail.com> wrote:
> >  > >> 2009/1/10 Roger Pack <r...@doachristianturndaily.info>:>> Anyone knows any web application for bounties? I'm interested in
> >  > >> >> contribute to this one and also one for Readline!!!
>
> >  > >> > Yeah I'm in for Readline :)
> >  > >> > Note that highline has a getc for windows, and cmd is also readline
> >  > >> > like, so I believe it is possible :)
>
> >  > >> > -=r
> >  > >> > As a note I get the same error message Luis does.
>
> >  > >> I guess you are looking for ruby_readline [1] .
>
> >  > > Hello Park,
>
> >  > > Actually I'm looking to an API-compatible replacement, which should
> >  > > work with IRB and auto complete capabilities.
>
> >  > > Not only that, but a solution that can work across platforms,
> >  > > including Windows :-)
>
> >  > > There is a solution called RawLine [1] but is not API-compliant with
> >  > > ReadLine.
>
> >  > > Thank you again for your pointer.
>
> >  > > [1]http://rubyforge.org/projects/rawline/
>
> >  > I found the pure Perl implementation of the readline libraries called
> >  > libterm-readline-perl-perl [1]
>
> >  > I hope some one could port it into Ruby.
> >  > If no one port it until the end of this month, I will port it.
>
> > Well, I don't have the time, but I can give some contribution for it,
> >  so :-D
>
> >  > BTW, I succeeded "gem build" as well as "gem install" with my zlib.rb
> >  > and rbzlib.rb.
> >  > You should test it with the latest zlib.rb and rbzlib.rb at  [2]
>
> > I've run the specs of Zlib for your code:
>
> >  Finished in 0.328 seconds
>
> > 38 examples, 25 failures
>
> I guess it's expected. Different implementations give different
> compression results.
>

The failures are related to wrong number of arguments.

if a replacement for Zlib is developed, it must comply with Zlib specs
located in rubyspec:

http://github.com/rubyspec/rubyspec/tree/b83d7e5f40659f551c7e94667ffeeb0ea7c6bcfe/library/zlib

It should implement the exact same compression strategies. If not, is
not a "drop-in" replacement.

> Of course, the spec might be more high-level but then an
> implementation that does not compress at all would likely work as
> well.

The spec is being used across IronRuby, Rubinius, JRuby and the Ruby C
(MRI). So the pure-ruby replacement must comply.

--
Luis Lavena