On Tue, Dec 30, 2008 at 8:15 PM, Eric Hodel <drbrain / segment7.net> wrote:
> On Dec 29, 2008, at 07:14 AM, Yasuhiro MISE wrote:
>>
>> 2. edit ``zipped_stream(entry)'' (line 203--216) as follows:
>>
>> def zipped_stream(entry)
>>  if defined? Rubinius then
>>    zis = Zlib::GzipReader.new entry
>>    dis = zis.read
>>    is = StringIO.new(dis)
>>  else
>>    # This is Jamis Buck's Zlib workaround for some unknown issue
>>    entry.read(10) # skip the gzip header
>>    zis = Zlib::Inflate.new(-Zlib::MAX_WBITS)
>>
>>    ######### EDIT START ###
>>    #is = StringIO.new(zis.inflate(entry.read))
>>
>>    __buf = ""
>>    while __temp = entry.read( 128 * 1024 ) do
>>      __buf << zis.inflate( __temp )
>>    end
>>
>>    is = StringIO.new( __buf )
>>    ######### EDIT END #####
>>
>>  end
>> ensure
>>  zis.finish if zis
>> end
>
> For what it's worth, I'd really like to see the root cause of this bug fixed
> so this method can be returned to a simpler form:
>
> def zipped_stream(entry)
>  zis = Zlib::GzipReader.new entry
>  dis = zis.read
>  is = StringIO.new(dis)
> ensure
>  zis.finish if zis
> end
>
> Or even:
>
> def zipped_stream(entry)
>  Zlib::GzipReader.new entry
> end
>
> I'm not sure where the bug actually exists, but I'd really like to see the
> bug fixed.  I'm fairly certain the bug doesn't exist in RubyGems and I'm
> tired of having to continually apply bandaids to RubyGems.
>
> While we're on the subject, what help can I provide so I can eventually get
> back to the bandaid-free version?
>

Gregory Brown offered backport the zlib tests from 1.9 to 1.8 in
[ruby-core:20602]

Maybe those can help pinpoint the root of the problem.

I offer my Windows resources to see if the problem is located on this
side of the bench.

-- 
Luis Lavena
AREA 17
-
Perfection in design is achieved not when there is nothing more to add,
but rather when there is nothing more to take away.
Antoine de Saint-Exupéry