From: Robert Klemme <bob.news@...>
Date: 2005-11-15T00:42:16+09:00
Subject: Re: trouble reading a gzipped xml-file

Guido de Melo wrote:
> Robert Klemme wrote:
>> You're not closing f here.
>>
> [...]
>>
>> I'm not so sure that it fails just because of this.
>
> It shouldn't, because ruby will determine the end of the file by
> itself and close the handle on exiting.

I didn't mean to say that it fails because of the open file.  My point
with the first remark was that it's a good habit to open files for only as
long as they are actually used.  The block form is the idiom of choice
here: it's not much longer as a simple File.open() or File.new() and it
ensures the file is always properly closed.

>> Did you try this?
>>
>> xmldoc = Zlib::GzipReader.open('file.dia') {|gz| REXML::Document.new
>> gz}
>
> irb(main):003:0> xmldoc = Zlib::GzipReader.open('file.dia') {|gz|
> REXML::Document.new gz}
> RuntimeError: Zlib::GzipReader is not a valid input stream.  It must
> be either a String, IO, StringIO or Source.
>
> This doesn't work either, I'm afraid...

I guess this is because GzipReader doesn't inherit IO:

>> Zlib::GzipReader.ancestors
=> [Zlib::GzipReader, Enumerable, Zlib::GzipFile, Object, Kernel]

But you can do this

xmldoc = Zlib::GzipReader.open('file.dia') {|gz| REXML::Document.new(
gz.read )}

Btw, you see differing output because you use different printing methods:

Zlib::GzipReader.open('file.dia') {|gz|
   print gz.read
}

vs.

p s

If there is no exception during GZIP reading I guess there might be a bug
somewhere.  As a test I'd write the gunzipped content to another file and
do a diff on the plain xml and this output to see whether GzipReader
actually yields the same content.

Btw, did you actually try to make REXML read the first variant?  Maybe you
have a problem in your XML file.

Kind regards

    robert