[Prev][Next][Index][Thread]

Content Encoding : a libwww bug?



Hi all

I am working on a browser based on libwww-5.0.
This would be a vrml browser, but this is of little concern.
I have the following problem.

I've created an HTStream that can ungzip data.
I've registered it as a content-decoder.

When I'm reading data from an HTTP server, everything comes out fine.
When I'm reading data from a file, my decoder stream is not inserted
into the stream stack.

Trace shows that libwww knows both content-type and 
content-encoding but nevertheless doesn't plug my strem in. It sez:

With http:// access:

...
  MIME header. Content-type: x-world/x-vrml
  MIME header. Content-encoding: x-gzip
...
  Building.... C-T stack from x-world/x-vrml to www/present
  Building.... C-E stack
...

With file:/ access:

...
  Get Binding. Look for 'wrl' Found!
  Get Binding. Look for 'gz' Found!
  Get Binding. Result for '/foo.wrl.gz' is: type='world/vrml',
  encoding='gzip', cte='unknown', language='unknown' with quality 1.00
...
(no interesting messages thereafter)

The question is:  is it me, or it's the bug in libwww?

Best regards.

-- 
mailto:anatoli@ptc.com (may or may not be speaking for my employer)