-dump_source
              dump page source into stdout
       -dump_extra
              dump HEAD, source, and extra information into stdout

etc. have the capability to send raw gzip matter to the terminal.

So one needs
$ w3m -dump_source http://...|zcat
to read it.

However in the case of -dump_extra, the first part is stdout plain, the
last part is gzipped, so one cannot just use zcat!

The solution is for w3m to take care of this for us internally for all
of these options, and add an -really_raw option for those who want
unprocessed output.

I never seemed to have this problem with GET(1).

Brian Keck wanted to send this, but it ended up elsewhere.
BK> handling programs to handle it, such as

BK>   % w3m -mime http//... >1
BK>   % folder +./.
BK>   % mhstore 1
BK>   storing message 1 as file 1.x-gzip
BK>   % file 1.x-gzip
BK>   1.x-gzip: gzip compressed data, was ...

BK> Currently the 'HTTP/1.1 200 OK' line makes the header not legal MIME,
BK> but obviously this is nice to have, so maybe it could be prefixed by
BK> something like 'HTTP-Status-Line:'.

BK> Currently mhstore doesn't work properly on the -dump_extra output
BK> even with the 'HTTP/1.1 200 OK' line deleted.  Maybe because there's no
BK> Content-Transfer-Encoding (a MIME, not HTTP, header).

BK> Maybe binary bodies should be base64 encoded, to make things less ugly
BK> in your pager.

base64 is still ugly. source means source, not any wrapped then half
unwrapped version.

BK> Brian Keck