Re: [BUG] wwwoffle misinterprets # symbol in 'Location' header

Maxim Kirillov <max630-JGs/[email protected]> writes:

> On Wed, Sep 21, 2005 at 07:14:53PM +0100, Andrew M. Bishop wrote:
> > It is not defined in HTTP specification to have a header
> > with a fragment identifier.  Actually to be honest the
> > only mention of the word fragment is the sentence "The URI
> > MUST NOT include a fragment." for a Referer header.  The
> > HTTP protocol does not handle fragment identifiers
> > anywhere, it is only browsers that understand what they
> > are and they are only valid for HTML type documents.
> 
> Hmm, you seem to be right. Sad. I've detected this issue on
> sending LiveJournal comments - it redirects to the new
> comment.
> 
> If anybody's interested:
> http://zilla.livejournal.org/show_bug.cgi?id=2528
> 
> What do you think should be done here (in WWWOFFLE)? How
> would you prefer it to react?

Before your e-mail I had already changed the code to handle a fragment
in strange places better than it was previously.  This case was not
one of the ones that I changed.  I have now changed the code so that
it will remove fragment identifiers in all places that they should not
be.

> BTW, does wwwoffle follows the "meta-refresh" html link?
> Does it consider (remove) the fragment there? Is there any
> standard regarding this? "html4 specification"
> (http://www.w3.org/TR/html4/struct/global.html#edef-META)
> seems to say nothing about it.

Yes, WWWOFFLE does follow a meta-refresh html link when it is
fetching.  It didn't handle the fragment correctly before, but it will
do now.

-- 
Andrew.
----------------------------------------------------------------------
Andrew M. Bishop                             amb-Uxr6IM1mbv2TY6FTCsQk+9Bc4/[email protected]
                                      http://www.gedanken.demon.co.uk/

WWWOFFLE users page:
        http://www.gedanken.demon.co.uk/wwwoffle/version-2.8/user.html