[PATCH] HTTP ioslave keeping POST data when it shouldn't

Dawit A. adawit at kde.org
Fri Jun 14 00:37:09 BST 2002


On Thursday 13 June 2002 09:42, Best, Jan-Pascal van wrote:
> On June 12, Dawit A.  wrote:
> > On Tuesday 11 June 2002 10:11, Best, Jan-Pascal van wrote:
> > > Hi all (especially Dawit),
> > >
> > > Just found the patch by Dawit about persistent connections.
> >
> > It almost
> >
> > > solves my problem, in that it resets the POST buffer in some
> > > circumstances and I'd like it to do that in more circumstances,
> > > especially after webdav queries. It seems like the check
> >
> > for HTTP_POST
> >
> > > is not necessary. If it is removed, my webdav trouble disappears.
> > > Authentication for webdav still works now.
> >
> > Nope, feel free to apply.  The check for the request method
> > is a vestage from
> > the days when kio_http only dealt with the original HTTP
> > protocol and not WebDAV as well :)
>
> There's still a problem with this. Some methods, like listDir,
> for webdav set m_bufPOST to some document to be sent, before calling
> httpOpen(). Now httpOpen calls resetConnectionSettings, which
> clears m_bufPOST because it thinks it's a relic from another
> session. 
>
> Would it be a good idea to put a flag in HTTPRequest
> that orders resetConnectionSettings to leave alone
> m_bufPOST? This flag could be set by the appropriate DAV member
> functions of HTTPProtocol and also by readHeader(), when it
> receives a 401 or 407. After use, it is reset by
> resetConnectionSettings() and also in resetSessionSettings().
>
> What do you think?

Come to think of it you are correct this does not work for HTTP at all either 
if the connection with the site has to be re-established before posting the 
data!  I now know why I originally moved it into ::readHeader :)  I will fix 
this the correct way.  There is no need for yet another flag...

Regards,
Dawit A.





More information about the kfm-devel mailing list