hc-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Oleg Kalnichevski <ol...@apache.org>
Subject Re: Server Cookies
Date Wed, 27 Apr 2011 18:52:31 GMT
On Wed, 2011-04-27 at 13:33 -0400, Bill Speirs wrote:
> On Tue, Apr 26, 2011 at 5:33 PM, Oleg Kalnichevski <olegk@apache.org> wrote:
> > For historical, mainly political, reasons HttpComponent project was
> > restricted from developing server side code which could be seen as
> > overlapping or even competing with other Apache projects.
> 
> In that vein, is there a separate package/project which is only the
> HTTP transport layer used by say Tomcat?
> 

Last time I looked at Tomcat's source, which admittedly was a long time
ago, there was something called Coyote connector. Coyote is basically a
low level server-side HTTP transport.


> > This restriction probably no longer makes any sense, so given enough interest
> > such components could be developed, but there is simply not enough
> > resources to do so.
> 
> If Tomcat (or any other Java HTTP server) were to use HttpCore as it's
> HTTP transport layer, then I would think there would be plenty of
> resources... my $0.02 not knowing any of the politics :-)
> 

Not entirely impossible but I would consider a Martian invasion more
likely. There is simply no convincing reason for any of established
Servlet engines to drop their custom HTTP code in favor of HttpCore.
Anyone to suggest such an idea would be laughed at. 


> > Mainly because there are different cookie standards and different ways
> > of parsing and formatting cookie headers.
> 
> Understood, but these different ways/standards appear to be well
> encapsulated/developed on the client side, just not the server side.
> 
> If someone were to put effort into developing these "server" cookies;
> what type of design decisions should go into it? It seems as though
> the ClientCookie (and classes BasicClientCookie & BasicClientCookie2)
> have all the methods required for a server cookie, only the toString
> method would need to be overwritten to generate the Set-Cookie (and
> Set-Cookie2) header strings respectively.
> 

While quite similar server side cookies are still different beasts. Some
attributes that make sense on the client side are meaningless on the
server side. I think server side cookies should be better represented by
specific interfaces / classes tailored specifically for server side
use. 

Hope this helps

Oleg 



---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
For additional commands, e-mail: dev-help@hc.apache.org


Mime
View raw message