hc-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Oleg Kalnichevski <ol...@apache.org>
Subject Re: Jakarta HttpXXX Charter
Date Mon, 29 Aug 2005 08:34:25 GMT
Here's my take for what it is worth

Goals and scope
(1) Low level HTTP components with the focus on transport aspects of
HTTP. The framework is content agnostic. No plans to define an
application API on top of it
(2) Continued development of Jakarta HttpClient based on the new
component framework. We get to keep the Jakarta HttpClient brand
(3) Collaboration with other projects if there is interest to develop
other HTTP services based on the component framework.

Release strategy
(1) HTTP component packages: httpcommon.jar, httpcookie.jar,
(2) httpclient.jar that requires dependent HTTP modules as external
(3) httpclient-full.jar that includes all dependent HTTP modules

Potential points of contention
(1) Lightweight server and other purely server side development efforts

During the squabble on the PMC list I _personally_ have not heard a
single convincing argument as to why we should be not permitted to
pursue such efforts save the breach of the project chapter.
* Sun themselves acknowledged that there's more to HTTP than
  javax.servlet API by initiating development of lightweight HTTP server
  for the next major java release
* We already have cases of HttpClient and SimpleHttpServer test code
  being used to develop server side HTTP services. This is not something
  we just invented due to boredom, excessive spare time, and lack of
  social life
* There enough Apache endorsed projects that compete with other Apache
  projects as well standards defined by Sun. Why should we be singled

Essentially I think this is a matter of whether we, as a community, want
to do that or not. If we all see this as a Good Thing driven by the user
demand, we should leave the option of pursuing server side development
open. My main concern is we may overstretched ourselves way too much. We
are presently down to 3-4 active committers, which is an absolute
minimum. At the same time I see the project scope exception as an
opportunity to bring more people on board

(2) The HTTP connector based on Tomcat Coyote API presently in SVN can
be kept as 'contrib', that is as a reference material without official
releases. If needed it can be moved to a non-Apache repository. Tomcat
is by far more important Apache project and their interests should take
precedence over those of HttpClient



On Mon, Aug 29, 2005 at 08:48:37AM +0200, Ortwin Gl?ck wrote:
> My 0.02?:
> Focus on ease of use:
> * Create a HttpClient facade that is simple to use for the most simple 
> case (think of HttpURLConnection)
> * Create a HttpClient that is feature rich and easy to use
> * Integrate standard technologies (like JMX) as optional dependencies
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: httpclient-dev-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: httpclient-dev-help@jakarta.apache.org

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

View raw message