Mailing-List: contact dev-help@hc.apache.org; run by ezmlm
Precedence: bulk
Reply-To: "HttpComponents Project" <dev@hc.apache.org>
Received-SPF: pass (nike.apache.org: domain of cookieme@gmail.com designates
 209.85.210.179 as permitted sender)
DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=gmail.com; s=gamma;
        h=mime-version:in-reply-to:references:date:message-id:subject:from:to
         :content-type;
        b=RM4wnE83EtpBUN0JQ/qfreR661zOi/10Bdw3cpaBrlQVX05CmRPByFjtntQWA/LimL
         lx0Qn14k63TX/7Q7X1H1AOYBmuCRPzotqUGMsU49MtNJy1soogMqWBDoXdzNk0mEsCVJ
         JWAP3XiATN9i7YCJgpHp+wckALC4QEChGK284=
MIME-Version: 1.0
In-Reply-To: <BANLkTimuU_4Eveu6RMnj3ZkgBngpsSbndQ@mail.gmail.com>
References: <BANLkTin8aTN4XioDpAPoTFJB98EfjM42ZQ@mail.gmail.com>
	<BANLkTimuU_4Eveu6RMnj3ZkgBngpsSbndQ@mail.gmail.com>
Date: Tue, 14 Jun 2011 14:34:12 +0300
Message-ID: 
 <BANLkTikQEMvwOe0ER7qSUL7T134xTHx3M0YV+hw7K-09hyDEUQ@mail.gmail.com>
Subject: Re: Fluent HttpClient
From: "Xu, Lilu" <cookieme@gmail.com>
To: HttpComponents Project <dev@hc.apache.org>
Content-Type: multipart/alternative; boundary=bcaec51b1c2b3d3e9404a5aa6afa

--bcaec51b1c2b3d3e9404a5aa6afa
Content-Type: text/plain; charset=ISO-8859-1

Hi Thom,

thank you very much! Your suggestions really help me a lot.

Currently, I added a pair of methods named encodingOn() and encodingOff() to
support the URI escaping. Example codes are in the Encoding Swicher part of
WikiPage of UriBuilder [1].
[1] http://code.google.com/p/fluent-hc/wiki/UriBuilder


On Tue, Jun 7, 2011 at 7:22 PM, Thom Nichols <tmnichols@gmail.com> wrote:

> Looks great, I'm really excited about this project. (I think I mentioned
> before, I wrote an equivalent for Groovy:
> http://groovy.codehaus.org/modules/http-builder/ )
>
> Here's some more quick feedback:
>
> FluentResponse:
>
> Have a convenience method to get just the content-type value (without any
> encoding part) or just the encoding without the content-type.  Note that
> the
> content-encoding header is NOT the character set.
> http://www.w3.org/Protocols/rfc2616/rfc2616-sec3.html#sec3.5  To get the
> character set, you have to do this:
>
> https://fisheye.codehaus.org/browse/gmod/httpbuilder/tags/0.5.1/src/main/java/groovyx/net/http/ParserRegistry.java?hb=true#to128
> (see the next function down to get just the content-type without any
> encoding parameter.)
>
>
I have already updated the HttpResponse part [2], and now the content-type
value and charset can be accessed directly.
[2] http://code.google.com/p/fluent-hc/wiki/FluentResponse


> FluentRequest:
>
> One possibility would be to have a builder with a base URI, e.g.
>  FluentRequest req = RequestBuilder.build("http://hc.apache.org",
> GET_METHOD);
>
> but then allow the executed request to be against different paths, which
> makes for easy REST interface implementation.  e.g.
>
> FluentResponse resp1 = req.exec( '/httpcomponents-core-ga' )
> FluentResponse resp2 = req.exec( '/httpcomponents-client-ga' )
> I suppose the best way to implement this might be to implement clone() on
> FluentRequest so you could essentially take a base configured request
> object
> and copy it to add your specific changes.  Then you could call all the
> setPath, addQueryParameter functions without needing an exec() method that
> only can take pre-defined arguments.
>
> So something like this:
>  FluentRequest req = RequestBuilder.build("http://twitter.com/statuses/",
> GET_METHOD);
>
> resp1 = req.clone().setPath("user_timeline.xml").exec();
> resp2 = req.clone().setPath("update.xml").set(POST_METHOD).set( new
> StringEntity("status=updated%20status") ).exec()
>

I use another way to implement this job. The main idea is using UriBuilder's
setPath(String) method. You may find the example code on the bottom of the
WikiPage of UriBuilder [1].
[1] http://code.google.com/p/fluent-hc/wiki/UriBuilder


>
> A shortcut to create a url-encoded post body from a HashMap with the
> correct
> content-type set automatically would be super-handy :)
>
> Now we have the setQuery(HashMap) method to set the query part of an URI.
But if we do the similar things to create a POST body, how to detect the
content-type if the value set of the HashMap consisted of multiple types
instances? For example:
entity:HashMap {
  "key1" : value1,
  "key2" : value2
}
where value1 is an Integer or String, and value2 is a binary stream (e.g.
byte[])?


> XML parsing:
>
> (wishlist/ possibility) - ability to pass an XmlStreamReader instance
> instead of getting the response as DOM?  Or pass an XPath string to get a
> DOM element?
>
> Also, note that if you're going to allow built-in XML parsing, you really
> should have a catalog so that the HTTP DTD doesn't get requested every time
> -- this is a recipe for disaster because eventually the w3c server will
> start rejecting HTTP requests to the DTD URI, and the XML parsing will
> fail.  See:
> http://www.w3.org/blog/systeam/2008/02/08/w3c_s_excessive_dtd_traffic/
>
> I am not quite familiar with the XML parsing. Why does the built-in XML
parsing need to request to W3C server every time? Does it mean that we would
not able to parse XML without an Internet connection?


> Here's an example of how to use an XML catalog:
>
> https://fisheye.codehaus.org/browse/gmod/httpbuilder/tags/0.5.1/src/main/java/groovyx/net/http/ParserRegistry.java?hb=true#to103
>
> https://fisheye.codehaus.org/browse/gmod/httpbuilder/tags/0.5.1/src/main/resources/catalog
>
> Another wishlist item - built-in support for common content-encodings.
> (Someone else may have mentioned it.)  Here's my implementation:
>
> https://fisheye.codehaus.org/browse/gmod/httpbuilder/tags/0.5.1/src/main/java/groovyx/net/http/ContentEncoding.java?hb=true
>
> https://fisheye.codehaus.org/browse/gmod/httpbuilder/tags/0.5.1/src/main/java/groovyx/net/http/GZIPEncoding.java?hb=true
>
> https://fisheye.codehaus.org/browse/gmod/httpbuilder/tags/0.5.1/src/main/java/groovyx/net/http/DeflateEncoding.java?hb=true
>
> Thanks!  Although I wrote a client for Groovy, there are plenty of times
> when I have to use Java (think Android) and this would be super-handy to
> have.  Good work so far!
>
> -Thom
>
>
> On Wed, Jun 1, 2011 at 11:40 AM, Xu, Lilu <cookieme@gmail.com> wrote:
>
> > Hi,
> >
> > I almost finished the request execution module and most classes have been
> > re-designed. Please have a look at the wiki pages [1] of this project and
> > drop me some comments.
> > The wiki pages contain some simple examples to illustrate how to use the
> > facade API of httpclient.
> >
> > [1] http://code.google.com/p/fluent-hc/w/list
> > --
> > Best regards,
> > Lilu
> >
>


-- 
Best regards,
Lilu

--bcaec51b1c2b3d3e9404a5aa6afa--