How does this look? Diff is attached.
On Thursday, October 6, 2011 at 3:13 PM, David Hosier wrote: > Ok, thanks. I'm taking a stab at a patch right now. > > -- David Hosier > > On Thursday, October 6, 2011 at 2:52 PM, Oleg Kalnichevski wrote: > > > On Thu, 2011-10-06 at 14:23 -0700, David Hosier wrote: > > > On Thursday, October 6, 2011 at 1:40 PM, Oleg Kalnichevski wrote: > > > > On Thu, 2011-10-06 at 13:21 -0700, David Hosier wrote: > > > > > Understood. The library does not support the use case of obtaining > > > > > the entity of a response via the recommended usage for any response > > > > > other than a 2xx. > > > > > > > > BasicResponseHandler does not, hence, unsurprisingly, the name Basic. > > > > However, there is _nothing_ that prevents you from writing a custom, > > > > better ResponseHandler which handles response entities differently. > > > > While the ResponseHandler interface is indeed the recommended way of > > > > handling responses, the BasicResponseHandler is just its very basic > > > > implementation which was never meant to be used for anything else but > > > > the most simplistic use cases. No one in their sane mind should _ever_ > > > > convert an HTTP response to a string in a productive application. > > > > > > > > > Additionally, the javadoc for BasicResponseHandler is incorrect. > > > > > > > > What exactly is incorrect? If you think javadocs are not clear enough or > > > > specific enough I'll happily apply a patch if you submit one. > > > I went and looked more closely, and the issue is that I was looking at > > > the class-level javadoc for BasicResponseHandler. The class-level javadoc > > > indicates that the response body is read before throwing the exception on > > > status codes >=300. However, the javadoc for the handleResponse() method > > > does not indicate that the response body is read. The statement about > > > reading the response body on >=300 really only occurs when used with > > > HttpClient, and it's that class that actually does the reading. That's > > > how I read the code at least. > > > > Fair enough. This may sound misleading if taken out of the usual context > > of always using ResponseHandler together with HttpClient. I'll tweak the > > javadocs tomorrow (unless you would like to submit a patch yourself) > > > > Oleg > > > > > > Oleg > > > > > > > > > So now that I understand better how things work, I can take action > > > > > accordingly. Thanks for the responses. > > > > > > > > > > -- David Hosier > > > > > > > > > > On Thursday, October 6, 2011 at 11:40 AM, Oleg Kalnichevski wrote: > > > > > > > > > > > On Thu, 2011-10-06 at 11:31 -0700, David Hosier wrote: > > > > > > > I'm using the DefaultHttpClient to make the call, yes. I want to > > > > > > > use DefaultHttpClient with the ResponseHandler the way I am > > > > > > > supposed to. However, the API does not give me the ability to get > > > > > > > a hold of the Entity if the status code is 404, because it throws > > > > > > > an Exception which does not contain the entity value. I need the > > > > > > > Entity value, even if the call returns 404. As far as I can tell, > > > > > > > I cannot get the information I need from the API the way it is > > > > > > > designed to be used. Is that clearer? Is my assessment correct? > > > > > > > > > > > > Yes, it is intentional that the exception thrown does not contain a > > > > > > response body, because it would involve reading the entire body > > > > > > content > > > > > > into a memory buffer. > > > > > > > > > > > > Oleg > > > > > > > > > > > > > > > > > > > -- David Hosier > > > > > > > > > > > > > > On Thursday, October 6, 2011 at 11:29 AM, Oleg Kalnichevski wrote: > > > > > > > > > > > > > > > On Thu, 2011-10-06 at 10:47 -0700, David Hosier wrote: > > > > > > > > > I am using this to interface with some REST services. One key > > > > > > > > > to a good REST service is to never let something like a 404 > > > > > > > > > spit out the server's generic 404 HTML page in response to a > > > > > > > > > REST request. So my service instead returns an entity with > > > > > > > > > the 404 that says something like "Could not find alert > > > > > > > > > 12334". I should be able to show this response entity. > > > > > > > > > However, given the way the ResponseHandler works with > > > > > > > > > HttpClient, this is not possible, because the entity is not > > > > > > > > > part of the Exception that is thrown when the ResponseHandler > > > > > > > > > encounters a 404. Without manually reading the entity after > > > > > > > > > ResponseHandler throws an Exception, I would only be able to > > > > > > > > > show the fields that are contained in the Exception. That > > > > > > > > > means I could only show the text 'Not Found', which is hardly > > > > > > > > > meaningful since the status code of 404 already tells me that. > > > > > > > > > > > > > > > > You are using ResponseHandler to interface with some REST > > > > > > > > services > > > > > > > > without using DefaultHttpClient? I am sorry but it still makes > > > > > > > > no sense > > > > > > > > to me. You might as well handle responses from that service > > > > > > > > _any_ way > > > > > > > > you like without using a ResponseHandler. > > > > > > > > > > > > > > > > Oleg > > > > > > > > > > > > > > > > > -- David Hosier > > > > > > > > > > > > > > > > > > On Thursday, October 6, 2011 at 5:39 AM, Oleg Kalnichevski > > > > > > > > > wrote: > > > > > > > > > > > > > > > > > > > On Thu, 2011-10-06 at 02:59 -0700, David Hosier wrote: > > > > > > > > > > > Ok, I see what the difference is in this situation. I am > > > > > > > > > > > not passing the ResponseHandler to the execute() method. > > > > > > > > > > > I am actually calling handleResponse() on the > > > > > > > > > > > ResponseHandler manually. > > > > > > > > > > > > > > > > > > > > I honestly see no sense in doing so. ResponseHandler is > > > > > > > > > > pretty much > > > > > > > > > > useless without the resource management code in > > > > > > > > > > AbstractHttpClient. > > > > > > > > > > > > > > > > > > > > What is the reason you want to invoke #handleResponse > > > > > > > > > > manually? > > > > > > > > > > > > > > > > > > > > Oleg > > > > > > > > > > > > > > > > > > > > > The problem I have with the implementation is that I > > > > > > > > > > > return error messages on error conditions. With the way > > > > > > > > > > > this works, you can only get very basic information from > > > > > > > > > > > the HttpResponseException. For example, on a 404, it > > > > > > > > > > > looks like the Exception only contains 404 and 'Not > > > > > > > > > > > Found'. I am able to pluck out the entity when invoking > > > > > > > > > > > handleResponse() manually by simply consuming the entity > > > > > > > > > > > myself, but it's not possible to get the entity if the > > > > > > > > > > > ResponseHandler is passed to execute() and the status is > > > > > > > > > > > not 2xx. Am I off base here or is my analysis correct? > > > > > > > > > > > Would you recommend that if I really need the entity on a > > > > > > > > > > > non-2xx response that I just keep manually consuming the > > > > > > > > > > > entity? I'm not sure it would make sense for your library > > > > > > > > > > > to attempt to consume the entity in BasicResponseHandler > > > > > > > > > > > and try to add it as an > > > > > > > > > > > other fi > > > > > > > > > > > eld to the HttpResponseException. The AbstractHttpClient > > > > > > > > > > > code you linked me to would have to change if you did > > > > > > > > > > > that. > > > > > > > > > > > > > > > > > > > > > > -- David Hosier > > > > > > > > > > > > > > > > > > > > > > On Thursday, October 6, 2011 at 2:30 AM, David Hosier > > > > > > > > > > > wrote: > > > > > > > > > > > > > > > > > > > > > > > On Thursday, October 6, 2011 at 2:22 AM, Oleg > > > > > > > > > > > > Kalnichevski wrote: > > > > > > > > > > > > > On Wed, 2011-10-05 at 13:44 -0700, David Hosier wrote: > > > > > > > > > > > > > > Perhaps I'm wrong, but the code for > > > > > > > > > > > > > > BasicResponseHandler in httpclient 4.1.2 does not > > > > > > > > > > > > > > satisfy the javadocs as written. The javadoc states > > > > > > > > > > > > > > the following: > > > > > > > > > > > > > > > > > > > > > > > > > > > > "If the response code was >= 300, the response body > > > > > > > > > > > > > > is consumed and an HttpResponseException > > > > > > > > > > > > > > (http://hc.apache.org/httpcomponents-client-ga/httpclient/apidocs/org/apache/http/client/HttpResponseException.html) > > > > > > > > > > > > > > is thrown." > > > > > > > > > > > > > > > > > > > > > > > > > > > > However, the code does not do that: > > > > > > > > > > > > > > > > > > > > > > > > > > > > StatusLine statusLine = response.getStatusLine(); > > > > > > > > > > > > > > if (statusLine.getStatusCode() >= 300) { > > > > > > > > > > > > > > throw new > > > > > > > > > > > > > > HttpResponseException(statusLine.getStatusCode(), > > > > > > > > > > > > > > statusLine.getReasonPhrase()); > > > > > > > > > > > > > > } > > > > > > > > > > > > > > > > > > > > > > > > > > > > HttpEntity entity = response.getEntity(); > > > > > > > > > > > > > > return entity == null ? null : > > > > > > > > > > > > > > EntityUtils.toString(entity); > > > > > > > > > > > > > > > > > > > > > > > > > > > > The code clearly throws the Exception without > > > > > > > > > > > > > > reading the entity. So what happens is that if you > > > > > > > > > > > > > > get a non-2xx response, connections are never > > > > > > > > > > > > > > released as can be seen by enabling DEBUG logging > > > > > > > > > > > > > > for the library. Am I misreading the code or > > > > > > > > > > > > > > javadocs, or is this really broken? If I catch the > > > > > > > > > > > > > > Exception and then read the entity manually like > > > > > > > > > > > > > > shown above, I can see the connections being closed. > > > > > > > > > > > > > > > > > > > > > > > > > > > > -David > > > > > > > > > > > > > > > > > > > > > > > > > > Hi David > > > > > > > > > > > > > The resource management is taken care of by > > > > > > > > > > > > > HttpClient [1]. I do not > > > > > > > > > > > > > think BasicResponseHandler is broken. The whole point > > > > > > > > > > > > > of ResponseHandler > > > > > > > > > > > > > is to free the user from having to worry about > > > > > > > > > > > > > resource management and > > > > > > > > > > > > > response entities. > > > > > > > > > > > > Interesting. Thanks for the link to the code. I can > > > > > > > > > > > > assure you that in my situation however, that the > > > > > > > > > > > > connections are not getting closed. I'll take a closer > > > > > > > > > > > > look at the code and compare it to this linked code to > > > > > > > > > > > > see if I'm using the right stuff. My assumption at this > > > > > > > > > > > > point then is that I'm just doing something wrong. > > > > > > > > > > > > Thanks. > > > > > > > > > > > > > > > > > > > > > > > > > > Oleg > > > > > > > > > > > > > > > > > > > > > > > > > > [1] > > > > > > > > > > > > > http://hc.apache.org/httpcomponents-client-ga/httpclient/xref/org/apache/http/impl/client/AbstractHttpClient.html#930 > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > --------------------------------------------------------------------- > > > > > > > > > > > > > > To unsubscribe, e-mail: > > > > > > > > > > > > > > [email protected] > > > > > > > > > > > > > > (mailto:[email protected]) > > > > > > > > > > > > > > For additional commands, e-mail: > > > > > > > > > > > > > > [email protected] > > > > > > > > > > > > > > (mailto:[email protected]) > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > --------------------------------------------------------------------- > > > > > > > > > > > > > To unsubscribe, e-mail: > > > > > > > > > > > > > [email protected] > > > > > > > > > > > > > (mailto:[email protected]) > > > > > > > > > > > > > For additional commands, e-mail: > > > > > > > > > > > > > [email protected] > > > > > > > > > > > > > (mailto:[email protected]) > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > --------------------------------------------------------------------- > > > > > > > > > > > > To unsubscribe, e-mail: > > > > > > > > > > > > [email protected] > > > > > > > > > > > > (mailto:[email protected]) > > > > > > > > > > > > For additional commands, e-mail: > > > > > > > > > > > > [email protected] > > > > > > > > > > > > (mailto:[email protected]) > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > --------------------------------------------------------------------- > > > > > > > > > > > To unsubscribe, e-mail: > > > > > > > > > > > [email protected] > > > > > > > > > > > (mailto:[email protected]) > > > > > > > > > > > For additional commands, e-mail: > > > > > > > > > > > [email protected] > > > > > > > > > > > (mailto:[email protected]) > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > --------------------------------------------------------------------- > > > > > > > > > > To unsubscribe, e-mail: > > > > > > > > > > [email protected] > > > > > > > > > > (mailto:[email protected]) > > > > > > > > > > For additional commands, e-mail: > > > > > > > > > > [email protected] > > > > > > > > > > (mailto:[email protected]) > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > --------------------------------------------------------------------- > > > > > > > > > To unsubscribe, e-mail: > > > > > > > > > [email protected] > > > > > > > > > (mailto:[email protected]) > > > > > > > > > For additional commands, e-mail: > > > > > > > > > [email protected] > > > > > > > > > (mailto:[email protected]) > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > --------------------------------------------------------------------- > > > > > > > > To unsubscribe, e-mail: > > > > > > > > [email protected] > > > > > > > > (mailto:[email protected]) > > > > > > > > For additional commands, e-mail: > > > > > > > > [email protected] > > > > > > > > (mailto:[email protected]) > > > > > > > > > > > > > > > > > > > > > > > > > > > > --------------------------------------------------------------------- > > > > > > > To unsubscribe, e-mail: > > > > > > > [email protected] > > > > > > > (mailto:[email protected]) > > > > > > > For additional commands, e-mail: > > > > > > > [email protected] > > > > > > > (mailto:[email protected]) > > > > > > > > > > > > > > > > > > > > > > > > --------------------------------------------------------------------- > > > > > > To unsubscribe, e-mail: [email protected] > > > > > > (mailto:[email protected]) > > > > > > For additional commands, e-mail: > > > > > > [email protected] > > > > > > (mailto:[email protected]) > > > > > > > > > > > > > > > > > > > > --------------------------------------------------------------------- > > > > > To unsubscribe, e-mail: [email protected] > > > > > (mailto:[email protected]) > > > > > For additional commands, e-mail: [email protected] > > > > > (mailto:[email protected]) > > > > > > > > > > > > > > > > --------------------------------------------------------------------- > > > > To unsubscribe, e-mail: [email protected] > > > > (mailto:[email protected]) > > > > For additional commands, e-mail: [email protected] > > > > (mailto:[email protected]) > > > > > > > > > > > > --------------------------------------------------------------------- > > > To unsubscribe, e-mail: [email protected] > > > (mailto:[email protected]) > > > For additional commands, e-mail: [email protected] > > > (mailto:[email protected]) > > > > > > > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: [email protected] > > (mailto:[email protected]) > > For additional commands, e-mail: [email protected] > > (mailto:[email protected]) > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > (mailto:[email protected]) > For additional commands, e-mail: [email protected] > (mailto:[email protected])
--------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
