[
https://issues.apache.org/jira/browse/NUTCH-1086?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14640872#comment-14640872
]
Nikolai Vasilev commented on NUTCH-1086:
----------------------------------------
Hello Peter,
the deprecation warning you see tells that you should no longer create
HttpClient with DefaultHttpClient, and use HttpClientBuilder instead:
http://hc.apache.org/httpcomponents-client-ga/httpclient/apidocs/org/apache/http/impl/client/DefaultHttpClient.html
{code}
Deprecated.
(4.3) use HttpClientBuilder see also CloseableHttpClient.
{code}
There is a flaw in Fabio's implementation. By default DefaultHttpClient uses
[BasicConnectionManager|http://hc.apache.org/httpcomponents-client-ga/httpclient/apidocs/org/apache/http/impl/conn/BasicClientConnectionManager.html],
which is not supposed to manage connections in multithreaded environment.
Which is crucial for Nutch. The
[PoolingClientConnectionManager|http://hc.apache.org/httpcomponents-client-ga/httpclient/apidocs/org/apache/http/impl/conn/PoolingClientConnectionManager.html]
should be used instead.
In our project we launch Nutch at Amazon EMR and we suffered some weird
dependency clashing, when tried to rewrite protocol-httpclient to
HttpClient4.X. Unfortunatelly I have lost logs with errors and cannot tell
exactly what was wrong.
> Rewrite protocol-httpclient
> ---------------------------
>
> Key: NUTCH-1086
> URL: https://issues.apache.org/jira/browse/NUTCH-1086
> Project: Nutch
> Issue Type: Improvement
> Components: protocol
> Affects Versions: nutchgora, 1.5
> Reporter: Markus Jelsma
> Assignee: Fabio Santagostino
> Fix For: 2.4
>
> Attachments: Http.java, HttpResponse.java
>
>
> There are several issues about protocol-httpclient and several comments about
> rewriting the plugin with the new http client libraries. There is, however,
> not yet an issue for rewriting/reimplementing protocol-httpclient.
> http://hc.apache.org/httpcomponents-client-ga/
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)