[
https://issues.apache.org/jira/browse/NIFI-1086?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15007146#comment-15007146
]
Aldrin Piri commented on NIFI-1086:
-----------------------------------
OkClient's isSuccessful excludes 300 as per
{code}
public boolean isSuccessful() {
return code >= 200 && code < 300;
}
{code}
Regardless, minor point and fine either way.
Concerning the status code 0, I think it's unlikely people are expecting an
exception as the processor class itself isn't accessible by external means. To
that end, when exceptions happen they leave the user with no recourse to handle
the item in question as it is simply kicked back to its respective input (or at
least it is never removed from its input and continuously retried). I think
that processors with this behavior and no chance for interaction serve more as
a bug than anything else as there is no way for the user to handle those items
short of expiring the queued item from the supplying connection. Would
certainly invite other thoughts on the topic as I definitely appreciate the
sensitivity to backwards compatibility and breaking changes.
> Refactor InvokeHttp
> -------------------
>
> Key: NIFI-1086
> URL: https://issues.apache.org/jira/browse/NIFI-1086
> Project: Apache NiFi
> Issue Type: Improvement
> Reporter: Joseph Percivall
> Assignee: Joseph Percivall
> Fix For: 0.4.0
>
> Attachments: NIFI-1086_02.patch, NIFI-1086_rebasing_to_master.patch
>
>
> InvokeHttp currently uses Java's HttpUrlConnection, which is lacking in it's
> features and ease-of-use. In order to support all the current InvokeHttp
> pending tickets it's clear that a new underlying library is needed.
> OkHttp looks to be a promising library that focusing on individual
> transactions (as opposed to Apache's HttpClient that focuses more on
> sessions).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
