Tim Dudgeon created CAMEL-13092:
-----------------------------------
Summary: Gzipping HTTP response prevents writing chunked output
Key: CAMEL-13092
URL: https://issues.apache.org/jira/browse/CAMEL-13092
Project: Camel
Issue Type: Bug
Components: camel-servlet
Affects Versions: 2.23.1
Environment: I encounter this using the REST DSL with the servlet
component, but I imagine its a more general problem. I
'm using Camel 2.18.0, but I checked and the problem seem to also happen with
the latest version.
Reporter: Tim Dudgeon
This has been discussed in the users mailing list [1], but there was no
conclusion, and it looks like a bug or quirk so I'm raising this issue to try
to get resolution.
The issue is that when writing the response of an HTTP request using a
PipedInputStream to the Message body and with the `Content-Encoding: gzip` set
as a response header the data does not seem to be sent until the entire stream
has been written to the PipedInputStream. This means the client cannot start to
read the response until the entire set of data has been written. When that
header is not set the data is streamed immediately.
Also, symptomatic of this, when using `Content-Encoding: gzip` the
`Content-Length` header gets set, which can only happen after the entire stream
has been processed. Without the gzip header the 'Transfer-Encoding: chunked'
header gets set and there is no Content-Length as you might expect. I would
have expected this to happen even if the `Content-Encoding: gzip` header is set.
Is this a bug, or am I missing some magical way to ensure that the response is
streamed when it is being gzipped?
[1]http://mail-archives.apache.org/mod_mbox/camel-users/201901.mbox/%3C9cc4adf0-91f2-a111-edce-e7a18fde9fdf%40gmail.com%3E
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
