RE: RangeFilter and redirects

2009-07-05 Thread Jerome Louvel
Hi David,

Thanks for the report. I've fixed the redirection conflict in SVN trunk and
1.1 branch. Let us know if it works better now.

Best regards,
Jerome Louvel
--
Restlet ~ Founder and Lead developer ~ http://www.restlet.org
Noelios Technologies ~ Co-founder ~ http://www.noelios.com


-Message d'origine-
De : webp...@tigris.org [mailto:webp...@tigris.org] 
Envoyé : mercredi 1 juillet 2009 20:09
À : discuss@restlet.tigris.org
Objet : RangeFilter and redirects

I ran across a problem that was causing the Daemon threads in our
WadlComponent-based server to enter a busy loop.  Essentially, if a Range
header was in the HTTP GET request but the server wanted to return a
redirection (e.g. via response.redirectSeeOther()), the entity would be
wrapped by a RangeRepresentation.  Tunnel in to that, and
RangeInputStream#read would loop here:

// Reach the start index.
while (!(position = startIndex)) {
position += skip(startIndex - position);
}

Aside from the likelihood that there's also a bug in the handling/wrapping
of a response entity that's too small for the range, my first question here
had to do with the status code of the response itself.  Should responses
that are not successful (2xx) be subject to wrapping in a
RangeRepresentation?

Thanks,
David

--
http://restlet.tigris.org/ds/viewMessage.do?dsForumId=4447dsMessageId=23671
49

--
http://restlet.tigris.org/ds/viewMessage.do?dsForumId=4447dsMessageId=2368108


RangeFilter and redirects

2009-07-01 Thread webpost
I ran across a problem that was causing the Daemon threads in our 
WadlComponent-based server to enter a busy loop.  Essentially, if a Range 
header was in the HTTP GET request but the server wanted to return a 
redirection (e.g. via response.redirectSeeOther()), the entity would be wrapped 
by a RangeRepresentation.  Tunnel in to that, and RangeInputStream#read would 
loop here:

// Reach the start index.
while (!(position = startIndex)) {
position += skip(startIndex - position);
}

Aside from the likelihood that there's also a bug in the handling/wrapping of a 
response entity that's too small for the range, my first question here had to 
do with the status code of the response itself.  Should responses that are not 
successful (2xx) be subject to wrapping in a RangeRepresentation?

Thanks,
David

--
http://restlet.tigris.org/ds/viewMessage.do?dsForumId=4447dsMessageId=2367149