users@grizzly.java.net

Re: (Cometd issue with encoded json)Re: [Jean-Francois Arcand's Blog] New Comment Posted to 'Updating Grizzly version in GlassFish v3'

From: Jeanfrancois Arcand <Jeanfrancois.Arcand_at_Sun.COM>
Date: Thu, 21 May 2009 10:37:45 -0400

Salut,

neekfenwick wrote:
>
> Jeanfrancois Arcand-2 wrote:
>> It is not Prelude build :-) It is v3. Which version have you downloaded
>> from here:
>>
>> http://download.java.net/glassfish/v3/promoted/
>>
>
> Ah I see, 'v3 promoted' is the 'latest pre release v3', and 'v3 prelude' has
> always been the "beta" version of glassfish V3. This is the first time I've
> strayed from the v3 prelude path. I have been confusing myself by assuming
> that nothing would be non-prelude.
>
> I downloaded
> http://download.java.net/glassfish/v3/promoted/latest-glassfish-unix.sh
> which is 52M, larger than any other unix .sh so I have no idea how it
> corresponds to the other downloads in that directory.
>
>
> Jeanfrancois Arcand-2 wrote:
>> Looks like the Bayeux spec added this timeout value recently. Can you
>> file an issue here:
>>
>> https://grizzly.dev.java.net/issues/
>>
>> And attach a small test case?
>>
>
> I don't think we're on the same page, here. The timeout value I referred to
> above is returned from the mortbay servlet.

yes, we are on the same :-) What I means is Grizzly Servlet isn't
returning the value at all. At the time we have implemented the spec, I
don't recall seeing that timeout value.

   A change in the Bayeux protocol
> that requires me to file an issue would concern something passed _to_ your
> servlet, not returned _from_ it, surely?
>
> My understanding of comet is that my app makes a request to the servlet,
> which will hold that request for a long period of time if there is no
> information to pass back. The mortbay servlet accepts one /meta/connect
> (with id=1) and returns, but the second (id=2) doesn't return, firebug shows
> a Post but no Response. However, the grizzly cometd servlet returns
> immediately from every /meta/connect Post.

Yes, that's a spec change.

>
> There seems to be nothing passed _in_ in these Post requests from the client
> that would tell the servlet how long to timeout each Post, so I presume it
> is an internal configuration.

Yes the connection is suspended.

>
> I'll file an issue anyway, and as always, appreciate your time :)

Thanks!

-- Jeanfrancois