So far I haven't heard back from others in the EG other than the folks
that participated in the user alias (Mark, and Alex). If others don't
have an opinion on this then I would like to proceed and make this
change in the spec.
Thanks
- Rajiv
On 12/3/12 10:47 AM, Rajiv Mordani wrote:
> Not sure how many of you'll are on the users alias so am moving the
> discussion to the EG mailing list (which the users will also be able
> to see).
>
> There is a thread on tying the new io APIs to async request processing
> APIs (from Servlet 3.0) only on the users alias. My first hunch is
> that it is OK but need to think a little more about
> this. But in the mean time I wanted to get others opinion on the topic.
>
> - Rajiv
>
>
> -------- Original Message --------
> Subject: [servlet-spec users] Re: Relationship between the
> non-blocking API and Servlet 3.0 async processing
> Date: Sat, 01 Dec 2012 16:51:01 +0000
> From: Mark Thomas <markt_at_apache.org>
> Reply-To: users_at_servlet-spec.java.net
> To: users_at_servlet-spec.java.net
>
>
>
> On 30/11/2012 23:11, Shing Wai Chan wrote:
> > NonBlocking IO is different from Async IO here.
> > In our JavaOne presentation, we do try to put Async and Non Blocking IO
> > together.
> > You can take a look at Example 3 of the Non-blocking IO in the
> > presentation below
> >https://blogs.oracle.com/swchan/entry/javaone_2012_what_s_new
>
>
> I can see the point of supporting NIO with the HTTP upgrade mechanism.
> I can see the point of supporting NIO with Servlet 3 Async.
>
> What I don't see is the point of supporting non-blocking IO in a
> traditional / synchronous servlet. The container thread is allocated
> until the response is completed so there is no benefit in use NIO to
> read the request body, nor to write the response body.
>
> I think it makes sense to limit the NIO APIs to Async and upgrade.
>
> Mark
>
>
>
>