jsr340-experts@servlet-spec.java.net

[jsr340-experts] Re: [servlet-spec users] Re: Re: Re: Initial draft of NIO proposal

From: Rajiv Mordani <rajiv.mordani_at_oracle.com>
Date: Fri, 11 Nov 2011 15:55:40 -0800

Ok here is the updated proposal with the comments and put it on the wiki at

http://java.net/projects/servlet-spec/pages/NIOAPI.

Please review and provide any comments.

Thanks

- Rajiv



On 10/24/2011 4:17 PM, Rajiv Mordani wrote:
> That's a good question. I think we should add a ServletReader and
> ServletWriter. What do others think?
>
> - Rajiv
>
> On 10/24/2011 11:12 AM, Shing Wai Chan wrote:
>> On 10/24/11 10:23 AM, Rajiv Mordani wrote:
>>> Trying to close on this. Since there wasn't a strong preference to
>>> use the NIO2 style API or the proposed API, I would like to move
>>> forward with the proposal with some of the changes
>>> that were suggested to the API. We can put a note in the spec asking
>>> for feedback when we
>>> do an early draft to get the community input.
>>>
>>> The changes that I have on the list to the proposal as of now are -
>>>
>>> * try and move the methods from NIOInputSource and NIOOutputSink
>>> to ServletInputStream and ServletOutputStream
>>>
>> In this case, will we have ServletReader and ServletWriter classes, too?
>> (Note that we have "BufferedReader getReader()" in ServletRequest.)
>>
>> Or will we only have NIO streams/bytes related API as in JDK?
>>
>> Shing Wai Chan
>>>
>>> * Rename any new NIOxxx API to AsyncIOxxx (although technically it
>>> really isn't Async but more non-blocking)
>>>
>>> Thanks
>>>
>>> - Rajiv
>>>> On 15 September 2011 18:00, Remy Maucherat<rmaucher_at_redhat.com> wrote:
>>>>> I don't quite understand why this is a good design over the usual
>>>>> blocking IO.
>>>> I don't represent that NIO.2 style is a better design, just that it is
>>>> the approach taken by java 7 and we should at least evaluate it.
>>>> The example I gave is definitely not good design - nor should it be
>>>> taken as representative of how NIO.2 style apis are used. It was just
>>>> a simple example.
>>>>
>>>> Perhaps we could get somebody from the NIO.2 team to give us a
>>>> presentation of why they did their API that way and we could then see
>>>> if the reasons they give are applicable to servlet API.
>>