[jsr-314-open] [ajax] My AI on PartialResponseWriter.

Ken Paulsen Ken.Paulsen at Sun.COM
Thu Dec 10 15:18:47 EST 2009



Jim Driscoll wrote:
>
>
> On 12/9/09 4:06 PM, Ken Paulsen wrote:
>>
>> 4b - Why not add them to the queue described in 4a instead of throwing
>> an exception (of course detect loops)?
>
> I thought of that, several reasons not to:
>
> 1)  There's a good use case to allow nesting one deep.  There's no 
> similar use case for multiple levels of depth.  (If you have one, let 
> me know).
Was that use case to call the other listed methods during "startUpdate"?

If you update component X which doing so causes component Y to be 
updated?  What if Y effects Z and so on?  You list startUpdate as one of 
the ones that could go an additional level deep... so doesn't that imply 
that it could go 'n' levels deep?  Or perhaps I don't understand correctly.

Thanks,

Ken
> 2) Ordering becomes complicated if you allow multiple levels of depth. 
> Complicated is bad - this proposal is already complicated, but no more 
> than absolutely necessary.
> 3) We can always add it later, but cannot take it away if we add it now.
>
>
>
>> Separate but related: I would like to override the behavior of
>> PartialResponseWriter. How is this currently pluggable? I have discussed
>> this a couple times with Ryan and have a solution working currently
>> where I simply call PartialViewContext.setPartialRequest(false) early
>> enough in the lifecycle, then do what I need... it's a hack, though. The
>> server-side is written to match the JS handling on the client-side... I
>> would like formal ways documented in the spec to customize the behavior
>> on the server and the JS on the client.




More information about the jsr-314-open-mirror mailing list