Re: SCTP Socket API - Objections to proposed extensions

Mark Butler wrote:
> Randall Stewart wrote:
> 
>> Usually a pdapi trunction is closely followed by an ABORT or
>> COMM_LOST more correctly... one could come up with the
>> same thing without it.. its just you would need to subscribe
>> to other things too... of course if you are doing 1-2-many you
>> really should be doing association events anyway...
> 
> 
> There is another case - if PR-SCTP is enabled, an implementation can 
> receive a pertinent FORWARD TSN after a partial delivery has already 
> occurred.  Then the delivery needs to be aborted without aborting the 
> association, or PR-SCTP is hazardous to use in real life.  In short 
> PR-SCTP implies truncated messages, whenever and wherever partial 
> deliveries are allowed.
True, but all things can be hazardous to use, at least if
they have any power of flexibility.

This is another area where the app designer should think
about what they are doing. If I design a application that
is sending 1Meg PR-SCTP messages then I had best be
prepared for having pd-api's be aborted... and thus
turn on notifications...

The reality of it, however IMO, are that a pd-api
writer is NOT going to send a 1Meg message.. more
like 160 bytes messages ... and should not ever
see any message fragmentation...

R

> 
> - Mark
> 
> 
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg <at> ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg
> 
> 

--

-- 
Randall Stewart
803-345-0369 <or> 815-342-5222(cell)