Re: question on get the whole FE object LFB?

Yes,I see.The protocol draft has the saying that when using FULLDATA TLV 
"The variable-length field requires the addition of a FULLDATA TLV within 
the outer FULLDATA TLV as in the case of element b above." in use case 
section example 3.Then just as what you have said when all optional field 
are in and the array are dense,we can use FULLDATA TLV encoding,treating 
the variable element using a nesting FULLDATA TLV,in other 
circumustances,SPARSEDATA TLV should be used.
Still,there is a question,when CE requesting a whole LFB,should the 
capabilities also be returned back just like attributes or just the 
attributes,the draft doesn't say this?
Yours,Fenggen


>From: "Joel M. Halpern" <[email protected]>
>To: "Jia Fenggen" <[email protected]>
>CC: [email protected]
>Subject: Re: question on get the whole FE object LFB?
>Date: Thu, 21 Sep 2006 23:50:35 -0400
>
>The CE simply asks for the table1 (or the whole LFB, or whatever.)
>If the FE sends back a FULLDATA TLV encoding then the CE knows that 
>all optional fields are present and all tables are dense.  The 
>FULLDATA encoding does include the table length.
>If the array is sparse, or there are optional fields missing (for 
>example if t2 is optional, and not present in all elements, then the 
>FE MUST use the SPARSEDATA encoding.  And the sparsedata encoding 
>includes the indexes / ids, so the CE can properly parse the whole 
>thing.
>
>Yours,
>Joel
>
>At 11:39 PM 9/21/2006, Jia Fenggen wrote:
>>I am afraid you have probably misunderstand what i am complaining 
>>about,for example,the LFB  foo have the following attributes 
>>defined:
>>   foo1, type u32, ID = 1
>>   table1: type array, ID = 2
>>           elements are:
>>           t1, type u32, ID = 1
>>           t2, type u32, ID = 2
>>   foo2,type u32,ID=3
>>   table2: type array, ID = 4
>>           elements are:
>>           j1, type u32, ID = 1
>>           j2, type u32, ID = 2
>>
>>table1 and table are variable arrays,let's take that in the 
>>instance of LFB foo,table1 has 3 rows and table 2 has 4 rows.if we 
>>encoding this LFB in FULLDATA,then because CE has no knowledge of 
>>the current rows of table1 and table2,then it can't decide how to 
>>get foo2 value,what do you think?
>>Yours,Fenggen
>>
>>>From: "Joel M. Halpern" <[email protected]>
>>>Reply-To: "Joel M. Halpern" <[email protected]>
>>>To: [email protected]
>>>Subject: Re: question on get the whole FE object LFB?
>>>Date: Thu, 21 Sep 2006 23:27:12 -0400
>>>
>>>I believe you are partially correct.
>>>That is, as you observe, a FULLDATA can only be used if all 
>>>optional data elements are present.
>>>This means that if all such elements are present, then a FULLDATA 
>>>can be used for the response.  It is up to the FE to check this 
>>>condition.
>>>If not all optional elements are present, then the SPARSEDATA 
>>>encoding is used.  That is what the SPARSEDATA encoding is for.
>>>This avoids having to introduce multiple Path-data-TLVs in 
>>>response to a single Path-data-TLV in a request.  It thereby 
>>>preserves the semantic information that what is being returned is 
>>>the whole LFB, while indicating which fields are present or 
>>>absent.
>>>
>>>So, the example is legal.  If we have a chance to update this, we 
>>>should probably note that it is only legitimate if all optional 
>>>fields in the FE Object are present.
>>>
>>>Yours,
>>>Joel M. Halpern
>>>
>>>At 11:06 PM 9/21/2006, Jia Fenggen wrote:
>>>>In draft-forces-portocol-8.txt Appendix D use case section,the 19 
>>>>use case is about how to get a whole LFB,it uses FE object LFB as 
>>>>a example: Get a whole LFB (all its attributes, etc.).
>>>>
>>>>For example:  at startup a CE might well want the entire FE
>>>>           OBJECT LFB.  So, in a request targeted at class 1, 
>>>>instance
>>>>           1, one might find:
>>>>
>>>>
>>>>
>>>>   operation = GET-TLV
>>>>           Path-data-TLV
>>>>                   flags = 0  IDCount = 0
>>>>
>>>>   result:
>>>>   operation = GET-RESPONSE-TLV
>>>>           Path-data-TLV
>>>>                   flags = 0  IDCount = 0
>>>>                   FULLDATA encoding of the FE Object LFB
>>>>here it uses FULLDATA to encoding the FE Object LFB,I think it's 
>>>>probably not correct,in FE Object LFB we have several variable 
>>>>arrays and when encoding in FULLDATA,the CE may not figure out 
>>>>what is the ending of a array,I think we could use flat Path data 
>>>>encoding or Nested Path data encoding to solve this like follows:
>>>
>>>... example details delete for length.
>>
>>_________________________________________________________________
>>璐逛杞 MSN Explorer:   http://explorer.msn.com/lccn/
>>
>

_________________________________________________________________
涓虹杩琛浜ゆ锛璇蜂娇 MSN Messenger:  http://messenger.msn.com/cn