[wsf-dev] 160 bit message id

[Greg Whitehead]

To comply with the IRI requirement you can format your uuid ids as urns,
e.g.
    urn:uuid:1eda95af-63e5-4048-8098-80ba89f7994c

-Greg

On 4/30/07 8:27 PM, "Greg Whitehead" <greg.whitehead at hp.com> wrote:

> I would suggest that you use the random UUID format and the cryptographic
> PRNG.
> 
> -Greg
> 
> 
> On 4/30/07 8:25 PM, "Asa Hardcastle" <asa.openliberty at zenn.net> wrote:
> 
>> Unfortunately, java can't get to the MAC address.  
>> 
>> According to wikipedia (which I realize can't be trusted) on random 128 bit
>> UUIDs:
>> 
>> "The number of theoretically possible UUIDs is therefore 2(128) = 256(16) or
>> about 3.4 × 10(38).  This means that 1 trillion UUIDs have to be created
>> every nanosecond for 10 billion years to exhaust the number of UUIDs."
>> 
>> My understanding from the specification:
>> 
>> * must ensure a negligible risk of collision (as Conor has said)
>> * if pseudorandom, must use at least 160 random bits
>> * according to WSAv1,  wsa:MessageId must be an IRI
>> - like "http://example.com/someuniquestring"
>> 
>> Which leaves me with: "http://someurl.com/160bitsofrandomfun" - which should
>> be extremely unlikely to collide - seems like the  "http://someurl.com/" is
>> somewhat unnecessary - and potentially a privacy concern.
>> 
>> asa
>> 
>> 
>> 
>> 
>>  
>> 
>> --
>> Asa Hardcastle, Technical Lead, openLiberty
>> Tel: +1.413.429.1044 Skype: subsystem7
>> 
>>  
>> 
>> On Apr 30, 2007, at 4:51 PM, Cahill, Conor P wrote:
>> 
>>>   
>>> There's an interesting history in the discussion of the message ID value. 
>>>  
>>>  
>>>  
>>> The current requirement (lines 557-559 of the SOAP Bindings specification)
>>> is:
>>>  
>>>>   
>>>>   
>>>> 
>>>> Any party that assigns a value to a <wsa:MessageID> header block MUST
>>>> ensure that   there is negligible probability that   that party or any
>>>> other party will accidentally assign the same   identifier to any other
>>>> message
>>>  
>>> This is not a 160 or 128 bit requirement.
>>>  
>>>  
>>>  
>>> I would presume that a 128 bit UUID that included the 48 bits of mac address
>>> of the issuing host would meet that requirement.
>>>  
>>>  
>>>  
>>> Conor
>>> 
>>>  
>>>>   
>>>>   
>>>> 
>>>>   From: wsf-dev-bounces at openliberty.org
>>>> [mailto:wsf-dev-bounces at openliberty.org] On Behalf Of Asa   Hardcastle
>>>> Sent: Monday, April 30, 2007 4:36 PM
>>>> To:   wsf-dev at openliberty.org
>>>> Subject: [wsf-dev] 160 bit message   id
>>>> 
>>>>   
>>>> Hi All,  
>>>> 
>>>>   
>>>> Does anyone know the origin of the 160 bit message id requirement in the
>>>> <wsa:MessageID> header?  Before I read that line I was thinking of   using
>>>> a 128 bit UUID.
>>>>   
>>>> 
>>>>   
>>>> thanks,
>>>>   
>>>> 
>>>>   
>>>> asa
>>>> 
>>>> 
>>>>   
>>>> 
>>>>   
>>>> 
>>>>   
>>>> --
>>>>   
>>>> Asa Hardcastle, Technical Lead, openLiberty
>>>>   
>>>> Tel: +1.413.429.1044 Skype: subsystem7
>>>> 
>>>> 
>> 
>> 
>> 
>> _______________________________________________
>> wsf-dev mailing list
>> wsf-dev at openliberty.org
>> http://lists.openliberty.org/mailman/listinfo/wsf-dev
> 
> 
> 
> _______________________________________________
> wsf-dev mailing list
> wsf-dev at openliberty.org
> http://lists.openliberty.org/mailman/listinfo/wsf-dev


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.openliberty.org/pipermail/wsf-dev/attachments/20070430/675f3763/attachment.html