Re: [kmip] Diagram showing difference between 32-bit and 64-bit binaryalignment proposals

Matt,

A 32-bit alignment constraint means
that TTL (Tag/Type/Length) can be 8 or 12 bytes, and Value a multiple of
4 bytes.

A 64-bit alignment constraint means
that TTL is 8 bytes (or even a multiple thereof), and Value a multiple
of 8 bytes.

Hence, the decisions on the TTL and
Value do not seem to be completely independent.

I'd suggest that: 

1) we decide first which alignment to
follow first: 32-bit or 64-bit, and 

2) we decide next the size of the TTL,
in accordance with the alignment, as explained above.

Regards,

-Robert

 wrote on 05/06/2009 11:29:47
PM:

> Hi Scott,

> 

> I was trying to reduce permutations, but maybe it makes sense to 

> have the following breakdown of independent decisions tomorrow:

> 

> Shall the Tag/Type/Length fields occupy 8 bytes (as described in the

> 64-bit proposal) or 12 bytes (as described in the 32-bit proposal)?

> Shall the Value field be aligned to 4 byte or
8 byte boundaries?

> 

> I suspect the preference may be to use 8 bytes for the Tag/Type/

> Length fields, and align the Value field to 4 byte boundaries -- 

> essentially a hybrid of the 32-bit and 64-bit proposals.  Do
folks agree?

> 

> Thanks!

> -Matt

> 

> Scott Kipp wrote: 

> Matt, 

> Thanks for putting this together.  It is
clearer when you have 

> diagrams like this.  It seems like we ought to reserve bits/octets

> if we aren't going to use them in a field.  It's kind of odd
that 

> the 32-bit proposal header takes up 3 words and the 64 bit proposal

> header takes up 2 words.  The TTLV format should be a different

> discussion than the bit alignment.  I prefer the TTLV format
of the 

> 64 bit alignment proposal, but either will do. 

> One thing your diagram doesn't show is that padding
in the 32-bit 

> proposal is either 1 to 3 bytes while the padding in the 64 bit 

> proposal is 1 to 7 bytes to get the alignment.  Your examples
in the

> other proposals clearly show this. 

> Thanks for clearing this up,

> Scott 

> 

> From:  [mailto:]

> Sent: Wednesday, May 06, 2009 3:41 PM

> To: 

> Subject: [kmip] Diagram showing difference between
32-bit and 64-bit

> binary alignment proposals

> 

> Hi Folks,

> 

> I've put together a diagram that shows the difference between the

> 32-bit and 64-bit proposals in graphic detail.  For these types
of 

> comparisons, words don't quite express the idea as well as pictures
do.

> 

> As the poll showed, the group is about halfway split between 32-bit

> and 64-bit, and I'm hoping that this diagram will show that the 64-

> bit proposal works just as well for 32-bit applications.  
:)

> 

> With any luck we'll reach a conclusion via e-mail and will be able

> to have a quick vote in tomorrow's meeting to pick one of these 

> proposals.  I do prefer the 64-bit option as more useful, but
would 

> be happy to approve the 32-bit version if that's the group's favorite.

> 

> Let me know if you have any comments or questions!

> 

> Cheers,

> -Matt

> 

> [attachment "matthew_ball.vcf" deleted by Robert Haas/Zurich/IBM]

> ---------------------------------------------------------------------

> To unsubscribe from this mail list, you must leave the OASIS TC that

> generates this mail.  Follow this link to all your TCs in OASIS
at:

> https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php