Mike

Attached is a review copy of the bit order document with Word comments added.  I am particularly concerned about the grammar region implications that you have added. We have not discussed this before on the calls.



Regards
 
Steve Hanson
Architect,
IBM DFDL
Co-Chair,
OGF DFDL Working Group
IBM SWG, Hursley, UK

smh@uk.ibm.com
tel:+44-1962-815848




From:        Mike Beckerle <mbeckerle.dfdl@gmail.com>
To:        Steve Hanson/UK/IBM@IBMGB, "dfdl-wg@ogf.org" <dfdl-wg@ogf.org>,
Date:        25/07/2014 14:45
Subject:        Re: [DFDL-WG] bit order documents updated




I will edit and share a discussion thread of the link16 requirements I have been having with Russ Hawkins of Lockheed-Martin discussing TDL/link16. He is a SME on this topic, having built and used processors for that format.

MIL-STD-2045 works today with just the leastSignificantBitFirst feature as described. We have some tests. I will publish this DFDL schema and tests.

MIL-STD-2045 doesn't require the byte order mixed-endian behavior.

Mixed endian behavior is something we originally thought link16 required that is not illustrated within MIL-STD-2045.

Our understanding of link16 was incorrect however. It turns out link16 only requires this byte/word swapping as a bulk preprocessing pass over the entire body of a message (but not the message header). This is not unlike decoding base64 or other encoded data found in a message payload. DFDL can't do this today in one pass - an application program that drives a DFDL processor is needed for multi-pass formats like this. This is fine for DFDL v1.0. 

Russ verified that this is the *only* byte/word swapping required for link16 - there is no per-element mixed-endian byte order required. Just the same least-significant-bit-first bitOrder that MIL-STD-2045 needs. Hence, I dropped the mixed-endian byte order from the proposal.

The remaining link16 requirement that couldn't be worked around is the 6-bit ascii format.

It is always going to be the case that for some formats, especially these big complex ones, that we will not really know if we can model it until the DFDL schema has been created. However, MIL-STD-2045 plus a minor additional 6-bit encoding is our best shot at having credible support for a wide array of MIL-STD formats.


Mike Beckerle | OGF DFDL Workgroup Co-Chair | Tresys Technology | www.tresys.com
Please note: Contributions to the DFDL Workgroup's email discussions are subject to the OGF Intellectual Property Policy



On Fri, Jul 25, 2014 at 6:02 AM, Steve Hanson <smh@uk.ibm.com> wrote:
I am very concerned that we do not understand these formats properly. We need to engage an SME to verify that these proposals are enough to model MIL-STD-2045 and Link-16 (and other related formats). Until that happens IBM will not be able to close action 233.

Regards
 
Steve Hanson
Architect,
IBM DFDL
Co-Chair,
OGF DFDL Working Group
IBM SWG, Hursley, UK

smh@uk.ibm.com
tel:
+44-1962-815848



From:        
Mike Beckerle <mbeckerle.dfdl@gmail.com>
To:        
"dfdl-wg@ogf.org" <dfdl-wg@ogf.org>,
Date:        
24/07/2014 23:08
Subject:        
Re: [DFDL-WG] bit order documents updated
Sent by:        
dfdl-wg-bounces@ogf.org




Of course as soon as I post this I get the information needed to resolve several things.

The new byte-order enum is not needed, so I removed that material. The bit order document is now a fairly clean thing trying to be an errata.

Please review: draft-gwdi-bit-order-features-v3.0.docx

I created a separate working document on "Mixed Endian Byte Order" should we ever take this up again.

I also removed the TDML stuff into the TDML document. (now v2).


Mike Beckerle | OGF DFDL Workgroup Co-Chair | Tresys Technology |
www.tresys.com
Please note: Contributions to the DFDL Workgroup's email discussions are subject to the
OGF Intellectual Property Policy



On Thu, Jul 24, 2014 at 5:16 PM, Mike Beckerle <
mbeckerle.dfdl@gmail.com> wrote:

There are now several working documents having to do with bit-order. The prior document has been taken down and replaced by 3 others.

These are all up on redmine:
http://redmine.ogf.org/dmsf/dfdl-wg?folder_id=5485

1) draft-gwdi-bit-order-features-v2.05.docx - Just describes the new features + TDML extension. The biggest section is organized like an errata specifying where things change in the spec. This should get incorporated into errata, or perhaps just referenced from an errata.

Section 3 of this doc is the proposed errata and spec language for review.

There is one major open issue here, which is it is unclear whether the mixed-endian byte order that was previously proposed is actually needed or not. The swapping of 16-bit words appears to be not a per-element thing, but something done as a pre-processing of an entire message body before parsing. This isn't something we can handle in DFDL, much like base64-encoded data and so forth. This may be the only place that needed the 16-bit word swapping.

2) Understanding Bit Orderings - draft-gwdi-mil-std-2045-understanding-bit-order-v2.05.docx - Material about bit order - the Wire model vs. Number model material. This is effectively just archiving this material for posterity.

If you already read this, you don't need to read it again.

3) draft-gwdi-mil-std-2045-additional-features-v2.05.docx - Material about additional DFDL features that would be helpful in modeling MIL-STD-2045.

This can also be reviewed. Not as urgent as (1) above.

4) draft-gwdi-dfdl-standard-encodings-v03.docx - This is material that will be integrated back into the spec as an appendix, but this incorporates feedback on prior versions, and adds a 6-bit ascii encoding that is used by the same binary format standards as the 7-bit one. Works same way, just 6 bits not 7, so there's some codepoint changes.

Mike Beckerle | OGF DFDL Workgroup Co-Chair | Tresys Technology |
www.tresys.com
Please note: Contributions to the DFDL Workgroup's email discussions are subject to the
OGF Intellectual Property Policy

--
 dfdl-wg mailing list
 
dfdl-wg@ogf.org
 
https://www.ogf.org/mailman/listinfo/dfdl-wg

Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU



Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU