choiceBranchKey - suggest change to List of DFDL String Literals ?
Our first experience with requiring and implementing choiceBranchKey in Daffodil and we're already dealing with the fact that many times the same branch has multiple different keys all of which indicate that it should be selected. Right now, dfdl:choiceBranchKey is a DFDL String Literal, so values like dfdl:choiceBranchKey=" " (that's two spaces) are legal and will work. If it was to be a whitespace separated value, then those would have to use DFDL Entities to express the whitespace. I know I have seen COBOL data where there were multiple tags that denote the same structure. I am curious what others experience with choiceBranchKey is, and how this issue was handled. I believe from prior emails that IBM has already implemented this in its DFDL implementation. There is an erratum specifying that choiceBranchKey comparison is supposed to be case-sensitive (for performance reasons), but nothing heretofore about it using DFDL Entities to represent whitespace. 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 <http://www.ogf.org/About/abt_policies.php>
Hi Mike If I recall, choiceBranchKey was a single DFDL String Literal because the value to be matched against it is computed by a DFDL expression (from choiceDispatchKey), which can do things like lower casing strings or other manipulations. Are you seeing use cases where multiple keys per branch are making the expression too complex? Regards Steve Hanson IBM Hybrid Integration, Hursley, UK Architect, IBM DFDL Co-Chair, OGF DFDL Working Group smh@uk.ibm.com tel:+44-1962-815848 mob:+44-7717-378890 From: Mike Beckerle <mbeckerle.dfdl@gmail.com> To: "dfdl-wg@ogf.org" <dfdl-wg@ogf.org> Date: 12/12/2016 19:48 Subject: [DFDL-WG] choiceBranchKey - suggest change to List of DFDL String Literals ? Sent by: "dfdl-wg" <dfdl-wg-bounces@ogf.org> Our first experience with requiring and implementing choiceBranchKey in Daffodil and we're already dealing with the fact that many times the same branch has multiple different keys all of which indicate that it should be selected. Right now, dfdl:choiceBranchKey is a DFDL String Literal, so values like dfdl:choiceBranchKey=" " (that's two spaces) are legal and will work. If it was to be a whitespace separated value, then those would have to use DFDL Entities to express the whitespace. I know I have seen COBOL data where there were multiple tags that denote the same structure. I am curious what others experience with choiceBranchKey is, and how this issue was handled. I believe from prior emails that IBM has already implemented this in its DFDL implementation. There is an erratum specifying that choiceBranchKey comparison is supposed to be case-sensitive (for performance reasons), but nothing heretofore about it using DFDL Entities to represent whitespace. 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
Yes this situation is that the expressions are big, because there are 100's of branches, and so the constants are remote from where the branch is expressed. And it ends up being quite inefficient. What one wants is a table based lookup. Not an expression like if (../key eq "11 or ../key eq "22") then "11" else if (../key eq "AA" or ../key eq "BB" then "AA" ... so on for dozens of cases. and so on. That's just a linear chain again, so equivalent to using discriminators on the branches. An interesting point expressed to us is that users would like the DFDL schema to not only work to parse and unparse the file, but for it to serve as a declarative expression of the data format - as the machine-readable documentation of the format ultimately. While a big expression will work, it's not as declarative as listing the possible dispatch keys for each branch right on that branch. A big expression really fails to capture the essence of the format in a manner that enables efficient implementation. 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 <http://www.ogf.org/About/abt_policies.php> On Tue, Dec 13, 2016 at 11:01 AM, Steve Hanson <smh@uk.ibm.com> wrote:
Hi Mike
If I recall, choiceBranchKey was a single DFDL String Literal because the value to be matched against it is computed by a DFDL expression (from choiceDispatchKey), which can do things like lower casing strings or other manipulations. Are you seeing use cases where multiple keys per branch are making the expression too complex?
Regards
Steve Hanson
IBM Hybrid Integration, Hursley, UK Architect, *IBM DFDL* <http://www.ibm.com/developerworks/library/se-dfdl/index.html> Co-Chair, *OGF DFDL Working Group* <http://www.ogf.org/dfdl/> *smh@uk.ibm.com* <smh@uk.ibm.com> tel:+44-1962-815848 <+44%201962%20815848> mob:+44-7717-378890 <+44%207717%20378890>
From: Mike Beckerle <mbeckerle.dfdl@gmail.com> To: "dfdl-wg@ogf.org" <dfdl-wg@ogf.org> Date: 12/12/2016 19:48 Subject: [DFDL-WG] choiceBranchKey - suggest change to List of DFDL String Literals ? Sent by: "dfdl-wg" <dfdl-wg-bounces@ogf.org> ------------------------------
Our first experience with requiring and implementing choiceBranchKey in Daffodil and we're already dealing with the fact that many times the same branch has multiple different keys all of which indicate that it should be selected.
Right now, dfdl:choiceBranchKey is a DFDL String Literal, so values like dfdl:choiceBranchKey=" " (that's two spaces) are legal and will work. If it was to be a whitespace separated value, then those would have to use DFDL Entities to express the whitespace.
I know I have seen COBOL data where there were multiple tags that denote the same structure.
I am curious what others experience with choiceBranchKey is, and how this issue was handled.
I believe from prior emails that IBM has already implemented this in its DFDL implementation.
There is an erratum specifying that choiceBranchKey comparison is supposed to be case-sensitive (for performance reasons), but nothing heretofore about it using DFDL Entities to represent whitespace.
Mike Beckerle | OGF DFDL Workgroup Co-Chair | Tresys Technology | *www.tresys.com* <http://www.tresys.com/> Please note: Contributions to the DFDL Workgroup's email discussions are subject to the *OGF Intellectual Property Policy* <http://www.ogf.org/About/abt_policies.php> -- 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
Yes it gets ugly and inefficient when you start having a large 'if' statement. An example of an IBM-schema with direct dispatch choice is SWIFT, which has dozens of messages but they all fall into 7 patterns so the expression 'if' statement has 7 branches. Not seen anything more complex than that so far. A List of DFDL String Literals is a possibility. Regards Steve Hanson IBM Hybrid Integration, Hursley, UK Architect, IBM DFDL Co-Chair, OGF DFDL Working Group smh@uk.ibm.com tel:+44-1962-815848 mob:+44-7717-378890 From: Mike Beckerle <mbeckerle.dfdl@gmail.com> To: Steve Hanson/UK/IBM@IBMGB Cc: "dfdl-wg@ogf.org" <dfdl-wg@ogf.org> Date: 13/12/2016 17:34 Subject: Re: [DFDL-WG] choiceBranchKey - suggest change to List of DFDL String Literals ? Yes this situation is that the expressions are big, because there are 100's of branches, and so the constants are remote from where the branch is expressed. And it ends up being quite inefficient. What one wants is a table based lookup. Not an expression like if (../key eq "11 or ../key eq "22") then "11" else if (../key eq "AA" or ../key eq "BB" then "AA" ... so on for dozens of cases. and so on. That's just a linear chain again, so equivalent to using discriminators on the branches. An interesting point expressed to us is that users would like the DFDL schema to not only work to parse and unparse the file, but for it to serve as a declarative expression of the data format - as the machine-readable documentation of the format ultimately. While a big expression will work, it's not as declarative as listing the possible dispatch keys for each branch right on that branch. A big expression really fails to capture the essence of the format in a manner that enables efficient implementation. 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 Tue, Dec 13, 2016 at 11:01 AM, Steve Hanson <smh@uk.ibm.com> wrote: Hi Mike If I recall, choiceBranchKey was a single DFDL String Literal because the value to be matched against it is computed by a DFDL expression (from choiceDispatchKey), which can do things like lower casing strings or other manipulations. Are you seeing use cases where multiple keys per branch are making the expression too complex? Regards Steve Hanson IBM Hybrid Integration, Hursley, UK Architect, IBM DFDL Co-Chair, OGF DFDL Working Group smh@uk.ibm.com tel:+44-1962-815848 mob:+44-7717-378890 From: Mike Beckerle <mbeckerle.dfdl@gmail.com> To: "dfdl-wg@ogf.org" <dfdl-wg@ogf.org> Date: 12/12/2016 19:48 Subject: [DFDL-WG] choiceBranchKey - suggest change to List of DFDL String Literals ? Sent by: "dfdl-wg" <dfdl-wg-bounces@ogf.org> Our first experience with requiring and implementing choiceBranchKey in Daffodil and we're already dealing with the fact that many times the same branch has multiple different keys all of which indicate that it should be selected. Right now, dfdl:choiceBranchKey is a DFDL String Literal, so values like dfdl:choiceBranchKey=" " (that's two spaces) are legal and will work. If it was to be a whitespace separated value, then those would have to use DFDL Entities to express the whitespace. I know I have seen COBOL data where there were multiple tags that denote the same structure. I am curious what others experience with choiceBranchKey is, and how this issue was handled. I believe from prior emails that IBM has already implemented this in its DFDL implementation. There is an erratum specifying that choiceBranchKey comparison is supposed to be case-sensitive (for performance reasons), but nothing heretofore about it using DFDL Entities to represent whitespace. 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
participants (2)
-
Mike Beckerle -
Steve Hanson