Gazelle HL7v2 validation and NULL values

22 views
Skip to first unread message

Dmytro Rud

unread,
Mar 16, 2015, 8:00:46 AM3/16/15
to ipf...@googlegroups.com
Hi,

Do I understand that HL7v2 conformance profiles -- and, correspondingly, the Gazelle validator based on them -- cannot handle NULL values (pairs of double quotes "") in composite fields?

NB: support for such NULL values shall be actually allowed only at special places like OBR-2, ORC-2, INV-1, OBX-5, etc.

Example:

Expected behavior: 

The following segments shall pass the validation:
INV|substanceId^^HL70451|OK^^HL70383
INV|""|OK^^HL70383

The following segment shall not pass the validation, because NULL cannot belong to any coding system:
INV|""^^HL70451|OK^^HL70383

Observed behavior:

The following segments pass the validation:
INV|substanceId^^HL70451|OK^^HL70383
INV|""^^HL70451|OK^^HL70383 (false positive)

The following segment does not pass the validation, because INV-1-3 "Coding System ID" is missing:
INV|""|OK^^HL70383

Proposal:

The XML Schema of conformance profiles (http://gazelle.ihe.net/xsd/HL7MessageProfileSchema.xsd) shall be locally extended -- the boolean attribute
/HL7v2xConformanceProfile/HL7v2xStaticDef/Segment/Field/@nullAllowed
shall be introduced and respected by the IPF-Gazelle validator: when set to "true", NULL will be accepted instead of whatever was originally defined for the field.  This attribute will never occur in "official" conformance profiles published by IHE and HL7, thus this local extension will not break anything.

Your opinions?

Thanks and bets regards
Dmytro

Dmytro Rud

unread,
Mar 19, 2015, 5:05:27 PM3/19/15
to ipf...@googlegroups.com
If nobody objects, I will implement my proposal on this weekend.

Best regards
Dmytro

Christian Ohr

unread,
Mar 23, 2015, 5:13:53 AM3/23/15
to ipf...@googlegroups.com
Sorry, I missed this thread completely.
Yes, if we remain backwards-compatible, I have no objections.
Note that we have tagged IPF 3.0 with IPF-Gazelle 1.0.6, but we can release updates to the gazelle lib as we wish.

Anyway you should document the extension - for now, the README.md is probably sufficient.

Christian

Dmytro Rud

unread,
Mar 23, 2015, 5:49:33 AM3/23/15
to ipf...@googlegroups.com
Thanks.  Done.  Shall I merge the changes from 1.1 to 1.0?

Best regards
Dmytro


--
You received this message because you are subscribed to the Google Groups "ipf-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ipf-dev+u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Dmytro Rud

unread,
Mar 23, 2015, 8:40:57 AM3/23/15
to ipf...@googlegroups.com
It seems that the next change will be the specification of condition predicates for conditional elements...  But I have neither clear timelines nor implementation ideas yet...
Reply all
Reply to author
Forward
0 new messages