20120910 Q4 Tooling WGM Minutes
Jump to navigation Jump to search
Tooling Meeting Template
|HL7 Tooling Meeting Agenda/Minutes
Location: Baltimore Conference Center Room 345
Time: 3:30 pm EDT
|Facilitator: Jane Curry||Note taker(s): Lynn Laakso|
|x||Jane Curry, Co-Chair|
|x||Peter Hendler, RIMBAA Co-Chair|
|x||Lynn Laakso, HL7 HQ Tooling Support|
|x||Massimo Frossi, Insiel Mercato|
|x||Peter Hendler, KP|
|x||Andrew Liu, CHI|
|x||Gordon Raup, Datuit|
|x||Jim Klein, Advisory Board|
|x||Brian Pech, KP|
|x||Michael Rossman, KP|
|x||Francesco Rossi, TBS Group|
|x||Amnon Shabo, IBM RIMBAA Co-Chair|
|x||Michael van der Zel, HL7 PC and RIMBAA liaison|
|Quorum Requirements Met (co-chair plus 3 counting staff): yes|
Tooling hosting RIMBAA
- Call to order 3:37 PM, by Jane
- Jane gives background on historical HL7 Tooling and evolution into new Tooling mandate beyond publication into RIMBAA realm of Software Implementation. We wish to find a way to more effectively acquire, develop, modify, support, etc our tools.
- Jane reviews a draft HL7 Tooling Strategy document.
- Risk assessments typically are those of not meeting users' needs. What types of risks should we take into account, especially those having downstream implications. Tools historically volunteer built now find themselves becoming obsolete without the developer's continued attention.
- Peter finds that folks are holding off on V3 tools once FHIR has come forward. Jane suggests we discuss tooling requirements and needs in general. Need tooling process for acceptance and absorption or demonstration of COTS tooling meeting needs.
- Michael does not find COTS tools a risk, but need to leverage them for us, industry wide. Jane agrees that it is a principle within the strategy to make suggestions to influence COTS tools to work for us, rather than build our own. This is the premise behind the Tooling Challenge. MAX project will also offer one benefit to move towards externalizing EA models and ability to import them back again.
- Michael Rossman asks a question on V3 Datatypes and RIM core classes as UML profiles. Jane notes that we have not updated those since UML 1.0 and its rudimentary available technology. We still have multiple tools that have to take outputs from other tools. He adds that there is overlap between the EA and Eclipse based tooling needs. Current tooling challenge is focused on EA because Sparx sponsored it but interop with other platforms gains you brownie points. Packaging modularization and interop is very important, Michael asserts. Jane recounts that the challenge is to create firstly a valid UML profile that will be validated by import into EA. Other UML platforms can use this to identify how their products can better interoperate with our models.
- Jane adds that we need to inject some energy into the OMG UML project to ensure that identified changes can be made to the UML standard.
- We also need resources for better training on use of tools up to and including vendor tools adopted by HL7 notes Jane.
- Our vertical market, with complex modeling requirements, and little money devoted to standards modeling development, is not attractive to commercial vendors. Michael adds that complex modeling needs exist in other industries and adding these features can help their competitiveness as well.
- RIMBAA as the Tooling window into the implementation space deal with users who deal with complex stuff. We need to understand what risks we can manage via "process", in stability and interoperability of our toolsets.
- Jane refers to Grahame's draft tooling strategy from the listserv.
- Peter asks about using RDF instead of UML for model comparison and computable semantics. Semantic web technologies are found to be more powerful than UML by some. Jane notes that it doesn't have to be an either-or, that you can convert to RDF triplets and adding to the translation.
- Jane asks what are the problem spaces that HL7 tools are trying to mitigate? Use of semantic types of tools may not fully satisfy the interaction-transaction models in messaging. She cites the importance of understanding the category of problems that the implementers' software is trying to solve. Peter notes they can make a list but hard to list an order for them for the tools we have right now.
- Policy exchange in security architectures is a new area and their workflow as an example of dynamic semantics, as compared to static semantics which have been of interest to implementers in the past.
- Michael comments that as a commercial software company their strategic position is on the provision of the 20%, not on the 80% as is focused on by FHIR.
- Jane asks again on the kinds of problems that implementers are seeking tools to solve, and what risks do tooling projects need to evaluate for accepting them for development. Expectation of cross reference of semantics and the functions in each space, as an architectural use of tools.
- Jane asks about the analytics. Amnon notes that they often work with machine analytics, seeking simplest possible patterns. Peter notes that analyzing other than database queries but SNOMED structures for examples, relies more on whether they can trust the field being analyzed will always be in SNOMED representation, and vocabulary handling. Handling assumptions on ontology i.e. SNOMED, and mapping vocabulary services are not our typical types of tools. Other tools for validating a pattern similar to Clinical Pattern or other SMIRFs (small isolated related fixed pattern). Pre-coordination and post-coordination are also functions needing to be addressed in the implementation space. Example of running post coordination through a reasoner to see what satisfied it created discussion on whether it would be an HL7 tool. Jane notes that there are areas in the problem spaces that are not necessarily HL7 but touch on HL7. In these cases we may wish to promote visibility to such tools that are not necessarily directly in the HL7 space.
- Peter sums it up that we're always interested in MIF to/from "other". Jane notes that for EHR-S FM profiling that MIF structures may be changing further. Requirements analysis is a new area of interest.
- Peter asks about tools that translate from non-HL7 representations of things to HL7 representations. Amnon jokes that it sounds like V2 to V3. Jane notes that conformance testing is not something that has been embraced by the Board. Michael adds that NIST has a CCD validation on their web site but it's not practical for the real world and their validation is not comprehensive. Determination of the quality of conformance, degree of conformance is at stake. Michael van der Zel (MvdZ) notes that CGIT is working on that, but others note that they focus more on V2. Michael Rossman (MR) notes that Java would not have gone far without Sun providing reference implementations. Jane notes that bridging gaps between requirements and design, explicit statements of fitness for purpose is a higher priority. Example instance generator is of interest both up- and down-stream.
- MvdZ asks if tutorials are tools. Jane says no, but we need tutorials about our tools, both those maintained and supported, as well as promoted, by HL7. MvdZ notes that Microsoft has labs where you can play with an API.
- Jane will also reserve a quarter for Phoenix in January, for Tooling to host and request the room.
- Michael van der Zel spells his name for people to reference as a write in candidate.
- Adjourned 4:59 PM EDT
|Next Meeting/Preliminary Agenda Items|
© 2012 Health Level Seven® International. All rights reserved.