The Methods & Tools Special Interest Group and the Interoperability Special Interest Group held a joint meeting from 9 pm til 10:10 pm Central European Time.
Versioning of CAPE-OPEN
U.S. Environmental Protection Agency (represented by Bill BARRETT), Bryan Research & Engineering (represented by Michael HLAVINKA), AmsterCHEM (represented by Jasper van BATEN), Michael HALLORAN (Contractor for CO-LaN), Michel PONS (Contractor for CO-LaN / Chief Technology Officer).
M. PONS stresses out the more and more urgent need to define a working strategy for versioning. M. HALLORAN had no time to investigate the technical aspects discussed at the previous meeting. M. PONS makes a quick summary of what was said about versioning at the CAPE-OPEN 2018 Annual Meeting for those not present in Ludwigshafen.He mentions that a fifth Use Case for versioning pops up following the discussions held at the CAPE-OPEN 2018 Annual Meeting on diffusion coefficients within the Thermodynamic and Physical Properties interface specification at version 1.1. While the proposal is meant to deprecate a property name and defines new property names, so with no impact on the Interface Definition Language rendering of the specification, it impacts the documentation set, either through an Errata & Clarification document or through a new version of the Thermodynamic and Physical Properties interface specification at version 1.1.
M. HLAVINKA reminds participants of the numbering scheme he proposed last time for 3rd version digits, i.e. year and day number in year.
M. HALLORAN precises then which tests have to be performed: this scheme only helps us if we can apply it to executables, we have to prove that we can version a DLL according to this scheme with TLB and PIA in it, and that we can have the Windows Installer correctly updating the DLL according to that scheme. We have to understand what it means that the CAPE-OPEN Type Library is in the DLL and we have to verify that the TLB still serve all the needs that the TLB currently serves (is it possible to refer to that TLB from a Visual Studio project?).
M. HLAVINKA states that it is common practice for them to reference a TLB from a DLL. However M. HLAVINKA fears issues with the PIA.
Jasper van BATEN advocates in favor of changing the IIDs of interfaces when going from version 1.1 to version 1.2 of the CAPE-OPEN standard even if the signature of interfaces has not changed.
The solution considered is that Thermo 1.1 belongs to CAPE-OPEN v1 just like Thermo 1.0. Makes a single TLB/PIA necessary, representing the entire v1 version. This version (v1) comes with the namespace 1.1.0. It has to be checked that the same namespace can be kept till there is no major changes to the standard requiring to move to v2. The current CATID assigned to a CAPE-OPEN component would represent the fact that the CAPE-OPEN component adheres to version 1 of the CAPE-OPEN standards. No change there so far but would need to have a different CATID for CAPE-OPEN component referring to version 2 of the CAPE-OPEN standards. That would help PMEs to figure out to which version the component adheres and the component could carry the CATID for v1 and the CATID for v2. The target is to give a namespace 2.0 for version 2 of the CAPE-OPEN standard.
A point raised is one to set a frequency of changes from one major version to another.
Next meeting is sheduled for Novembre 26, 2018.