When the various profile teams (e.g. DAF, QICORE, HSPC, etc.) are creating profiles, valuesets, examples, etc. the FHIR core publish tool build fails completely for validiation failures. Validation failures aborting the build is acceptable when the FHIR core publisher is running on travis-ci.org and being used to publish the project to the public continuous integration web site.

However, for the sub-teams building the FHIR spec locally and developing profile artifacts, one team accidentally breaking the build, breaks it for all even locally. Given a 10-20 minute run to publish locally causes many hours of downtime for many people across the sub-teams.

So for example, if a valueset or example for HSPC fails validation then members of the QICORE sub-team can still build and publish the profiles, etc. locally. One sub-group shouldn't cause stoppage for other sub-groups unless the failure is in a domain/core resource.

This situation happened again yesterday (15-Apr-2015) with fhir-svn build failure # 2606 also subsequently in # 2607 and 2608. Took over a day later to get the build back to passing for everyone.

Must have configuration parameters to ignore validation errors for particular IG sets and/or change errors to warnings for particular contexts to allow the full publish run to complete even with errors.