You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@daffodil.apache.org by Mike Beckerle <mb...@tresys.com> on 2018/09/17 20:56:56 UTC

schemas and branches of schema repos

So there are all these github DFDL schemas.


Testing against them is becoming too painful.


Right now they have a master branch pointing at daffodil 2.1.0, and a for-daffodil-2.2.0 branch which is obsolete and pointing at 2.2.0-SNAPSHOT daffodil, which presumably doesn't exist anymore, and never will again.


My suggestion to revise these conventions:


master branch - IBM's branch. Files as they left them.


daffodil-master branch - Runs daffodil most recent release (update with each release) - files organized according to standard project layout. Only exists if we need to diverge from IBM's master branch file-structure, or need to be temporarily incompatible with IBM DFDL.


daffodil-dev branch - Runs daffodil current development snapshot release (updated to next dev branch each release) - That is, currently 2.3.0 is the branch for development. This branch of a schema should be maintained to work with what is pushed to 2.3.0 branch. (The build.sbt will say 2.3.0-SNAPSHOT.)


Once we have a cross-testing rig that drives either or both daffodil and IBM dfdl, then the reason for daffodil-master distinct from just master for schemas IBM created, can go away, as a single file organization will work, but until then we have potential need for master, daffodil-master, and daffodil-dev


Comments?