[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
[qvtd-dev] Epsilon builds
|
Hi Horacio
I've just taken another shot at the Epsilon build available at
https://hudson.eclipse.org/ocl/job/qvtd-branch-tests/ws/MMT-QVTd.downloads/mmt-qvtd-incubation-Update-201503100711.zip
It should be complete: it includes antlr 3.1.b1 (Piggy back CQ 9369 raised.)
First attempt failed to install on M5 since the Hudson build gives
strict lower bounds for vague dependencies. (core.runtime has
incremented since M5).
While checking other releng niceties
- I've eliminated all use of ANT as you said it wasn't needed
- I've eliminated the org.eclipse.ui activators that aren't needed
- I've defined non QVTd version dependencies
- I've added about.html
- I've refined plugin.properties
- I've refined build.properties
Then I started checking schema, expecting to have to add comments etc,
but the schema folder is not for extension point definition, rather
redistribution of yWorks schema. If we do this we need a CQ. I don't
think we need yWorks for Mars. Post Mars we may have some debugging
tools that might exploit it so we could raise a CQ then. For now I can
just omit the schema from the build.properties so they are not
distributed and the extras graph scripts will be broken. Once
org.eclipse.qvtd.build.etl renames to org.eclipse.qvtd.compiler,
org.eclipse.qvtd.build.etl.extras can stay behind pending
re-implementation as part of org.eclipse.qvtd.debug.ui.
Until build installable, this is available as commits immediately
following master on ewillink/460516
I have also included the 1 QvtrToQvtcTests and the 1 passing
QVTdMtcTests in all tests.
Next - generalization of the QVTi launch config for QVTc / QVTr.
In due course the debugger needs access to the trace / middle models so
that it can do QVTr / QVTc rather than QVTi source debugging.
Regards
Ed Willink