Only build Core+Tests in the macOS CI build

[msmk0]

This (temporarily) disables building the examples and the DD4hep- and TGeo plugin. This sidesteps the build issue with the additional HEP-specific dependencies (ROOT, Geant4, DD4hep, ...).

[codecov]

Codecov Report

Merging #141 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #141   +/-   ##
=======================================
  Coverage   45.79%   45.79%           
=======================================
  Files         349      349           
  Lines       17395    17395           
  Branches     8071     8071           
=======================================
  Hits         7966     7966           
  Misses       4332     4332           
  Partials     5097     5097           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 3adfa46...e425bb0. Read the comment docs.

[paulgessinger]

I’ve got the dependency build running. Should be able to reenable this right away

[msmk0]

Ok. Can you create a PR for the necessary changes. Then I will close this one.

[paulgessinger]

The CI job in the other repository now builds all dependencies. I still need to test if our code builds on top of that (which I expect should work), and then switch over the tarball URL in the macOS job here. I'll do that as soon as I have it.

[paulgessinger]

Ok so it seems not to be possible to convince DD4hep's invocation of find_package(XercesC) to not fail if XercesC is not at the exact same location as when DD4hep was built. Building XercesC explicitly doesn't seem to help, and since brew also changes install locations if the version changes, it will break every time there's a new version unless we rebuild the dependencies.

DD4hep doesn't seem to have any issues finding root, even if the location is different as when DD4hep was built, so this points to being a 'problem' with FindXercesC.cmake, but I'm not sure.

Honestly, I don't know what to do here.

I can still make the deps CI deploy up-to-date versions of DD4hep, G4, root and so on, and the CI in this repo can pick that up, but that's not at all a synchronized process here, and could be fragile.

[msmk0]

I would suggest we merge this change in, then. That way we compile at least the core on macOS for now. We can discuss options next week without having a broken build.

[paulgessinger]

TGeo should be fine though?

[msmk0]

That requires ROOT which is not available via homebrew, correct?

[paulgessinger]

But it is available from the tarball that the CI currently downloads.

[msmk0]

Won't we have the same problem that we have with Geant4 now, i.e. ROOT depends on a library from homebrew (expat or something similar) with a version-specific path that will break in the future?

[asalzburger]

Hi, I agree with the strategy here - if we have a macOS build for the Core, that should be fine.

DD4hep is a pain in the way it handles dependencies, I have it working on my Mac locally and will NEVER touch it again.

Let's quickly discuss in the briefing tomorrow, but I would tend to approve and merge as such then.