Restore compatibility with EDM4hep 0.10

[jmcarcell]

BEGINRELEASENOTES

• Revert some of the changes of Fix a few compiler warnings #1375 to build a vector of floats instead of doubles since the necessary constructor doesn't exist in 0.10.X.
• Use edm4hep::labels::CellIDEncoding only when they are available (starting at version 0.99)

ENDRELEASENOTES

See #1333 (comment)

[github-actions]

Test Results

   16 files     16 suites   5h 56m 5s ⏱️
  369 tests   366 ✅ 0 💤 3 ❌
2 904 runs  2 895 ✅ 0 💤 9 ❌

For more details on these failures, see this check.

Results for commit 1ef0a0b.

♻️ This comment has been updated with latest results.

[tmadlener]

Given the discussion in #1379. How hard would it be to setup a workflow that builds an older version of EDM4hep on the fly to make sure things still actually work?

[andresailer]

related to the deprecation of a constructor

Can you add why these are reverted please!

[andresailer]

Given the discussion in #1379. How hard would it be to setup a workflow that builds an older version of EDM4hep on the fly to make sure things still actually work?

Which old version of EDM4hep should we test against? Maybe it exists in one of the LCG releases.

[tmadlener]

Which old version of EDM4hep should we test against?

Letting @wdconinc confirm, but I think 0.10.5 is what we should be aiming for.

[wdconinc]

Which old version of EDM4hep should we test against?

Letting @wdconinc confirm, but I think 0.10.5 is what we should be aiming for.

We are using 0.10.5 in EIC (because we cannot upgrade to acts-38 yet, which is the only version that supports 0.99). However, it should probably be 0.10, and not rely on a patch release.

[andresailer]

@jmcarcell Could you cherry-pick #1381 please!

[tmadlener]

Can we also put a the 0.10.5 (or whatever we settle on for CI) into the find_package(EDM4HEP) call in the appropriate CMakeLists.txt?

[wdconinc]

Can we also put a the 0.10.5 (or whatever we settle on for CI) into the find_package(EDM4HEP) call in the appropriate CMakeLists.txt?

Or, in the interest of forward compatibility, the same structure that's used for podio to allow both 0.99 and 1.0?

[tmadlener]

It looks like DDDigi is also built with LCG_102 and LCG_101, but that doesn't yet have a recent enough version of EDM4hep.

[jmcarcell]

This should be ready now.

Can we also put a the 0.10.5 (or whatever we settle on for CI) into the find_package(EDM4HEP) call in the appropriate CMakeLists.txt?

Or, in the interest of forward compatibility, the same structure that's used for podio to allow both 0.99 and 1.0?

I do not mind this, but maybe we can do it once we get closer to EDM4hep 1.0 since I'm not sure when it's going to happen, and probably more changes will be needed in DD4hep by then?

[tmadlener]

probably more changes will be needed in DD4hep by then?

I think there will be potentially some more changes, similar to #1374 for the MCParticle. And there is still #1373. So at least from my point of view, I think we have still some PRs ahead of us where we can add the compatibility with 1.0.

[andresailer]

Something in ROOT master broke

   	 62 - t_ClientTests_Check_Shape_PseudoTrap (Subprocess aborted)
  	 63 - t_ClientTests_Check_Shape_PseudoTrap1 (Subprocess aborted)
  	 64 - t_ClientTests_Check_Shape_PseudoTrap2 (Subprocess aborted)

[andresailer]

Since test failures are unrelated

[tmadlener]

Minor nitpick from my side. We can also leave it in, it doesn't really hurt at this point.

[andresailer]

I would like to merge this after it is done on LCG_104, so that I can test the reader on LCG_104 as well.