Need to configure RoIBResult BS encoder correctly.

Delete the obsolete `Reconstruction/RecExample/RecExCommission` package.
Also delete files in other packages that are related and obsolete. The
`rec.Commissioning` flag remains as that seems to be set to `True` in
various places. Even for those cases, there should be no change in job
configuration.

Move only remaining `MinimalCommissioningSetup.py` to RecExCond package.

Frank Berghaus authored 4 years ago and Frank Winklmeier committed 4 years ago

There should be no need to expose the header of the FileMetaDataTool to
other packages, it can be treated as a component and called through its
interface. Remove some unneeded headers. Enable checking for thread
safety.

TAG making ("doWriteTAG") is no longer supported and this package only
contains unused algorithms. Remove references from other packages to
this package or delete them entirely if obsolete.

The `EventTagGlobal.AttributeList` variable is still required in
RecExCommon. But in many other places we anyhow just use the hard-coded
value `"SimpleTag"`. So do the same here to avoid this trivial
dependency.

Missing import of TriggerFlags.

Hello,

This MR is hopefully the last one in the "adding tau cells to xAOD" series :-)
We now save the cells of photon shot clusters, which are built from a handful of EM1 cells.
There is almost no xAOD size increase due to shot cells being added to the AllCalo container, because these cells are already part of other EM clusters (tau pi0s) for a large part.
The only extra disk footprint comes from the small container of cluster-to-cell links (0.3 kb/event seen in standalone reconstruction of MC ESD).

The only change to the xAOD output is, as expected:
```
Py:diff-root      WARNING the following variables exist only in the new file !
Py:diff-root      WARNING  - [TauShotClustersAuxDyn.CellLink.m_persKey]
Py:diff-root      WARNING  - [TauShotClustersAuxDyn.CellLink]
Py:diff-root      WARNING  - [TauShotClustersAuxDyn.CellLink.m_persIndex]
Py:diff-root      WARNING  - [TauShotClusters_links]
```

Cheers,
Bertrand

Hello,

This MR adds the functionality to thin the CaloCalTopoClusters_links container and only keep the entries associated with tau clusters.
This container is needed for the tau cluster-to-cell navigation in xAODs, but it too large to be written out without thinning.
As discussed in ATLTAU-1733, the CaloClusterCellLinkContainer thinning was not supported so far, and Scott just implemented it (not merged yet).
Once it's in Athena, the CaloCalTopoClusters_links size should be reduced significantly (I'm already adding the container to the xAOD output to minimise the number of MRs, so we expect a jump in xAOD size, ATLASRECTS-5684).
According to q221 run on 40 events, thinning reduces the container size by a factor of 2.5 (3.0 kb/evt after thinning).

Cheers,
Bertrand

Hello,

This MR is implementing an improved version of the tau pi0 cell thinning in xAODs (ATLASRECTS-5684, ATLTAU-1733).
The pi0 clusters are initially built from cells stored in a separate container (TauCommonPi0Cells).
An algorithm (ClusterCellRelinkAlg) is introduced to recreate pi0 clusters with cells from the main cell container (AllCalo), thereby eliminating the overhead from our separate cell container.
The ESD and AOD contents have been adjusted accordingly: TauCommonPi0Cells is dropped, and the relinked pi0s are written out instead of the initial pi0s.
There will be a follow-up MR that will drop some other pi0 container we used to use in TauPi0ClusterCreator, which should no longer be needed.
For now, only the minimum amount of TauPi0ClusterCreator code was changed.

Another algorithm (TauCellThinningAlg) is introduced to save the cells from the tau seed-jet topoclusters in xAODs.
Only clusters within 0.2 of the tau are considered, to limit the xAOD size increase.
Note: in the xAODs, we are still missing CaloCalTopoClusters_links (the CaloClusterCellLinkContainer) for the tau seed-jet clusters.
An attempt at thinning this container was made, but it was not successful.

Cheers,
Bertrand

If we configure LAr bytestream writing, add CaloNoise as an explicit
dependency of the output stream.

Hello,

As discussed in ATLASRECTS-5684, the pi0 cell thinning implemented in !37160 does not work for the RAWtoALL workflow.
This MR should fix this.

Cheers,
Bertrand

Addresses ATR-21701

Hello,

As discussed in ATLASRECTS-5684 and ATLTAU-1733, this MR is implementing the easiest mechanism we have to decrease the size of the tau pi0 cell container in the xAOD.
We simply keep the cells that belong to pi0 topoclusters.
It is not the optimal solution, though, but merging TauCommonPi0Cells and AllCalo will require more work.
This is under study.

Cheers,
Bertrand

Frank Berghaus authored 4 years ago and Adam Edward Barton committed 4 years ago

This has just laid dormant ever since the CMT->CMake migration. :-(

At the same time introduced some tests for the implicit copy-constructor
and copy-operator of xAOD::EventFormat.

Peter Onyisi authored 4 years ago and Edward Moyse committed 4 years ago

Need <cstdlib> for size_t.

Frank Winklmeier authored 4 years ago and Vakhtang Tsulaia committed 4 years ago

Delete some obviously obsolete job options (e.g. referencing
TrigSteer_L2).

With athenaMT it is best to rely on the default (ROOT) signal handler in
CoreDumpSvc, which uses gdb to produce the stacktrace for all threads.
Setting `FatalHandler = 438` makes use of the `SealSignal` handler,
which only produces one trace. For production environments where the gdb
traces are not needed, one should override the setting in the transform
or AMI tag as is already done today.

Relates to ATEAM-629.

Delete the TrigTier0 package and move the remaining three files to
TriggerJobOpts. Adjust references to these files in other packages.

SealSvc does not have any useful function any more

ThinningSvc is no longer used and is being removed.

Frank Berghaus authored 4 years ago and Walter Lampl committed 4 years ago

The ByteStreamEventStorageInput service provided a legacy job option called FullFileName to specify bytestream input. With the move to EventSelector this option was all but removed from ByteStreamEventInputSvc. The EventSelectorByteStream provieded the input option to specify input. It captured the FullFileName option of the InputSvc. This commit removes the relevant parts of the ByteStreamEventStorageInputSvc and EventSelectorByteStream. It also attempts to move all references in other job options (unless they specify being for an older release).

Run LumiBlockMuWriter on BS files or if the input file already
has xAOD::EventInfo.  Needed to fix some trigger ART tests.

It will fill the decorations with 0 in this case.

fix the setting of beam_energy and AMITag to avoid saving more than one value in the TagInfo metadata

Introduced new LumiBlockMuWriter algorithm, which fills pileup information
on xAOD::EventInfo using  WriteDecorHandle-s.

Enabled LumiBlockMuWriter algorithm in RecExCommon_topOptions.py for RAWtoESD
configurations over ByteStream input, and also unconditionally in
skeleton.RDOtoRDOtrigger.py and in runHLT_standalone.py

Ported one client - JetCalibrationTool - to accesing pileup
values in xAOD::EventInfo using ReadDecorHandle

Fixed ref_RDOtoRDOTrig_mt1_build.ref with updated trigger counts

StatusCodeSvc is deprecated. Remove all references to it from job
options.