Exit gracefully if no IDVAlgorithm is found

LoKi functors, as well as some other tools, assume they are owned by a DVAlgorithm and use the tools configured within. Typically, BPVIPCHI2() assumes the existence of the geometry tool. This was reported bu @amathad in https://gitlab.cern.ch/lhcb-dpa/project/-/issues/173 following https://mattermost.web.cern.ch/lhcb/pl/tddzihtuitbdtq1ifajt4pj6zh.

There's no point fixing LoKi as eventually its functors will be replaced by ThOr. BPVIPCHI2 is already. In the meantime make sure there is no crash but exit gracefully instead. With this MR the message is

FunctionalParticleMaker.LoKi::Hy...   FATAL  Exception with tag=Gaudi::Utils::getIDVAlgorithm is caught
FunctionalParticleMaker.LoKi::Hy...   ERROR Gaudi::Utils::getIDVAlgorithm 	 Cannot find IAlgorithm. You probably cannot use this tool/functor in this context	 StatusCode=FAILURE
ToolSvc                               ERROR Error initializing tool 'FunctionalParticleMaker.LoKi::Hybrid::TrackSelector'
HiveDataBrokerSvc                     FATAL in sysInitialize(): exception with tag=FunctionalParticleMaker is caught
HiveDataBrokerSvc                     ERROR FunctionalParticleMaker 	 Failed to retrieve tool LoKi::Hybrid::TrackSelector/LoKi::Hybrid::TrackSelector	 StatusCode=FAILURE
ServiceManager                        ERROR Unable to initialize service "HiveDataBrokerSvc"
HLTControlFlowMgr                     FATAL Error retrieving HiveDataBrokerSvc
ServiceManager                        ERROR Unable to initialize Service: HLTControlFlowMgr
ApplicationMgr                        ERROR Application Manager Terminated with error code 1

I suggest to backport to run 2.

added DPA-WP3 Functors backport run2 labels

requested review from @rmatev

assigned to @lpica

added 1 commit

• fa4915c4 - Fixed formatting

Compare with previous version

added 1 commit

• 4a86722a - Remove returns

Compare with previous version

resolved all threads

added lhcb-run3-cleanup label

Hi @pkoppenb, can I test this?

For me, yes. Thanks

/ci-test --merge

Started reference and integration test builds. Once done, check the comparison of build and test results.

Throughput Test Moore_hlt2_reco_baseline: 0.0 Events/s -- change of -100.00% vs. reference

Throughput Test Moore_hlt2_fastest_reco: 0.0 Events/s -- change of -100.00% vs. reference

Throughput Test Moore_hlt2_pp_thor: 0.0 Events/s -- change of -100.00% vs. reference

Throughput Test Moore_hlt1_pp_default: 0.0 Events/s -- change of -100.00% vs. reference

Hi @pkoppenb, does the message look like the one you were expecting?

There are many unrelated errors which I do not understand. I don't understand what is going on here.

@rmatev, can you take a look at the results?

This was also built in lhcb-run3-cleanup which seems happy https://lhcb-nightlies.web.cern.ch/nightly/lhcb-run3-cleanup/372/

@pkoppenb I believe the right link is https://lhcb-nightlies.web.cern.ch/nightly/lhcb-run3-cleanup/373/ which has the same errors as above (already some LHCb tests fail, probably best to start there).

Indeed. I must have been too aggressive with exceptions.

added ci-test-triggered label

• [2021-11-22 16:36] Validation started with lhcb-master-mr#3172
• [2021-11-23 00:55] Validation started with lhcb-run3-cleanup#373

added hlt2-throughput-decreased label

added hlt1-throughput-decreased label

Indeed. It seems LoKi internally is sometimes checking for IDVAlgorithm to be nullptr and then do something else. This MR is thus not how https://gitlab.cern.ch/lhcb-dpa/project/-/issues/173 can be solved.

closed

mentioned in issue Moore#356 (closed)