Shaun Roe authored 1 year ago and Frank Winklmeier committed 1 year ago

main-cppcheck_followup-EventLoop

Shaun Roe authored 1 year ago and Frank Winklmeier committed 1 year ago

main-cppcheck-EventLoop

Ruggero Turra authored 1 year ago and Vakhtang Tsulaia committed 1 year ago

Fix condor driver without shared filesystem

* Don't use /tmp
* Move to cpack to create the tarball

Ruggero Turra authored 1 year ago and Frank Winklmeier committed 1 year ago

correctly pass x509userproxy to condor

Johannes Elmsheuser authored 1 year ago and Edward Moyse committed 1 year ago

EventLoop: fix gcc13 compilation

Andrii Verbytskyi authored 1 year ago and Adam Edward Barton committed 1 year ago

Drop unused find_package(boost) and boost headers

Andrii Verbytskyi authored 1 year ago and Johannes Elmsheuser committed 1 year ago

Hotfix of the MR 66055

Andrii Verbytskyi authored 1 year ago and Johannes Elmsheuser committed 1 year ago

Remove almost unused StringUtils

Xingyu Wu authored 1 year ago and Johannes Elmsheuser committed 1 year ago

implement nEventsPerJob options for PrunDriver

Well, I guess that the lesson here should be always to write unit
tests when if statements are involved, not just rely on the
integration tests.  On th plus side the new infrastructure made it a
lot easier to test this, and I may have fixed a longer-standing bug
since the previous implementation had never been tested.

The existing code would just use/replace the existing directory.

This decouples the actual selection of input files and event ranges
from the implementation of the worker.  And it is part of the overall
goal of splitting the worker functionality into a series of modules.

I thought I had fixed the `DuplicateChecker` test in !65652, but while
I created a unique directory in each test now, I wasn't actually
putting anything into that directory. 

This should be fixed now.

Nils Krumnack authored 1 year ago and Nils Erik Krumnack committed 1 year ago

That was creating some issues in nightly tests.

GridDriver is based on Ganga, which is not really the current
recommendation anymore, or even last year's recommendation.  Given
that some problems are being reported in conjunction with GridDriver
I'm just removing it (as agreed with the AMG conveners).

Teng Jian Khoo authored 1 year ago and Vakhtang Tsulaia committed 1 year ago

Make DualUseConfig/ConfigAccumulator CA-compatible and create CA FullAlgorithmTest instances

We are trying to make it possible to run user n-tuple production as
part of central production.  For that it is very helpful if the jobs
do the proper error reporting.  For GridDriver and PrunDriver that is
done by default, and that is the recommended way to do it.  However,
some users have their own wrapper scripts and use DirectDriver on the
grid.  Those users can now enable the grid reporting functionality in
their jobs as well.

In the existing code it tries to read the sample metadata based on the
dataset.  This works fine, but breaks for central n-tuple production,
as that wants to run on new datasets that weren't known at submission.
I now added a default behavior which is hopefully sufficient for the
purpose.

Andrii Verbytskyi authored 1 year ago and Walter Lampl committed 1 year ago

Use C++17 std::filesystem instead of boost::filesystem where possible

Zach Marshall authored 1 year ago and Vakhtang Tsulaia committed 1 year ago

Update to batch driver to use CMAKE env variables

In case the workdir is not set, trying to use the first part of the
CMake path to provide the appropriate setup area for the event loop
batch driver.

It was pointed out that the name in AnalysisBase wasn't matching the
name in Athena, so I added an alias in AnalysisBase.  I kept the old
version around because removing it would break users code.

While it was seemingly supported so far, it didn't actually work.
This should now be fixed, and a test for it added.

I also went ahead and made the basic function a template, so that
users can use it with any histogram-like class deriving from TObject.
Though that part will only work in AnalysisBase, as I only modified
the AnalysisBase part.

Nils Erik Krumnack authored 2 years ago and Walter Lampl committed 2 years ago

add "fixes" to make AnalysisBase compile natively on MacOS

This mostly involves adding some missing includes, but I also had to
disable some things that xcode doesn't seem to support.

Zach Marshall authored 2 years ago and Walter Lampl committed 2 years ago

Custom error handler for ROOT file opening

This uses WithRootErrorHandler to catch ROOT errors during file opening
in EventLoop. If an error message is detected, an exception is thrown.

This follows on from the discussion in !61079 and replaces that MR.

This is very experimental, but at the very least it would allow to do
configuration based on the meta-data.

See JIRA ATLASG-2449 , using `TPython::Eval` with string as return type
causes issues with python3.7+. This is used in `PrunDriver.cxx`
to get JEDI task status for wait/retrieve. Changing type to int based on
enum defined both in `ELG_jediState.py` and `PrunDriver.cxx` to solve the issue.

Also cleaning-up some commented lines.

Note to return of the ELG_jediState:

Task states not mentioned above the state as a string used to be added
to nc_ELG_state_details meta information. I had a look and in whole athena
this is used in PrunDriver and only for print-outs so I think its okay
to just say "its not these states"?

Actually covering all the cases with enum would be messy
(there is around twenty possible states, see here:
    https://github.com/PanDAWMS/panda-bigmon-core/blob/master/core/constants.py#L85
)

Zach Marshall authored 2 years ago and Adam Edward Barton committed 2 years ago

Adding support for TEfficiency in AnaAlgorithm

Title says it all. Came up on a PATHelp thread.

Shaun Roe authored 2 years ago and Tadej Novak committed 2 years ago

23.0-cppcheck-SampleHandler

Scott Snyder authored 2 years ago and Walter Lampl committed 2 years ago

RootCoreUtils: Suppress cppcheck warning.

substr()

Andrii Verbytskyi authored 2 years ago and Walter Lampl committed 2 years ago

cppcheck: Reduce the number of useless string operations

Andrii Verbytskyi authored 2 years ago and Walter Lampl committed 2 years ago

cppcheck: 	uselessCallsSubstr

Nils Erik Krumnack authored 2 years ago and Atlas Nightlybuild committed 2 years ago

Bug Fix for external files added to tarball sent to the grid when using `EL::PrunDriver`

See merge request atlas/athena!59296

(cherry picked from commit fc62ed6b)

e671f969 Adding changes

Teng Jian Khoo authored 2 years ago and Edward Moyse committed 2 years ago

Actually check data type for large-R jet sequence

Romain Bouquet authored 2 years ago and Vakhtang Tsulaia committed 2 years ago

Manual Sweep atlas/athena!57533: PrunDriver: enable adding additionnal files to the TarBall sent to the grid thanks new option `optGridPrunShipAdditionalFilesOrDirs`

Manual Sweep atlas/athena!57533: PrunDriver: enable adding additionnal files to the TarBall sent to the grid thanks new option `optGridPrunShipAdditionalFilesOrDirs`

Zach Marshall authored 2 years ago and Edward Moyse committed 2 years ago

BugFix to GridTools for proxy time output parsing

If grid tools aren't installed as an RPM (as they are on lxplus by default), then we get slightly different output from the associated commands. This requires a change to the parsing. Addresses an issue raised today in PATHelp.