Make prinout of differences in Counter, 1DHistograms, and 1DProfiles deterministic
As an example for how it is right now, have a look at the brunel-extraptester
test in :
- https://lhcb-nightlies.web.cern.ch/nightly/lhcb-master-mr/2562/Brunel/x86_64_v2-centos7-clang11-dbg/tests
- https://lhcb-nightlies.web.cern.ch/nightly/lhcb-master-mr/2562/Brunel/x86_64_v2-centos7-clang11-opt/tests
The order of the algorithms and counters reported isn't fixed, which makes the comparison harder than it needs to be.
This MR tries to ensure we print these differences in alphabetical order.
Merge request reports
Activity
added Build label
- Resolved by Rosen Matev
/ci-test --merge Moore!771 (merged) Brunel!1138 (merged) Rec!2376 (merged)
Starting a normal ci-test wouldn't do much, since we need differences to check that this MR does what it says.
Thus I'll run a test and include @lohenry's seeding MR as that should introduce plenty of diffs.
And one more free round of testing for @lohenry
added ci-test-triggered label
- [2021-07-13 17:22] Validation started with lhcb-master-mr#2568
- Resolved by Christoph Hasse
I wonder if it is possible to have a look only at relative differences, through an option. I expect 0.1%-ish differences all over the place, but by eye I need to each time estimate if it is acceptable or not.
mentioned in commit fe8bc300
added enhancement label