Uh oh!
There was an error while loading. Please reload this page.
Avoid std::map in EventSelectionModule - #16396
Conversation
ktf
commented
May 26, 2026
@ddobrigk can you have a look here? Looking up in a map is very expensive due to the pointer traversal and constructing it requires one allocation per node which is also expensive. A binary search in a vector of pairs should be much faster. Also, @aalkin was suggesting to use a index table for this? |
ktf
commented
May 27, 2026
@ekryshen can you have a look at this? |
ekryshen
commented
May 28, 2026
@ktf Thanks, looks reasonable. Actually I suspect that we don't need this map or vector of pairs any longer. It was introduced as a workaround for early passes of pp2022 where trigger info wasn't aligned with actual bcs. As far as I understand, the trigger functionality is used by EMCAL analysers only. I will check with them if they still need support of old passes. If not, we can simply remove this workaround. |
ekryshen
commented
May 29, 2026
@ktf I've just got a confirmation that early passes are not used by ECAL analysers so this functionality is not needed any longer. How should we proceed? We can either merge or close this PR, then I will clean up this code. |
ktf
commented
May 29, 2026
removing the mapping completely is of course fine with me. can you take care of it? |
ekryshen
commented
Jun 1, 2026
Done here #16489. I guess we can close this PR |
ktf
commented
Jun 1, 2026
Thanks |
No description provided.