Use bit manipulation with the UniqueIDGenSvc and add a check for repeated numbers #247
Conversation
and add a check for the case where an ID is repeated.
jmcarcell
changed the title
Co-authored-by: Thomas Madlener <[email protected]>
Co-authored-by: Thomas Madlener <[email protected]>
| std::tie(std::ignore, inserted) = m_uniqueIDs.insert(hash); | ||
| } | ||
| if (!inserted) { | ||
| warning() << "Event number " << evt_num << ", run number " << run_num << " and algorithm name \"" << name |
There was a problem hiding this comment.
Maybe this should be fatal, with an option to turn it into a warning? Nobody reads warnings.
There was a problem hiding this comment.
Is it possible/likely to enter this condition 'by chance' while everything is actually sane? If not then I would be in favor
There was a problem hiding this comment.
Possible hash collision, but it shouldn't be likely. And if it happens too often, we have a problem we should know about.
There was a problem hiding this comment.
I changed the default to throw whenever the same ID is asked, with an option not to if needed. Indeed if this is buried in thousands of lines of logs it's not very useful. This will throw if someone runs the same algorithm multiple times with the same name; for example, if digitizing multiple collections with the same digitizer and running it multiple times.
BEGINRELEASENOTES
UniqueIDGenSvc:UniqueIDGenSvcwhich is about 1000 times faster than the string manipulation that was being doneENDRELEASENOTES
The difference between the bit manipulation versus strings is so big that even when adding the check it will be faster than before. I checked that this PR doesn't change the hash that is returned.