In 0.6 findfirst(::Function, ::A) would return the index of the first element that caused the predicate function to return true, and 0 if it is not present. Now it returns ‘nothing’… of type ‘Nothing’, so will this cause issue with type inference and compilation of efficient code, if you don’t know …

Not sure, but you can replicate part of findfirst’s behaviour with map() fairly easily. It will run over the whole array though, so it would be more like a findall function. You’d probably have to check the output for zeros/nothings at some point though, except where the value is certain to be in t…

Yes, it’s not type stable, but the language has been enhanced (and is in the process of being further enhanced) to allow small unions to be handled efficiently, so there’s no performance penalty anymore.

[image] Keno: Yes, it’s not type stable, but the language has been enhanced (and is in the process of being further enhanced) to allow small unions to be handled efficiently, so there’s no performance penalty anymore. Brilliant! Quick follow-up: are there also plans to stop highlighting small…

They’re highlighted in yellow on master.

New findfirst behaviour - type unstable or no? - #4 by Keno - General Usage

Topic		Replies	Views	Activity
New find* methods and type stability General Usage	6	1138	February 1, 2018
The new output of findfirst is dangerous General Usage	21	4372	February 24, 2018
Suggestion for more general/performant sentinel for find* functions Internals & Design	17	1730	January 29, 2018
Findfirst for Dicts with `nothing` keys General Usage	33	3654	March 4, 2018
Findfirst that returns the value/element New to Julia	16	4065	January 23, 2018