Closed Tortar closed 10 months ago
I have to say that this means that probably there is something to improve in Base, since that the previous method is so slow sounds strange
Merging #897 (737312d) into main (a71cc12) will increase coverage by
12.09%
. The diff coverage is60.00%
.:exclamation: Current head 737312d differs from pull request most recent head 90a22d6. Consider uploading reports for the commit 90a22d6 to get more accurate results
@@ Coverage Diff @@
## main #897 +/- ##
===========================================
+ Coverage 80.27% 92.37% +12.09%
===========================================
Files 44 32 -12
Lines 3027 2334 -693
===========================================
- Hits 2430 2156 -274
+ Misses 597 178 -419
Files | Coverage Δ | |
---|---|---|
src/core/model_abstract.jl | 89.06% <66.66%> (-0.03%) |
:arrow_down: |
src/core/model_concrete.jl | 88.23% <50.00%> (-1.18%) |
:arrow_down: |
... and 18 files with indirect coverage changes
:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more
Actually it was a O(N) operation while this one is not, this means that the improvement in speed is arbitrary large!
I think it should be https://github.com/JuliaLang/julia/issues/51605 , let's see :-)
It was actually very slow before, consider this silly model I was using to benchmark a different thing:
before:
after:
A micro-benchmark tells that now random_agent for a StandardABM is 40x faster!
I exported a new function
random_id
in the process, which is in line to the other random functions which export also the id version