Closed Zentrik closed 10 months ago
Merging #600 (87f72a9) into master (5a4517d) will increase coverage by
0.14%
. The diff coverage is18.18%
.
@@ Coverage Diff @@
## master #600 +/- ##
=========================================
+ Coverage 8.89% 9.03% +0.14%
=========================================
Files 40 40
Lines 2688 2690 +2
=========================================
+ Hits 239 243 +4
+ Misses 2449 2447 -2
Files | Coverage Δ | |
---|---|---|
src/cache.jl | 37.03% <100.00%> (ø) |
|
src/utils.jl | 35.55% <50.00%> (+0.67%) |
:arrow_up: |
lib/OptimizationOptimisers/src/sophia.jl | 0.00% <0.00%> (ø) |
|
lib/OptimizationBBO/src/OptimizationBBO.jl | 5.40% <0.00%> (ø) |
... and 1 file with indirect coverage changes
:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more
This fixes a performance regression due to 92974b8526ca031f9c6ed295462b8a9ba45eeacb. The regression is because the defaults for
callback
anddata
were changed without changing the code that branched on whether they were set to their defaults.I removed
fieldnames
as that seems to not run at compile time (it was taking about 50% of the time for my use case due to having to iterate over it), probably because its effects are(!c,!e,!n,!t,!s,!m,+i)′
. I believe you could fix this by using generated functions but I don't know how to. Cthulhu complains that callinggetproperty
withu0
andp
has anAny
return type, though I don't see it in the profiler so maybe Cthulhu is wrong here? If this is an issue, I'm not sure how to fix it, perhaps this feature should be removed.Some tests should be added to prevent this regression in the future, I've added a test to check for how many bytes are allocated.