Conversation
Codecov Report
@@ Coverage Diff @@
## master #66 +/- ##
==========================================
+ Coverage 90.45% 91.70% +1.25%
==========================================
Files 6 6
Lines 398 398
==========================================
+ Hits 360 365 +5
+ Misses 38 33 -5
Continue to review full report at Codecov.
|
mmiller-max
force-pushed
the
mm/speed-up-tests
branch
5 times, most recently
from
2fca078 to
2abcac1
Compare
|
OK this is better, but still not great. On Ubuntu and Julia V1 its gone from 8m to 5m30s. Changes were:
I think most of our time is in the creation of a new Julia process each time Open to any suggestions, but also happy just to close this and move on. Can always revisit it later. |
| @testset "Runner tests" begin | ||
| @testset "Installed packages by name in primary environment" begin | ||
| # Pkg.add | ||
| Pkg.UPDATED_REGISTRY_THIS_SESSION[] = true |
There was a problem hiding this comment.
| Pkg.UPDATED_REGISTRY_THIS_SESSION[] = true | |
| # for speed avoid updating registry, we have a fresh one anyway | |
| Pkg.UPDATED_REGISTRY_THIS_SESSION[] = true``` |
| end | ||
|
|
||
| # Own time_taken field | ||
| mutable struct TimeTakenTestSet <: AbstractTestSet |
There was a problem hiding this comment.
I see this have been move to not be in a seperate process, that makes sense
- Don't activate test package environment - Stop checking registry as much - Remove tests that duplicate testing - Simplify all test packages - Don't clone a fresh registry - we're not testing it
mmiller-max
force-pushed
the
mm/speed-up-tests
branch
from
2abcac1 to
30877bc
Compare
2 participants
Not sure how much this will affect test speed, so a WIP PR.
Will close #51 eventually.