Benchmarking ACCP when it's bundled with JDK #360
Conversation
amirhosv
changed the title
| } | ||
| } | ||
|
|
||
| static void removeAllProviders() { |
There was a problem hiding this comment.
I believe the original reason to remove/setup individual providers was to isolate utilities classes or other crypto instances that may be requested internally by providers without specifying a provider explicitly. One instance of that can be found in JDK11's MGF1, some others here.
My concern here is that without isolating individual providers/provider suites, then we'll get less accurate measurements due to cross-provider instance sourcing (e.g. getting a MGF1 instance from JDK11 that internally uses a higher-JCA-priority MessageDigest instance from BouncyCastle). Do we need this particular change (installing all providers all the time) to satisfy the overall intent of the PR?
There was a problem hiding this comment.
Thanks for catching that.
Description of changes:
The change allows to run benchmarks with ACCP when it's bundled with JDK.
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.