Skip to content
Snippets Groups Projects
Phil de Joux's avatar
Phil de Joux authored
- Satisfy -Wmissing-signatures in test-runtime-deps
- Satisfy -Wx-partial in HackageBenchmark
- Satisfy -Wunused-imports in QuickCheck.Instances.Cabal
- Use partial pattern for filtering in list comprehension
- Don't error on deprecated import
8453ee0b
History
Code owners
Assign users and groups as approvers for specific file changes. Learn more.

Dependency Solver Benchmarks

hackage-benchmark

The goal of this benchmark is to find examples of packages that show a difference in behavior between two versions of cabal. It doesn't try to determine which version of cabal performs better.

hackage-benchmark compares two cabal commands by running each one on each package in a list. The list is either the package index or a list of packages provided on the command line. In order to save time, the benchmark initially only runs one trial for each package. If the results (solution, no solution, timeout, etc.) are the same and the times are too similar, it skips the package. Otherwise, it runs more trials and prints the results if they are significant.