You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Aug 22, 2024. It is now read-only.
Currently a set of constants in checker/src/k_limits.rs provides limits on how much computation MIRAI will do in various places. With these limits in place, largish crates can be analyzed in a somewhat reasonable time frame. The limits do, however, impact on precision and may lead to false negatives. By making these into variables that can be set via command line options, it will become possible to schedule longer runs that may find more issues than more limited runs. The greater precision allowed by such runs can also be used to weed out false positives.
The recommended way to complete this task is to turn the constants in k_limits into the fields of a struct and to provide a constructor that sets the fields to default values, along with a command line processor that updates the values from options in the command line.
The limits struct should then be made part of the visitor state (checker/src/visitor.rs).
Ideally, there should also be a way to set these options to non default values for integration test cases. See also #52.
The text was updated successfully, but these errors were encountered:
I began working on this issue, it appears that the k_limit constants also need to be integrated into callbacks (checker/src/callbacks.rs), is this correct?
Issue
Currently a set of constants in checker/src/k_limits.rs provides limits on how much computation MIRAI will do in various places. With these limits in place, largish crates can be analyzed in a somewhat reasonable time frame. The limits do, however, impact on precision and may lead to false negatives. By making these into variables that can be set via command line options, it will become possible to schedule longer runs that may find more issues than more limited runs. The greater precision allowed by such runs can also be used to weed out false positives.
The recommended way to complete this task is to turn the constants in k_limits into the fields of a struct and to provide a constructor that sets the fields to default values, along with a command line processor that updates the values from options in the command line.
The limits struct should then be made part of the visitor state (checker/src/visitor.rs).
Ideally, there should also be a way to set these options to non default values for integration test cases. See also #52.
The text was updated successfully, but these errors were encountered: