Using VSCode extension to measure variability (e.g. verificationLogger, randomSeedIterations) #437
-
I'm seeing a lot of variability in the time of verification (see What advice to folks have for debugging this within VSCode? Can you run with verificationLogger and randomSeedIterations easily? Can you see resources used? Thanks, |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 1 reply
-
You can right-click on a Dafny file, Build with custom arguments, and replace everything with |
Beta Was this translation helpful? Give feedback.
-
Thanks, I ended up getting things working on the command line. Here are my notes on the commands (in a Windows CMD console):
Others might want different values for the time limit, cores, # seed iterations, log format, and whether to use Split. (So, I now have this running, but I don't have a good understanding of which log format to use or how to use the logging results to improve my verification.) |
Beta Was this translation helpful? Give feedback.
You should use the dafny-reportgenerator tool. It's a bit heavyweight but can summarize the results nicely.
What it does is that it regroups results for a single method, so you can see how many times they fail and how many time they succeed.
If you have 1 success and 9 failures for a single method, it's that your proofs are brittle. If you have 10 successes and 0 failures, your proofs are currently stable.
If you have 0 success and 10 failures, Dafny can't prove it.