cce
cce
Invalid points would have led to false before too, right?
Yes the issue was the linter was not correctly blocking builds with these issues, I think I fixed it in the reviewdog.yml change, and @winder has made this linter part...
huh, fixing these errcheck linter warnings led to another issue: ``` Failed panic: flag "app-arg" does not exist goroutine 1 [running]: github.com/algorand/go-algorand/cmd/goal.panicIfErr({0x1b3bd20, 0xc00059e160}) /opt/cibuild/project/cmd/goal/application.go:199 +0x88 github.com/algorand/go-algorand/cmd/goal.init.1() /opt/cibuild/project/cmd/goal/application.go:194 +0x1dcc FAIL github.com/algorand/go-algorand/cmd/goal...
A few releases ago, I was worried that if we made a change to add a new case where we call os.Remove() on part keys, it might surprise users still...
Well, I still worry that since we currently only delete your part keys when you explicitly provide `--delete-input` to goal, it may be a big surprise to some to have...
Some ideas: - use [test splitting by timing data](https://circleci.com/docs/parallelism-faster-jobs#splitting-by-timing-data) which uses CircleCI's previous test timing data to optimally assign tests to shards (vs random assignment like we are doing today)....
works for me? ``` ubuntu@ip-172-31-31-128:~/go/bin$ ldd algorelay statically linked ubuntu@ip-172-31-31-128:~/go/bin$ ldd algotmpl statically linked ubuntu@ip-172-31-31-128:~/go/bin$ ldd algocfg statically linked ubuntu@ip-172-31-31-128:~/go/bin$ ldd algons statically linked ```
There is some anecdotal evidence that this change makes the time.Sleep() timing assumptions in the metrics tests flakier.. the metrics tests are failing on all three platforms. https://app.circleci.com/pipelines/github/algorand/go-algorand/3189/workflows/b08be5df-27a3-4aaa-b7c6-04c492bdc3e9/jobs/42614/tests https://app.circleci.com/pipelines/github/algorand/go-algorand/3189/workflows/b08be5df-27a3-4aaa-b7c6-04c492bdc3e9/jobs/42610/tests https://app.circleci.com/pipelines/github/algorand/go-algorand/3189/workflows/b08be5df-27a3-4aaa-b7c6-04c492bdc3e9/jobs/42619/tests...
Well, the tests pass now, and coverage reports more accurately reflect what is being tested, but collecting atomic coverage reports from all the lines in these previously-uncounted go-algorand package import...
Can revisit later