Closed tcharding closed 2 months ago
We need to be careful before merging this one. Its a good chance to check all our repos that use this tool to see that they either:
If anything is not using a specific commit hash then we need to change them to do so. We can't paralyze this repo by having its master
branch automatically deployed to a dozen other projects.
In 3c31ce7794d0b12fa3c2a54edbf43e193ea7ff80:
This fix is fine though I think we should just stop treating the name fuzz
specially. I don't treat it specially in my local CI setup. I look for honggfuzz deps and run cargo hfuzz
on all binaries in crates that they appear (and recently, I even choose the correct version of honggfuzz based on the lockfile).
CI here is failing. I think we should fix it.
You were reviewing while I was hacking :)
If anything is not using a specific commit hash then we need to change them to do so. We can't paralyze this repo by having its master branch automatically deployed to a dozen other projects.
Agreed, I'll check the users to make sure this is the case.
I'm going to come back to this, its more important to check all the users. (And I'm confused why shellcheck -x foo
is warning, that is a sign to come back with a fresh brain.
Only miniscript
was missing the use of rev
(https://github.com/rust-bitcoin/rust-miniscript/pull/720).
This fix is fine though I think we should just stop treating the name fuzz specially.
I had a go doing this but if we treat it like other crates it requires a contrib/test_vars.sh
file which seems clunky.
I'm not sure what is going on with the CI job, shellcheck ci/run_task.sh
runs cleanly for me locally. Can you check on your machine please @apoelstra
I had a go doing this but if we treat it like other crates it requires a
contrib/test_vars.sh
file which seems clunky.
Doing this would correctly capture the fact that some of our fuzz harnesses still have honggfuzz
features while others don't.
But in any case, it seems less clunky to me to have a contrib/test_vars.sh
than to have a particular workspace which we special-case based on its directory name, and don't even try to run the unit tests on.
@tcharding when I run shellcheck locally the error appears. It is complaining that the say_err
function is defined but never called.
We probably want to globally whitelist "unused code" type warnings on this repo.
Cheers, I found a place that we were using echo
to print an error message so I used say_err
. I'm still unsure why my shellcheck
was not showing the warning.
We probably want to globally whitelist "unused code" type warnings on this repo.
Not done but I'll keep it in mind.
This PR dose a few things:
say_err
to clear anothershellcheck
warning (that I was unable to repro locally)cargo build
andcargo test
for all crates instead of treatingfuzz
as a special caseResolve: #10