Eric McDonald

I also see this fairly often, though it is not consistently reproducible, which makes me think that there is a timing or data race issue involved. (All occurrences that I...

Also sometimes see the following, which further corroborates the notion of this being a data race: ``` Scanning 6 files with 158 python rules. 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████|6/6 tasks [Errno 2] No such...

As I mentioned earlier, it definitely bears the characteristics of a race. In my case, I can see it happening occasionally on the standard Github Actions workers (CI job) as...

> Off the top of my head, my first suspect for where a latent race might hide is the `Parmap` mechanism, and thus I might suggest trying to reproduce it...

Finally had a chance to dive on the code some. When I saw `clone` calls in the syscall trace, I assumed that threads were being used for the workers, but...

Sure thing. @emjin. I was testing under 0.111.1 last week, but I am still able to sporadically reproduce the issue with 0.112.1. ``` $ semgrep --version 0.112.1 $ python -c...

@emjin: If you are running Semgrep under PyPy, you can get 100% reproducibility by invoking `gc.collect` just inside the `with` block, like so: ``` with open(rule_file_name, "w+") as rule_file, open(...

Thanks for merging the PR that I submitted. I suspect that the PR only addressed the failures that I was seeing. One thing that might help others would be to...

I would expect this to be a plugin for the Python Markdown-It. It is more than just JS code, since a relatively-positioned HTML button must be generated as part of...

Code coverage error appears to be a Codecov rate-limiting issue. However, I just realized that I missed the `Literal` case that you mentioned in the issue. I need to cover...