Frank Harrison
Frank Harrison
@joshua-cannon-techlabs said: > From profiling, most of the time taken is traversing the file's dependencies and parsing them as well. @cdce8p said: > As far as I'm aware most time...
It looks like the _intent_ is for the license to be MIT, but because it statically links in ODGF (LGPL) it is actually an LGPL license. So, buyer beware.
The route-cause would be fixed by #1167
See #33. We'll probably merge those changes in soon-ish, once the licensing issue is fixed. @nateharms we probably won't support conda in the future.
The map/reduce mixin (`pylint.checkers.mapreduce_checker.MapReduceMixin`) itself isn't designed to distribute computation (pylint already is trying to do that via `multiprocessing` in `lint.parallel.check_parallel`), it just helps collate data about those runs. Really...
My view remains that `lint.parallel.check_parallel` should be fully incorporated into `pylint.lint.pylinter.PyLinter.check` and not just invoked by it. A better candidate may be `pylint.lint.pylinter.PyLinter._check_files` because that is where we know what...
> the documentation needs to be updated Maybe. Although I'm not sure the work has been done to demonstrate/understand the ins and outs here to document it properly. For example,...
I imagine there would be several stages of attack, in order: 1. gen the ast on main and share with sub-procs. 2. multi-proc/thread the ast-generation. 3. cache the ast (ala...
Thanks for sharing the fascinating differences between `flake8` and `pylint`. Really interesting. > we still rely on linter for all checkers Yes, I noticed. I don't understand quite why that...