Re: tclint, pre-commit and patch submissions

Hi Tom,

I'm in favor o enforcing clean commits. If we are using a linter for TCL,
then I'd personally go with this option:

- we make the repo refuse commits that are not pre-commit clean

Or a pre-commit checker that we carry on our repo that gets run
automatically.


On Wed, Oct 1, 2025, 12:55 Tom de Vries <[email protected]> wrote:

> Hi,
>
> I've recently added a tclint pre-commit hook, and started cleaning the
> testsuite.
>
> The current status is that all gdb.* dirs are done, with the exception
> of gdb.stabs (which is going to be removed, so I skipped it).
>
> My question is what to do with patches adding new test-cases or
> modifying existing test-cases.
>
> If these patches are not tclint-clean, then after commit this command:
> ...
> $ pre-commit run tclint --all-files
> ...
> will start showing new tclint errors.
>
> One thing that could be done at that point is to ask the
> submitter/committer to fix the tclint errors.
>
> But there's no formal agreement atm that this need to be fixed.  I've
> proposed the hook, and Tom Tromey approved it, but that's just two
> maintainers.
>
> So I'd like to know the opinion of other maintainers.
>
> Possible outcomes of this discussion could be that:
> - we get rid of tclint in pre-commit and forget about it
> - we drop one or more error categories in gdb/tclint.toml
> - we require submitters to run pre-commit before submission
> - we make the repo refuse commits that are not pre-commit clean
> - nothing changes, and submitters can use all/some/no pre-commit hooks
>    for their own submissions
>
> Thanks,
> - Tom
>