Re: git: 1df431576f99 - main - sh/tests: Cut down builtins/read12.0 by 2 seconds

On Mon, May 25, 2026 at 09:30:36AM +0200, Dag-Erling Smørgrav wrote:
> Jilles Tjoelker <jilles@stack.nl> writes:
> > Dag-Erling Smørgrav <des@FreeBSD.org> writes:
> > > Please revert the non-functional parts of this change.
> > Please explain what you are worried about in the change. Is this about
> > false positives or false negatives?

> The purpose of a test is to reliably fail if the functionality it tests
> is broken.  Your change turns my test from one that is fairly certain to
> fail if the timeout functionality is broken to one that has a reasonable
> chance of passing regardless.  What's more, you justify the change with
> a logical argument so fallacious that you _still_ refuse to explain it.
> All that just to save yourself a couple of seconds once in a while, even
> after I showed you how to save much more than two seconds simply by
> enabling parallelism.  Please revert, or I will.

I have reverted the entire commit. Let's leave it as is until I
understand what the problem is.

-- 
Jilles Tjoelker