Re: Stochastic bisection support

From: Jan Kara <jack@suse.cz>
Date: Nov 22, 2021, 11:57 UTC
Message-ID: <20211122115740.GA24453@quack2.suse.cz>
In-Reply-To: <CAPx1GvfT36SvJ6Lwf1-2KUebVXkCkNvYTUw=FU+dHBy76VN5RQ@mail.gmail.com>

On Fri 19-11-21 23:54:12, Chris Torek wrote:

Show 19 quoted lines

> On Fri, Nov 19, 2021 at 8:51 AM Johannes Schindelin
> <Johannes.Schindelin@gmx.de> wrote:
> > An interesting problem, for sure!
> >
> > It is slightly related to a scenario that has been described to me
> > recently: in a gigantic project whose full test suite is too large to run
> > with every Pull Request, where tests are more likely to become flaky
> > rather than simply break, a stochastic CI regime was introduced where a
> > semi-random subset of the test suite is run with every CI build. That team
> > also came up with the concept of attaching confidences as you describe.
> >
> > I only had time to look at the first patch closely so far. I hope to find
> > more time next week to review further.
> 
> I only scanned for obvious items myself, but the idea of a
> probabilistic test is indeed interesting.
> 
> I do wonder why you (Jan Kara, not Dscho :-) ) used fixed-point
> arithmetic here though.

Ah, very good question. I guess because I'm primarily Linux kernel programmer and we don't have floating point arithmetics in the kernel so I'm used to fixedpoint :). Secondarily because fixedpoint arithmetics provides me more control over where I'm loosing precision but looking at this now I agree the ugliness in the code is not probably worth the imagined win. I'll rewrite stuff using floating point. Thanks for the suggestion.

								Honza

-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR