From: "Randy Brukardt" <randy@rrsoftware.com>
Subject: Re: Five Years After
Date: Wed, 24 Jan 2018 21:09:00 -0600
Date: 2018-01-24T21:09:00-06:00 [thread overview]
Message-ID: <p4bhod$r9e$1@franka.jacob-sparre.dk> (raw)
In-Reply-To: 292710c7-3520-406f-9e6b-1fb030ada211@googlegroups.com
"Maciej Sobczak" <see.my.homepage@gmail.com> wrote in message
news:292710c7-3520-406f-9e6b-1fb030ada211@googlegroups.com...
>> Also having a validated compiler doesn't seem to matter as much as it did
>> in the last century. ;-)
>
>Interestingly, within the safety-critical domain, "validation" (or
>"qualification", which
> is a more specialized term) means more than ACATS tests can offer, and
> outside
> of this domain nobody cares about it anyway. In some sense, this effort,
> even
> though very well-intended and representing an objectively good engineering
> practice, was simply misplaced.
Everyone agrees that the ACATS test suite offers a lot of value to both
implementers and users (implementers because an independently sourced and
trustworthy set of tests is very helpful to creating and maintaining an Ada
compiler -- it is especially good IMHO in preventing errors of omission,
which are often the hardest to find; users because the test suite provides a
modicum of portability; Ada compilers accept essentially the same language
since they've all been tested with the same set of tests). This certainly is
true since I'm paid money every year to maintain and enhance the test
suite -- that wouldn't be the case if people didn't see a value.
The value to formal testing ("conformity assessment") is substantially less,
especially nowdays when machines are fast enough that anyone can run the
ACATS in a few hours. That is, you can do your own testing in a day or two
to prove that your vendor is actually telling the truth about conformity.
Why would you want to pay some formal testing $$$ to do what you can do
yourself?
Thus, I've been spending some of the ACATS effort on making it easier to
self-test. Using the new(ish) grading tools and some other tools that I'll
be posting shortly as "submitted tools", one can essentially test a GNAT
implementation against the ACATS without a lot of hand work. (You do have to
understand enough about your implementation and your target to configure the
ACATS itself.) The hope is to get the value of independent testing without
the costs.
Of course, formal third-party testing is still available if someone wants
it.
Randy.
next prev parent reply other threads:[~2018-01-25 3:09 UTC|newest]
Thread overview: 40+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-01-13 12:36 Five Years After Jeffrey R. Carter
2018-01-13 16:45 ` Jacob Sparre Andersen
2018-01-13 17:22 ` gautier_niouzes
2018-01-13 19:45 ` Nasser M. Abbasi
2018-01-13 21:15 ` Jeffrey R. Carter
2018-01-14 19:13 ` Luke A. Guest
2018-01-23 4:35 ` Robert Eachus
2018-01-23 5:23 ` Luke A. Guest
2018-01-23 6:26 ` bozovic.bojan
2018-01-24 13:16 ` Nasser M. Abbasi
2018-01-24 13:39 ` Dmitry A. Kazakov
2018-01-25 2:44 ` Randy Brukardt
2018-01-24 13:50 ` Simon Clubley
2018-01-24 13:55 ` Björn Lundin
2018-01-24 19:07 ` Simon Clubley
2018-01-24 15:12 ` Mehdi Saada
2018-01-24 16:04 ` Shark8
2018-01-24 19:06 ` Simon Clubley
2018-01-24 20:34 ` Lucretia
2018-01-25 0:29 ` Simon Clubley
2018-01-25 20:14 ` Lucretia
2018-01-26 21:25 ` Shark8
2018-01-27 0:07 ` Dennis Lee Bieber
2018-01-27 1:52 ` Nasser M. Abbasi
2018-01-25 0:43 ` Robert Eachus
2018-01-25 2:51 ` Randy Brukardt
2018-01-24 16:05 ` Dennis Lee Bieber
2018-01-24 17:07 ` Robert Eachus
2018-01-24 18:43 ` Dennis Lee Bieber
2018-01-24 17:55 ` Jeffrey R. Carter
2018-01-24 18:44 ` Dennis Lee Bieber
2018-01-24 20:17 ` Lucretia
2018-01-25 0:14 ` Simon Clubley
2018-01-23 21:57 ` Maciej Sobczak
2018-01-24 18:14 ` G. B.
2018-01-25 3:09 ` Randy Brukardt [this message]
2018-01-25 21:09 ` Maciej Sobczak
2018-01-25 23:07 ` Robert Eachus
2018-01-26 4:28 ` Randy Brukardt
2018-01-24 20:41 ` Mehdi Saada
replies disabled
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox