From: Austin Ziegler <Austin.Ziegler@...>
Date: 2004-06-15T02:27:57+09:00
Subject: Re: [ANN] celsoft.com/Battery 0.1.1

Sean O'Dell [mailto:sean@celsoft.com] :
> On Monday 14 June 2004 08:14, Austin Ziegler wrote:
>> Sean O'Dell [mailto:sean@celsoft.com]:
>>> When you're given no choice, as far as you're concerned as a
>>> library user, they're run in an arbitrary order. [...]
>> This is a different claim than you made initially.
> I learned that tests were run alphabetically, but originally I
> thought they were run randomly, according to the order Ruby
> returned them (which I think was actually correct, if I read
> Nathaniel right). Either way, I felt the ordering was arbitrary. I
> didn't say random because I wasn't sure it was random, I just knew
> that the order was, as far as I was concerned, in some arbitrary
> order (decided not by me for reasons so vague I don't even know
> them and can't interpolate them from the docs).

This does not address the fact that it is a different claim than you
made initially.

>>> As far as I'm concerned, the order that tests are run is a black
>>> box.
>> As they should be. Tests should be independent of one another and
>> not depend on any particular ordering.
> I run my tests in any order I want. I suspect Test/Unit will also
> be allowing controlled test orders, so you might want to try and
> convince Nathaniel it's a bad idea before he implements it. You
> might be able to stop him, but my library isn't changing.

You may want to take a less arrogant tone; you're starting to stink
things up around here since you obviously cannot or will not read
what people have actually written in your haste to defend yourself
against attacks that don't exist. It also decreases the interest in
your library given your inept behaviour over a simple, very polite
request of someone who felt that you misspoke.

What I said was -- very clearly -- TESTS should be written to be run
in any arbitrary or random order. I didn't suggest that a testing
framework shouldn't be able to specify a particular test running
order. Frankly, I would love to be able to do:

  Test::Unit::Runner.order = :random

And *force* the tests to be run in a random order to ensure that I
am not introducing dependencies between test cases.

-austin
--
austin ziegler * austin.ziegler@evault.com