[Axiom-developer] Re: AxiomUnit

[Bill Page]

On 6/12/07, Christian Aistleitner wrote:
> On Fri, 08 Jun 2007 15:35:00 +0200, Bill Page wrote:
>
> > On 6/8/07, Christian Aistleitner wrote:
> >>
> >> ...
> >> > Given that, what remains for AxiomUnit?
> >>
> >> As we've already discussed in private mail some time ago, there
> >> are more things. But I suggest to switch to private communication
> >> again, as I do not want to abuse the Axiom Mailing list for third
> >> party projects.
> >
> > [...], but I am very confused as as to why you
> > would prefer private communication about a subjec that is
> > important to so many other Axiom users.
>
> If I'd know that AldorUnit could be nicely ported to Axiom and if I'd
> also know that this approach would be the only valid one, I'd probably
> discuss the whole process on the mailing list. However, I've been told
> that currently there are some scripts around to test Axiom code—
> which one is the best?

Right now there is only a very rudimentary implementation of regression
testing that is part of what will become Axiom Silver. There is currently
no systematic method for doing unit testing in Axiom. So I would strongly
encourage you (Do I need to plead with you?) to please discuss AldorUnit
on this list.

> I assume you would not want full length discussion about every single
> issue of all such scripts on the list. AxiomUnit would not automatically
> be better.

Clearly one purpose of discussion is to consider what approach might
be better, but in open source projects it seems that one seldom
actually reach consensus. Instead sooner or later one (or several)
developers get tired of the discussion and find the motivation to just
strike-off in one of the directions discussed. Sometimes other developers
follow. I know that sounds like a crazy unworkable development model
but it is hard to argue with the success of a lot of other open source
projects...

> To me AxiomUnit should be developed as an add-on library to Axiom.
> Along with tools to facilitate writing test (suites/cases/s) and executing
> them. If later on people decide to integrate them into Axiom, it's fine.

I agree.

> It's similar to other add-on libraries of Axiom.

I agree.

> I suppose, you do not want to have the discussion of every Axiom
> add-on library on the list ;)

Well I do not wish to limit discussion in any manner. If there is
discussion of every Axiom add-on library, then the result is a
potentially rich source of documentation about these aspects of
Axiom.

> > What do you mean by "third party projects"?
>
> AldorUnit has been developed for Aldor. But it is not entangled with
> Aldor.org in any way. AldorUnit is coming from RISC, which is not
> entangled with Aldor.org as far as I can tell. Ralf might chime in here,
> if this not accurate.

I think that is a good thing. If there are issues relating to rights that
should be attributed to RISC or any other organization then of course you
should be specific about the license and copyright that you associate
with your work.

> By "Third party project" I meant a project not endorsed or encouraged
> by Axiom.

Since Axiom is a collaborative open source project I think it is quite
hard to define exactly what is and is not "endorsed or encouraged by
Axiom". Essentialy what is encouraged is that you and other people
contribute to this project in what ever way you find appropriate to you.
In return about the only thing that the Axiom project can promise is that
there will be a large number of other people who will be very grateful for
your work.

Regards,
Bill Page.