|Subject:||Re: [Axiom-developer] merge branches/daly to trunk|
|Date:||Fri, 08 Jun 2007 10:47:43 +0200|
|User-agent:||Opera Mail/9.21 (Linux)|
But Christian might be better in saying something why AxiomUnit does not yet exist.
it is actually quite some time since I attempted to get AldorUnit to Axiom. As the crucial parts af the discussion happen orally between me and Ralf, parts of the arguments are lost.
The most important aspect is improving the Axiom experience--during the invostigation of Axiom, it turned out that the most important improvements AldorUnit provides cannot be brought to Axiom. Having a full blown AxiomUnit, one would have tools to debug segfaults (although they do not seem to be an issue for Axiom) and other really bad behaving code. One should get really usable output from the Testing environment. Ease of writing tests and its integration into Axiom is another important issue.
Aldor's TextWriter abilities where crucial for providing usable textual output. These possibilities are not present in Axiom -- but it seems Martin did some work in this direction.
For evaluating the possible benefits of AxiomUnit I played around with using the Aldor compiler for Axiom and loading the resulting code. Although I do not have the details at hand, I managed to crash Axiom badly without writing evil-looking Aldor code. Axiom behaved unexpectedly when loading functions compiled with Aldor. I remember race conditions when loading libraries. Recovery from errors proofed to be a serious issue. All in all, my experience of using the Aldor compiler with Axiom wasn't the best. I do not want to spread FUD, it's just some time off. So the details faded a bit.
However, I will have a look at Martin's work. Maybe this will help in finally providing AxiomUnit.
Kind regards, Christian
|[Prev in Thread]||Current Thread||[Next in Thread]|