Re: [Axiom-developer] Fwd: [M#73697383] Re: Disk-quota Request

[Ralf Hemmecke]

On 09/15/2006 04:35 AM, root wrote:
> > > But look at that file: src/input/mapleok.input.pamphlet
> > > -rw-------   1 hemmecke hemmecke 234633 2006-09-13 10:03 
> > > mapleok.input.pamphlet
> > >
> > > Who is going to read that???
> > I think this is a good example of something that belongs elsewhere
> > if we agree that the purpose of this archive is source code control.
> > I think keeping these sorts of documents on the Axiom Wiki (or the
> > Axiom Portal if it is personal research), makes better sense than
> > mixing it into the source. I understand Tim's purpose in associating
> > it with the "input" files since these are the closest thing that
> > Axiom has to actual test files.
>
> mapleok IS an input file. I just took the time to document the
> expected results so you could tell if you got the right answer.
> It is partially on the way to automating regression testing.

Do you really believe that somebody is going to read that file? Even if 
you documented it? I won't I am not a computed that checks whether the 
output that appears on my axiom session agrees byte by byte with what is 
in the .input file. I would have to count at least 100.000 bytes by 
inspection. I guess even if a computer can make errors. In such a stupid 
procedure I am far worse.

> Testing files are clearly part of the source control pile.

Right, but output should be checked automatically. And isn't it better 
to compare the actual Axiom data structures than the OutputForm result?

> Especially since they are run as part of the source code build.

Maybe you should look at Christian's AldorUnit. I think it's much better 
than just inventing a new test environment.

http://www.risc.uni-linz.ac.at/software/aldor/aldorunit

Ralf