Re: [Axiom-developer] merge branches/daly to trunk

[Waldek Hebisch]

Gabriel Dos Reis wrote:
> On Mon, 4 Jun 2007, Martin Rubey wrote:
> 
> | "Bill Page" <address@hidden> writes:
> | 
> | > As far as I know, the only thing in silver (aka. axiom.daly,
> | > aka. axiom.trunk) is a first attempt at some regression testing.  I 
> believe
> | > that Gaby is trying to port this downstream [...]
> | 
> | I think it is a mistake to do that.  We should port Christian Aistleitner's
> | code to use libaxiom instead of libaldor.
> 
> Since I do not have resource to allocate for that task, should I go with
> no form of regression test at all?  
> 

I have posted some time ago scripts which help regression testing.
I must admit that I consider technique applied in current silver as
step backwards:  AFAICS results of running input files are still
compared using text comparison, but since expected result is
embedded in .input files we have extra maintenance burden when
output changes (I use separate tree for comparison).

Concerning regress script: using md5sums is quite crude, there is
a lot of differences which give functionally identical result.
One of the scripts I posted is a Lisp comparator -- I would be
happy if somebody provied a better one, but ATM it is _way_
better then byte comparison/md5sums.

-- 
                              Waldek Hebisch
address@hidden