RE: [Axiom-developer] reproducibility of a segmentation fault when the compiler encounters a macro (with $lisp) badly indented

[Bill Page]

> Vanuxem Grégory <address@hidden> writes:
> 
> > If someone can try to compile, in a fresh new axiom session,
> > the attached file. I get a segmentation fault.
> > 
> > )abbrev domain MYIARR MyIndexedOneDimensionalArray
> > MyIndexedOneDimensionalArray(S:Type, mn:Integer):_
> >                  Exports == Implementation where
> >    Qsetelt ==> SETELT$Lisp
> >    Qnew ==> GETREFV$Lisp
> > 
> >   Exports == OneDimensionalArrayAggregate S with
> > 
> ... 

On August 19, 2006 1:00 PM Martin Rubey

> Yes it is. However, I think that the macro definitions should 
> all be indented by the same amount. I.e., try
> 
> )abbrev domain MYIARR MyIndexedOneDimensionalArray
> MyIndexedOneDimensionalArray(S:Type, mn:Integer):_
>                  Exports == Implementation where
>   Qsetelt ==> SETELT$Lisp
>   Qnew ==> GETREFV$Lisp
> 
>   Exports == OneDimensionalArrayAggregate S with
> ...

On August 19, 2006 1:15 PM Ralf Hemmecke wrote:
> 
> ... 
> Anyway that is a very good example why pile syntax is
> terrible.
> 

Ralf, I am sorry but your comment makes be rather angry. :(
I cannot understand why an otherwise very intelligent
programmer would blame the language for how a compiler
handles a syntax error. Arrrgh!

(: Of course I mean this a the spirit of friendly criticism... :)

In my opinion the notation without the {}; decoration is still
superior even when the compiler writers get it wrong. There are
several other instances when SPAD does not consistently implement
the language.

If you compile this in #pile mode in Aldor you get the expected
result.

http://wiki.axiom-developer.org/SandBoxPileNotation