Re: [Axiom-developer] Re: root chunks

[Waldek Hebisch]

Ralf Hemmecke wrote:
> On 06/22/2007 07:56 PM, Waldek Hebisch wrote:
> > Ralf Hemmecke wrote:
> >> I am pretty sure that someone can confirm that gcc has an option 
> >> that compiles the #line information into the executable. I am not 
> >> so sure about gcl. Camm?
> >> 
> >> You don't need to debug with sbcl. If a bug happens via SBCL and 
> >> not via GCL then that is a bug in the compiler and not in the 
> >> algebra code. So you would not be interested in debugging unless 
> >> you are a compiler writer.
> 
> > Sorry, that very naive:  Axiom in many places makes unportable 
> > assumptions.   So, in most cases when code works in one Lisp but 
> > fails in another the code is wrong.  FYI SBCL caught a number of bugs
> >  that were hidden when running under gcl (some of them were likely to
> >  give you "memory may be damaged" message).
> 
> Do you really mean that you had to change the *algebra* code? I did not
> speak of anything else. That the lisp/boot/shoe code is something else,
> I completely understand. And I also understand that for such issues
> different compilers are good. (Sorry for the gap that I left in my
> statement.)
> 

I was thinking mostly about boot code.  But SBCL helped also debugging
algebra -- because ATM "safe" gcl build takes much longer than safe
SBCL build.  In case of issue 359 it seems that SBCL catches
error earlier than "safe" gcl.  Issue 359 seem to be at boot
level, but already illustrates that it is hard to separate algebra
from the rest of the system.  And while I do not have example
of algebra problem that shows in SBCL but do not show in 
"safe" gcl I belive that such problems are possible and we
will hit such problem sooner or later.


-- 
                              Waldek Hebisch
address@hidden