[Ecls-list] Status of CVS

Gabriel Dos Reis gdr at integrable-solutions.net
Mon May 12 16:20:06 UTC 2008

On Mon, May 12, 2008 at 10:27 AM, Juan Jose Garcia-Ripoll
<jjgarcia at users.sourceforge.net> wrote:
> On Mon, May 12, 2008 at 4:53 PM, Waldek Hebisch
>  <hebisch at math.uni.wroc.pl> wrote:
>  >  I do not understand your argument.  AFAICS if code is correct it should
>  >  run both in sbcl and in other Lisps.  If code is wrong than implementations
>  >  which insert checks nicely report error while other Lisps crash or run
>  >  code ignoring problem.  Ignoring problems seem to be in Lisp spirit,
>  >  but it makes finding bugs much more difficult.
>  But this is what I mean. There may be code around that relies on those
>  automatic checks for functioning properly, expecting that errors will
>  be signaled when the functions are passed the wrong arguments.

Let me be clear.  The case of Axiom I reported was NOT relying
on the check being inserted.  Rather, what I'm saying is that
SBCL's insertion of the checks *helped* a lot in porting Axiom
to SBCL, CLisp and ECL.  Said differently, now, OpenAxiom
and FriCAS don't need the check for proper behaviour, but before
getting there, the checks were valuable as debugging tool.

>  >  Concerning programmer written check forms: to require them for debugging
>  >  looks backwards for me
>  You are twisting what I say. What I said is that writing code that
>  relies on type declarations to find errors, expecting that the lisp
>  will signal them, is a bad design.

But then, that has no relevance to my message you replying to
earlier.  What I said earlier amounted to an appraisal of your
proposed behaviour because now it would make life easier for
debugging mysteries with the ECL compiler disappearing on
faulty codes.

-- Gaby

More information about the ecl-devel mailing list