given the origin and complexity of the translate
facility, I am surprised that it works at
all on such items (like sum) with special
evaluation methods.
Good for you if you get it working..
RJF
Wolfgang Jenkner wrote:
> Raymond Toy <toy@rtp.ericsson.se> writes:
>
>
>>This also fixes things. However, when I load the translated file and
>>try harm_foo(1,10), I get a warning about |$i| being undefined with
>>CMUCL, but not Clisp. It does, however, return the correct value.
>>
>
> Does this warning come from
>
> harm_foo(l,u):=block(mode_declare([l,u],fixnum),
> sum(1/i,i,l,u));
> or rather from
>
> harm_bar(l,u):=sum(1/i,i,l,u);
>
> I am asking this because translating harm_foo only involves the trans1
> patch but not the the fcall patch, and the generated code is
> essentially a DO loop with |$i| bound by an init form, which should
> not give any trouble.
>
> Anyway, in the case where CMUCL gives the warning could you please
> check if the generated lisp code is essentially the same as the code
> generated by Clisp? (Sorry for asking you to do this but I have only
> Clisp installed.)
>
> Wolfgang
>