Re: [Axiom-developer] [Q] Function for testing for a partof the output?

[William Sit]

Bill's solution of course is mathematically sound if what one is looking for is
a kernel, and is preferrable. However, if one really wants to check some strings
embedded, then the following would work:

match?("*sss*",tex(ooo::TEX).1,char("*"))

where sss is what you are searching for (perhaps "texified", like \%\% for %%),
and ooo is the output of a computation. One advantage is we do not have to know
the structure of the output and so can be applied uniformly to all outputs.

There seems to be a bug with match (without the ?), which is supposed to return
the index if the string match succeeds. But

match("*to*","yorktown",char("*"))
  
   >> System error:
   Wildcard must be a character

protected-symbol-warn called with (NIL)

Also, the interpreter will not coerce "*" to a character.

William
---

"Page, Bill" wrote:
> On Thursday, February 10, 2005 4:58 PM Vladimir Bondarenko wrote:[
> 
> > Please help also about the following where I have tried
> > several variations but without result
> >
> > -> member?(integrate,kernels(integrate(erf(z)*sin(z)^2,z)))
> >
> >   false
> 
> Try this:
> 
>   is?(integrate(erf(z)*sin(z)^2,z),integral)
> 
> which tests that the first expression is a kernel with
> the operator integral.
> 
> > while I'd like to have a test1 working like this
> >
> > -> test1(integrate, integrate(erf(z)*sin(z)^2,z))
> >
> >  true  -- Currently, AXIOM cannot get this integral and
> >         -- returns it unevaluated
> 
> Perhaps this is a little more accurate:
> 
>   not empty? [i for i in kernels integrate(erf(z)*sin(z)^2,z)
>     | is?(i,integral)]
> 
> which will be true if at least one integral remains.
> 
> > and
> >
> > -> test2(%%, integrate(1/(a+z^5), z))
> >    true  -- Currently, AXIOM returns the terms like  %%ID1
> 
>   not empty? [i for i in kernels integrate(1/(a+z^5), z)
>     | is?(i,rootOf)]
> 
> > Let me thank you cordially for your valuable help in advance.
> 
> Of course you are most welcome!
> 
> Regards,
> Bill Page.