Bill Allombert on Sun, 30 May 2004 22:56:13 +0200 |
[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]
Re: reporting the GP name of functions in error message |
On Wed, Mar 10, 2004 at 11:31:03PM +0100, Bill Allombert wrote: > Hello PARI-dev, > > I tried to implement reporting of the GP name of the function called > where an error occur in the new parser, but I found it could be done > the same way with the current parser, so here a patch. > > It is for evaluation purpose, it could be perfected. > > With this patch, we get for example: > ? cotan(0) > *** cotan: division by zero in R/R > > instead of > ? cotan(0) > *** division by zero in R/R > > or > ? for(i=2,100,factormod(x^4+1,i)) > *** factormod: composite modulus in mpsqrtmod: 9. > > instead of > > ? for(i=2,100,factormod(x^4+1,i)) > *** composite modulus in mpsqrtmod: 9. > > which is less confusing when it appear inside a large GP routine. Well, I have commited this patch. We might want to extend it to be able to report user functions name as well, like in ? f(x)=1/x ? f(0) *** f: division by zero in gdiv, gdivgs or ginv but I am not sure it is practical without a call stack, since e.g. ? f(x)=cotan(x) ? f(0) *** cotan: division by zero in R/R will not talk about f. Note that in library mode, you will always get traditional error messages. Cheers, Bill.