[boost] Re: Including abs() in rational.hpp

From: jsiek_at_[hidden]
Date: 1999-12-10 17:37:42

I suppose when the compilers all get up to snuff this won't
be a problem, but it currently is a headache. The "using std::abs"
breaks under certain conditions with a compiler from
a town in the pacific northwest.

Gabriel Dos Reis writes:
 > jsiek_at_[hidden] writes:
 > 
 > | So the letter of the law makes it illegal, but was this really the
 > | intention? Seems to me this causes nasty problems for generic
 > | libraries.
 > 
 > Can you show any nearly real-world problem in our specific case?
 > 
 > | ...  Since std::abs() is not all-applicable, it forces
 > | the *user* to do "using std::abs;" etc, for every operation
 > | of this nature that is used in the algorithm. 
 > 
 > No. The user will write std::abs() only when he knows the
 > argument type comes from ::std or is defined by the language.
 > 
 > In an hypothetical ultra-generic case, he'll count on Koenig
 > lookup. That is why that rule was put in in the first place.
 > 
 > -- Gaby
 > 
 > ------------------------------------------------------------------------
 > GRAB THE GATOR! FREE SOFTWARE DOES ALL THE TYPING FOR YOU!
 > Gator fills in forms and remembers passwords with NO TYPING at over
 > 100,000 web sites! Get $100 in coupons for trying Gator!
 > http://clickhere.egroups.com/click/2092
 > 
 > -- 20 megs of disk space in your group's Document Vault
 > -- http://www.egroups.com/docvault/boost/?m=1
 > 
 > 
 >