> Bill Page wrote: >> Normally I would just define local functions: >> >> rep(x:%):Rep == x pretend Rep >> per(r:Rep):% == r pretend % >> >> but in this case I get the message: >> ... >> compiling local rep : $ -> Union(acomp: A,bcomp: B) >> SUM;rep is replaced by x >> Time: 0.14 SEC. >> >> compiling local per : Union(acomp: A,bcomp: B) -> $ >> Semantic Errors: >> [1] per: cannot pretend r of mode (Union (: acomp A) (: bcomp >> B)) to mode $ >> >> ****** comp fails at level 1 with expression: ****** >> error in function per >> >> --- >> >> I don't understand this error message from FriCAS. Perhaps this is >> a bug that has been corrected in OpenAxiom? >>
On Thu, May 14, 2009 at 9:58 PM, Waldek Hebisch wrote: > > Well, the compiler specifically checks if pretend is from a union > to something which is no a union and rejects such pretends. I can > only guess that original Axiom developers considered such pretend > to be too dangerous. In fact, union has rather special representation, > so most of the time such pretend would be wrong. OTOH compiler > contains no such check for records and records have the same > problem with possibly different representation... > But you would agree I think, that the use of pretend in the context of rep and per are in fact type-safe, right? Could we simply remove this check from the compiler? Gaby: Is that what was done in OpenAxiom? > Concerning Sum, the following (which is just a tiny part of Sum) > compiles fine: > > )abbrev domain SUM Sum > Sum (A:SetCategory,B:SetCategory) : C == T > where > C == SetCategory with > selectsum : % -> Union(acomp:A,bcomp:B) > ++ selectsum(x) \undocumented > in1 : A -> % > ++ makefirst(a) \undocumented > in2 : B -> % > ++ makesecond(b) \undocumented > > T == add > > --representations > Rep := Union(acomp:A,bcomp:B) > import Rep > > rep(x:%):Rep == x :: Rep > per(r:Rep):% == r :: % > > Note that currently in FriCAS (like in original Axiom) Rep has > magical properties, in particular it is enough to use '::' to > convert between Rep and %. > I don't understand this use of :: It does not seem to me that coercion should be involved in the definition of rep and per. By "magical" do you really mean "inconsistent"? ;-) Regards, Bill Page. ------------------------------------------------------------------------------ Crystal Reports - New Free Runtime and 30 Day Trial Check out the new simplified licensing option that enables unlimited royalty-free distribution of the report engine for externally facing server and web deployment. http://p.sf.net/sfu/businessobjects _______________________________________________ open-axiom-devel mailing list open-axiom-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/open-axiom-devel
