On Fri, May 16, 2008 at 10:50 PM, Bill Page <[EMAIL PROTECTED]> wrote: > Gaby, > > I wanted to you send a "thank you!" for the implementation of 'rep' > and 'per' in OpenAxiom. I frequently found the liberal and irregular > use of 'pretend' in other Axiom dialects to be confusing - so much so > that sometimes I had to resort to first writing code in Aldor to > benefit from the improved error messages and consistent use of 'rep' > and 'per' in order to get something to compile. Once it was working in > Aldor then I could manually translate it back to SPAD and be > reasonably confident that it would compile in SPAD also. > > The page: > > http://axiom-wiki.newsynthesis.org/SandBoxSum > > is the first time I have been able to do this directly in SPAD using > OpenAxiom. Consistent use of 'rep' and 'per' is a pleasure. And there > seem to have been other less obvious improvements to the compiler. > Thank you very much for this!
You're welcome. Kudos to Luke Wagner who kindly reported the problem and explained how painful to work around when he was working on a class project. It is good to see that you can now express things that were difficult or impossible to express before. I agree with your assessment about the use of `pretend'. I wish my students never heard of `pretend' :-) > > --------- > > Then I'll ask just one question about Union: > > (1) -> )sh Union(a:Integer,b:Float) > > Union(a: Integer,b: Float) is a domain constructor. > ------------------------------- Operations -------------------------------- > > ?=? : (%,%) -> Boolean ?case? : (%,a) -> Boolean > ?case? : (%,b) -> Boolean coerce : % -> OutputForm > construct : Float -> % construct : Integer -> % > ?.? : (%,b) -> Float ?.? : (%,a) -> Integer > ?~=? : (%,%) -> Boolean > > shows only 'construct' as an overloaded method for creating members of > the Union. > > (1) -> x:Union(a:Integer,b:Float):=2 > > (1) 2 > Type: Union(a: Integer,...) > > (2) -> y:Union(a:Integer,b:Float):=3.14 > > (2) 3.14 > Type: Union(b: Float,...) > > But suppose I wanted to create a Union with two copies of 'Integer'? This is impossible in all versions of AXIOM right now -- as documented in the Jenks&Sutor book, under the euphemism `can be difficult to work with'. Types in branches must not overlap. > This is a tagged union so this should be possible, right? But what > about the exports? Currently, it is not. And it can be frustrating -- almost a year ago, I spent an incredible amount of time chasing that problem. I did not have time to solve it, but I believe you should open an bug report for it. I know the implementation reasons why it is that way, but I don't think there is any `high level' justification for that limitation. > > (3) -> )sh Union(a:Integer,b:Integer) > Union(a: Integer,b: Integer) is a domain constructor. > ------------------------------- Operations -------------------------------- > > ?=? : (%,%) -> Boolean ?case? : (%,a) -> Boolean > ?case? : (%,b) -> Boolean coerce : % -> OutputForm > construct : Integer -> % construct : Integer -> % > ?.? : (%,a) -> Integer ?.? : (%,b) -> Integer > ?~=? : (%,%) -> Boolean > > There seem to be two operations named 'construct' with the *same* > signature! Surely one of these was intended to create an member of the > Union tagged 'a' and the other a member tagged 'b'. But how can I call > the correct one? > > (3) -> z:Union(a:Integer,b:Integer):=2 > > (3) 2 > Type: Union(a: Integer,b: Integer,...) > (4) -> z.a > > (4) 2 > Type: PositiveInteger > (5) -> z.b > > >> Error detected within library code: > (0 . 2) cannot be coerced to mode (Integer) > > (5) -> z.b := 2 > > The form on the left hand side of an assignment must be a single > variable, a Tuple of variables or a reference to an entry in an > object supporting the setelt operation. > > (5) -> z:Union(a:Integer,b:Integer) := [b==3] > > There are no library operations named OPTARG > Use HyperDoc Browse or issue > )what op OPTARG > to learn if there is any operation containing " OPTARG " in its > name. > > Cannot find a definition or applicable library operation named > OPTARG with argument type(s) > Variable b > PositiveInteger > > Perhaps you should use "@" to indicate the required return type, > or "$" to specify which version of the function you need. > > ------- > > How can I create a member tagged with 'b'? > > In panAxiom I can write: > > x:Product(Integer,Integer) := [1,2] > > With the dual construction of 'Sum' shown at the link above, I would > also like to be able to write: > > x:Sum(a:Integer,b:Integer) := ? > > and have it populate the correct copy of Integer. In Aldor I can use > the "keyword arguments" like: > > x:Sum(a:Integer,b:Integer) := [b==3] > > or equivalently 'union(b==3)' > > http://www.aldor.org/docs/HTML/chap13.html#8 > > but so far I have not found anything equivalent in any flavor of Axiom. > > Maybe 'setelt' should be exported by Union? > > I wonder if the reference to "OPTARG" in the error message shown above > indicates that there is some incomplete support for this somewhere in > the Interpreter? When you wrote [b == 2], the interpreter understood that you were passing in value for an optional argument -- much like value for `keyword parameter' in some other programming languages. I don't like much that double interpretation of `==' which depends on whether '==' appears at toplevel or inside expression. I would rather use a different symbol for `keyword argument', than a fuzzy interpretation of '=='. (To be honest, as much as I dislike that notation for default values, I found myself extending Boot with that notation in order not to fallback to Lisp. I tried some other symbols but it would have required me to do an extensive modification to the parser for the local problem I wanted to solve.) > > Any ideas about this? > It is a limitation documented in the Axiom Book. Please fill a `feature request'. -- Gaby ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ open-axiom-devel mailing list open-axiom-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/open-axiom-devel
