[MOBY-dev] [moby] Re: Suggestion for newtag in Parameter(Secondary Input specification)

[Paul Gordon]

Sure.

>Hi Paul, 
>
>Could you collect together the various ideas that have come up in this
>thread and present a formal proposal?  
>
>Cheers! 
>
>M
>
>
>
>On Tue, 2006-02-28 at 08:09 -0700, Paul Gordon wrote:
>  
>
>>Given that the size of the human genome exceeds the limits of a xsd:int, 
>>and that we often use e-values exceeding the limits of xsd:double, may I 
>>suggest that we mandate the equivalents to xsd:integer and xsd:decimal?  
>>I have built my jMOBY code using arbitrary precision numbers, but I'm 
>>sure we'll need to change other parts of it, and of the Perl libraries...
>>
>>    
>>
>>>I think we should choose the corresponding XML Schema types for each 
>>>primitive object. For instance:
>>>
>>>moby:String -> xsd:string (it would not allow XML content) or 
>>>xsd:anyType (any content)
>>>
>>>moby:Integer -> xsd:int [-2147483648,2147483647] or xsd:integer (*any* 
>>>integer)
>>>
>>>moby:Float -> xsd:double (IEEE double-precision 64-bit floating point 
>>>type), xsd:float (IEEE single-precision 32-bit floating point type) or 
>>>xsd:decimal (*any* real number with a finite number of decimal digits)
>>>
>>>moby:DateTime -> xsd:dateTime
>>>
>>>moby:Boolean -> xsd:boolean or an enumerated xsd:string 
>>>{'0','1','false','true'}
>>>
>>>	Best Regards,
>>>		José María
>>>
>>>Paul Gordon wrote:
>>> 
>>>
>>>      
>>>
>>>>The main problem:
>>>>
>>>>If somebody specifies a float, can I legally submit a e-value cutoff 
>>>>like "1.0e-180" (i.e. are we going to assume bit capacity, such as 
>>>>2^-149 for 16-byte IEEE floating point, or are we supporting arbitrary 
>>>>precision)? Underflow and overflow can cause problems on many 
>>>>systems...  same thing for integers > 2^32...
>>>>
>>>>Actually, do we support scientific notation?  That isn't mentioned either.
>>>>
>>>>Cheers,
>>>>Paul
>>>>
>>>>P.S. Yes, you are right Pieter.  You could enumerate integers, or even 
>>>>floats for that matter: this distinction matters for a server with 
>>>>strong types, but not for the client.  I've been too client centric 
>>>>lately :-)
>>>>
>>>>   
>>>>
>>>>        
>>>>
>>>>>On Mon, 2006-02-27 at 23:44 +0100, Pieter Neerincx wrote:
>>>>>
>>>>>
>>>>>
>>>>>     
>>>>>
>>>>>          
>>>>>
>>>>>>Hmmm I use enum for some integers as well. I think it's perfectly  
>>>>>>normal to say for example: enum [1,2,4,8].
>>>>>>  
>>>>>>
>>>>>>       
>>>>>>
>>>>>>            
>>>>>>
>>>>>Perhaps Paul can clarify what problem he is trying to solve.  My
>>>>>instincts tell me that maybe he is having difficulty with casting un-
>>>>>typed XML blocks as either Integer or String, as appropriate... is that
>>>>>a correct interpretation of the problem Paul?  
>>>>>
>>>>>I think the combination of the <datatype> block and the <enum> block
>>>>>should be able to indicate whether the ENUM is of type String or of type
>>>>>Integer (or Float, or whatever).  Is that not sufficient?  Or am I
>>>>>misunderstanding what the root of the problem is?
>>>>>
>>>>>M
>>>>>
>>>>>
>>>>>
>>>>>_______________________________________________
>>>>>MOBY-dev mailing list
>>>>>MOBY-dev at biomoby.org
>>>>>http://biomoby.org/mailman/listinfo/moby-dev
>>>>>
>>>>>
>>>>>     
>>>>>
>>>>>          
>>>>>
>>>>_______________________________________________
>>>>MOBY-dev mailing list
>>>>MOBY-dev at biomoby.org
>>>>http://biomoby.org/mailman/listinfo/moby-dev
>>>>
>>>>
>>>>   
>>>>
>>>>        
>>>>
>>> 
>>>
>>>      
>>>
>>_______________________________________________
>>MOBY-dev mailing list
>>MOBY-dev at biomoby.org
>>http://biomoby.org/mailman/listinfo/moby-dev
>>    
>>