[Os-project-managers] optional doubles with no default values

[Jun Ma]

I read this thread of emails [ optional doubles with no default values]
In principle:
1. NaN is a number which is different than all other number. Of course, any 
one number is different from others
2. Think of NaN as the quotient of 1/0.
3. In Java and most programming languages, NaN is a double, which is usually 
considered as equivalent of a number.
4. NaN as a default value is of course different from no value. No 
interpretation is to be made on "no value"
5. Usually in any implementation, no value means null.
But since null is for objects, for primitive types like int and double in 
Java, which cannot be null, it's than up to the particular implementation to 
be designate  a specific value which will never be used or different from 
all practical values.

Of course, the optimization services elements can overwrite the rules by 
either schema technology or English annotation, such as we use for most of 
the sparse technologies, where we mean 0 if an element is not .

Given the above principles, with no OS-specific interpretations for maxTime:
<maxTime/> should mean not specified!

Jun
--------------------------------------------------
From: "Horand Gassmann" <Horand.Gassmann at dal.ca>
Sent: Thursday, January 27, 2011 4:52 PM
To: <os-project-managers at list.coin-or.org>
Subject: [Os-project-managers] optional doubles with no default values

> Hi guys,
>
> following the talk today, I started cataloguing all the doubles in our
> schemas that do not have default values (and thus can be NaN). In OSoL
> I found five:
>
> <xs:complexType name="InitVarValue">
> <xs:complexType name="InitObjValue">
> <xs:complexType name="InitConValue">
> <xs:complexType name="BranchingWeight">
> <xs:complexType name="MaxTime">
>
> InitVarValue we talked about, and it seems that the next two should be
> treated the same way by analogy. However, I have my doubts about
> BranchingWeight and especially about MaxTime. Suppose I say
>
> <maxTime/>
>
> what should happen? Maybe I am not interpreting this one properly; all
> the others appear as attributes, this one is element content. Perhaps
> the above construct is actually not allowed by the schema.
>
> Any thoughts?
>
> Cheers
>
> gus
>
>
>
>
> _______________________________________________
> Os-project-managers mailing list
> Os-project-managers at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/os-project-managers
>