Re: Some X++ construction issues.

[Michael McClary <michael>]

> 	 - Should pseudoconstructors be static member functions?
> No.
> 
> As it is pseudoconstructors don't have to be declared in the class thet
> they construct.  Often we have a pseudoconstructor declared with some
> overlaoading in the class that defines the parameters for the overloading.
> This is a good thing.  We also have a pseudoconstructor for an abstract
> class defined in the concrete class that actually instantiates it.

Sounds good.  (Though it could be argued that the .hxx defining parameters
for the overloading could be included by the class, it can be counter-
argued that this allows the overloading to be >added< to an existing
class pulled from an immutable library.)

> This is a good thing, why change it.

I thought I made it clear I wasn't proposing changing it.  I wanted to
examine whether there was anything to be gained or lost by changing it.
Looks like this is a case where conatainment barriers hinder, rather
than aid, the programming effort, so function would be lost.

(And now, when a developer, holding his copy of the ARM open to page 266,
asks us "Why did you do THAT?", we can say "Because <insert quote from
roger's paragraph here>".  Sounds much better than "Because we didn't
think of it at first, then didn't want to spend the time changing it just
for neatness."  B-) )

	michael