Re: First draft standard text
JohnPryce wrote:
P1788
On 29th November I sent, to the P1788 officers, a first draft of text
of the standard, roughly as far as our motions have taken it. I asked
for their comments. These were not too unfavourable, so I think it is
worth circulating the same version to the whole P1788 group, which I
do hereby.
Your comments please on any aspect of the text that you feel relevant:
- Content, and agreement (or not) with substance of motions that have
passed.
John,
Just a cursory note about section 5.3 Exception Handling:
In the original Motion 8.02 text, I intentionally did not specify that bare
intervals, bare decorations, or decorated intervals were _types_. Although
an implementor might choose to treat them as such, another practical option
exists. An implementor may represent them as _states_, instead.
For example, an implementor may choose to define bare interval [a,b] as a
basic type; but bare decoration as a _state_ of a bare interval when a or b
is NaN.
The difference is subtle but important, since thinking of the distinctions
in terms of states rather than types reduces the overall number of types
required by the implementation while at the same time providing all the
necessary exception handling functionality.
Also, thinking in terms of states makes it trivial for implementor to add
support for Kaucher arithmetic, since the improper interval [b,a] is simply
another _state_ of a bare interval, not a new _type_.
Nate