Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: P1788/D9.6 -- the recirculation ballot



On 2015-04-13 20:01:11 +0100, John Pryce wrote:
> On 13 Apr 2015, at 16:17, Vincent Lefevre <vincent@xxxxxxxxxx> wrote:
> > In the first paragraph of §9.7.2 (page 30 line 22):
> > 
> >  It is called unsigned if the sign is absent.  A positive-natural
> >  literal is an unsigned integer literal whose value is not zero.
> > 
> > I wonder whether "unsigned" should be changed to "natural"
> 
> It's a marginal point, but yes, done. It now reads
> "It is a natural-number literal if the sign is absent. A positive-natural
> literal is a natural-number literal whose value is not zero."

OK.

Now, I've found a more serious error. There's an inconsistency:

Page 31 line 8 (in §9.7.4): "r is empty or is a non-negative integer
literal ulp-count"

Since only "integer literal" has been defined previously, I read this
as: r is an integer literal whose value is non-negative, i.e. larger
or equal to 0. For instance, "+17".

But Table 9.5 says:

    natural         {decDigit} +
[...]
    radius          {natural}
    uncertIntvl     {sign} ? {decSignificand} "?" {radius} ? {dir} ?
                    ( "e" {integerLiteral} )?

i.e. {radius} is a natural-number literal (a sign is not allowed).

I think that the best fix is to disallow the sign (in all the examples,
there is no sign). Thus in page 31 line 8, one should have:
"r is empty or is a natural-number literal ulp-count".

To be clear: r = 0 is allowed, isn't it?

A suggested editorial change: page 31 line 7 in "m ? r u E" and line 8
in "u is empty or...", I think that "u" should be replaced by another,
unused letter. The reason is that italic u is already used for the
upper bound just above. Moreover "u" is also used for "up", though
with a different typography. The letters that shouldn't be used for
this (because they already have another meaning in §9.7.3 and §9.7.4):
b p d x s l u m r e.

-- 
Vincent Lefèvre <vincent@xxxxxxxxxx> - Web: <https://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)