Re: Zeroes and infinities

[Thorsten Siebenborn <7_born@xxxxxx>]

David James schrieb:
> Thorsten,
>
> Thanks for the interesting comments.
> Is your concern that:
>   1) The 754R draft should be changed.
>   2) The DVJ tables are inconsistent from the 754R draft.
>
> I suspect your concern is with (1), although the concern
> only became visible when the compacted/pseudo-encrypted
> text was concerted to a table.

What happened: I saw the table and got into a discussion
about the sense or nonsense of the return values.
Thinking about IEEE 754 I encountered some oddities which
I don't like very much.

> If (1) is the concern, then a rogue comment submission
> would be helpful to the BRC.

Some remarks:
a) How long should the standard be open to changes ?
   I thought it should be already complete last year ?

b) If it is nearly complete, there is quite a natural
   resistance to include changes. Heck, I would be
   resistant, too.

c) As there are different members with different competence,
   it also boils down to competence. I am skeptical that
   the BRC is open to discuss proposals with an amateur having
   no papers published before; especially because much in
   754 R has been adopted from C99 and if Nick had problems,
   well...

And it's easy to criticize, but hard to walk the talk.
I think I am going to test my ideas first by implementing
my own system; if I can show that this thing DOES function
and is superior to 754, then I am much more confident
with proposals. I am currently searching for errors in
my argumentation.

Only in implementation you can encounter and analyze
problems you haven't recognized before.

> If (2) is the concern, then a clear delineation of the
> conflicting text in 754R and DVJ tables would be most
> helpful to myself (DVJ).

What exactly *may* be changed if I want to propose sth ?
Errors, ambiguities, definitions ? And who can object ?
I don't want to waste time by proposing sth and it is
silently dropped.

Best,
Thorsten