Version 1.1 of acs/ac-00328.txt

Unformatted version of acs/ac-00328.txt version 1.1
Other versions for file acs/ac-00328.txt

!standard 2.4.2(6)          20-03-22 AC95-00328/00
!class Amendment 20-03-22
!status received no action 20-03-22
!status received 20-01-10
!subject Extended number bases
!summary
!appendix

From: Peter Hermann
Sent: Friday, January 10, 2020  10:37 AM

A comment on this:

 >"Randy Brukardt" <randy@rrsoftware.com> 2020-01-10, 05:14  
 >"'Ada-Comment List'" <ada-comment@ada-auth.org>  > ...
 >(We've all "seen the need" for some feature or other over the years that  
 >didn't make the cut!)

So was the fate of
http://www.ada-auth.org/cgi-bin/cvsweb.cgi/acs/ac-00070.txt?rev=1.1
on a tiny improvement of numberbase.
It had been dropped in 2003.
Moreover, in 2010, my minimal refinement of the core language had been diluted 
by Babylonian baloney (e.g. AI05-0226-1) and other off_road thoughts.

But a few understood , e.g.

 >From: Tucker Taft Sent: Tuesday, June 29, 2010  7:54 AM  
 >... not in the ivory towers of language lawyerism.

 >From: Martin Dowie   Sent: Thursday, June 12, 2003  4:37 PM  
 >...
 >This would be _very_ handy for things like digital maps systems and map  
 >preparation facilities.

 >From: Martin Dowie   Sent: Friday, June 13, 2003  2:43 AM  >...
 >Also, you'd still have the prepend the literal with "33#" or  
 >whatever, so that should give a hint that special care and  
 >checking is required.
 >I'd rather check that than some of the more convoluted  
 >algorithms currently used!

 >From: Robert Dewar  Sent: Tuesday, June 29, 2010  8:35 AM  
 >... it is trivial enough to add, a few minutes work  
 >in the compiler nothing more, so I don't seriously object ...
 >
 >From: Robert Dewar Sent: Tuesday, June 29, 2010  9:23 AM  
 >I am completely convinced we should NOT go beyond base 36...

I confess and respect:
A negative point is the broken backward compatibility.

Ada with its representation clauses has substantial gains for practical 
programming close to hardware. This was one of the many(!) reasons, why I saw 
Ada as the best successor of Fortran.
We had to run our finite element systems on all existing hardware.

***************************************************************

From: Randy Brukardt
Sent: Sunday, March 22, 2020  7:19 PM

> Moreover, in 2010, my minimal refinement of the core language had been 
> diluted by Babylonian baloney (e.g. AI05-0226-1) and other off_road 
> thoughts.

That proposal was rejected because it was judged by many to harm readability.

Quick: what value does 'R' represent in base-32?? I have plenty of trouble 
with 'D'. :-)

Additionally, the pairs '1' and 'I' and 'O' and '0' are easily confused.

Note that Ada 202x includes a user-defined literal feature, so one can add any 
form of literal one wants to a numeric type using a custom string literal. 
(There's also a user-defined 'Image, so it can be made consistent).
So the need can be met with a slightly less convenient notation.

***************************************************************

Questions? Ask the ACAA Technical Agent