Re: [sv-ec] Mantis 958: array query functions

From: Michael Burns <michael.burns_at_.....>
Date: Fri Dec 14 2007 - 08:26:12 PST
To be honest, my long-term preference would be to junk the idea of 
getting a dimension size altogether; I'd like to preserve the idea that 
arrays have sizes, dimensions (in general) don't. You can always get the 
same idea from selecting an array from that dimension:

  bit a [0:1][4:27];

rather than

  $left(a,2);

use

  $left(a[0])

or, if you don't even know the range,

  $left(a[$left(a)])

or something. This may go against the whole idea behind these system 
functions (I wasn't around when they were introduced), but I'd really 
like to avoid all the complex special-case rules, and also avoid using 
the concept of array dimension entirely (except when describing the 
special syntax for multiple dimension arrays).

On the other hand, I don't think we have any room now to make 
non-backwards-compatible changes; pragmatically speaking, I think your 
change is the best we can do in the next few hours :)

--Mike

Jonathan Bromley wrote:
>> it shall be illegal to call any query function with a dimension
>> number N>1 if the variable in question has more than one 
>> variable-sized dimension among its dimensions <= N.
>>     
>
> Actually it's easier than that:  if dimension N is variable-size,
> and N>1, then it's illegal.
>
> I could add exceptions; for example, all variable-size dimensions
> always have $increment = -1.  But I think it's far easier just
> to outlaw any query on such a dimension.
>
> I've uploaded a new version of the proposal, proposal-958-2,
> with a new clause 19.7.1 added.  The existing proposal-958-1a
> is still there, since it's the subject of the vote.
>
>   


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
Received on Sat Dec 15 02:38:10 2007

This archive was generated by hypermail 2.1.8 : Sat Dec 15 2007 - 02:38:33 PST