Re: [TCLCORE] CFV: TIP 283

[Donald G P. <dg...@ni...>]

Kevin Kenny wrote:
> (2) This is the only chance that we'll get to do it right;

Yes, which is why it is important to actually get it right.

> (3) The [info level] issue is very much a red herring.  The only code
>     broken by the change is [info level] within a procedure that is
>     a member of an ensemble.  

Ah, but there is no special command type known as "member of an 
ensemble".  A key feature of ensembles is that they can dispatch by name 
to any command at all, whether or not that command was initially 
designed for the purpose of being the target of an ensemble subcommand.

I think your analysis would be correct if ensemble subcommand targets
were something new, like the methods proposed by TIP 257, that could 
never be reached except via new features only in 8.5.  That's not the case.

As I understand the current state of TIP 283, it creates a new situation 
where a proc can no longer rely on prior assumptions about what [uplevel 
1 [info level 0]] will do.  That's worth either correcting, or embracing 
and declaring in huge bold letters in the proposal as an incompatibility.

-- 
| Don Porter          Mathematical and Computational Sciences Division |
| don...@ni...             Information Technology Laboratory |
| http://math.nist.gov/~DPorter/                                  NIST |
|______________________________________________________________________|