From octave-maintainers-request at bevo dot che dot wisc dot edu  Mon Jul 31 15:04:44 2000
Subject: Re: Ordered NaN on purpose? 
From: "John W. Eaton" <jwe at bevo dot che dot wisc dot edu>
To: Edward Jason Riedy <ejr at EECS dot Berkeley dot EDU>
Cc: Joao Cardoso <jcardoso at inescporto dot pt>,         octave-maintainers@bevo.che.wisc.edu
Date: Mon, 31 Jul 2000 15:05:07 -0500 (CDT)

On 30-Jul-2000, Edward Jason Riedy <ejr at EECS dot Berkeley dot EDU> wrote:

| So would a patch that does the following be accepted:
| 
|   1) Make a global flag for the default min/max behavior.
| 
|   2) Add an optional parameter for min/max to override the default.
| 
|   3) Make the default be to return a number if at all possible, so
|      min([NaN, -Inf]) == -Inf.

Sure, but I would prefer to just use the `correct' solution instead of
introducing a new flag.  If you want to have different behaviors,
perhaps it is best to use a different function, or an option to the
min/max functions.  Global flags for things like output formats are
not too bad, but for this purpose (changing the way a function works)
they usually cause trouble (I know, Octave already has too many of
them that behave this way, but it may be too late to change most of
them).

jwe