[Omake] Re: [CUCS Bug 545] "or" and "and" take each argument to be a _sequence_ of values.

[Jason Hickey]

Aleksey Nogin wrote:
> Right. And it would be much better to have _this_ one be the default 
> "and" than what we have now.

Wait wait.  Before you redefine "and", you should consider the 
ramifications.

> a) Create a mixed version that acts as the non-array one for >= 2 
> arguments and acts as the array one for 1 argument

I believe this is a bad idea.  There are many complicated functions in 
OMake, but "and" should not be one of them.

> b) Create a mixed version that acts non-array-algebraically on each 
> individual OMake value, but as a regular "and" _across_ arguments. (BTW, 
> unless you insist on the misleading "bool list list -> bool" 
> interpretation, this is completely logical once the whole space of 
> non-strictly-boolean values is considered; arrays of length 0 is _not_ 
> the only meaningful interpretation of an empty OMake value!).

Huet mentioned this line of argument, as in (his example was for the 
sake of argument), "Stupid de Bruijn thinks numbers stand for 
variables."  The responder first has to spend time saying de Bruijn is 
not stupid before he can make a real response.

So let me just say, I did mention the type "bool list list -> bool", but 
I had no intent that it should be misleading/canonical.  Next, let me 
use the Aydemir argument, "Oh, whatever...":)

If you are sure that $(and) needs to be redefined, let me just insist:
    - The identity for "and" is "true"!  Let's repeat!!!

Jason

-- 
Jason Hickey                  http://www.cs.caltech.edu/~jyh
Caltech Computer Science      Tel: 626-395-6568 FAX: 626-792-4257