RE: [Yaml-core] NULL or Magic values bad

[Oren Ben-K. <or...@ri...>]

Michael G Schwern wrote:
> I've noticed here:
> http://yaml.org/type/null/
> 
> That these values:
>      null|Null|NULL|undef|Undef|UNDEF
> 
> Indicate a special null type.  As there are already two other ways to do
> it,
> I'm not sure two (or six depending how you count) more are needed.

The status of the types isn't even "last call" - we really should spend some
time working out the final details there. The issue of the various ways to
write "null" has been raised and it hasn't been finalized yet.

> Worse, it adds another special case for the plain flow scalar.

Nope. We are simply building on the feature of "implicit types"...

> Something
> which should be done with Great Care as it reduces its DWIM value.  Or
> does
> it add to it by guessing that the user wants a hard null rather than the
> string 'null'... hmmm.

The answer is application dependent, but in principle, yes.

> Tangentally, if we have Perl/Ruby's "undef" should we also have Python's
> "None"?  If not, why have "undef"?

And there's also 'nil' (LISP)... perhaps we shouldn't be trying to cover all
bases and simply stick with just null/Null/NULL. Like I said, this is still
TBD.

Have fun,

    Oren Ben-Kiki