Re: [boost] Singular variants?

From: Gabriel Dos Reis (gdr_at_[hidden])
Date: 2003-10-11 03:37:05

"E. Gladyshev" <egladysh_at_[hidden]> writes:

| --- Gabriel Dos Reis <gdr_at_[hidden]> wrote:
| > "E. Gladyshev" <egladysh_at_[hidden]> writes:
| > 
| > | If the variant interface says that the variant can
| > | go 'singular' or 'unspecified content' in *some* cases,
| > | The poor guy will have to handle it as a normal
| > | normal variant's interface. He would not care when
| > | and how the variant goes 'singular' or 
| > | 'unspecified content'. If any of these can happen, 
| > | he should handle it in his f() function.
| > | It is like f( char* p ). A good f() behaves well
| > | when p = NULL.
| > 
| > I disagree.  A singular value is not like any other values.  It is a
| > mistake to make people into believeing otherwise.  
| 
| What is so special about it?

I guess if it weren't special, it won't get the qualification
"singular" or "unspecified value".

| > encouraged to read docs and be careful about specs.  Variant isn't the
| > kind of tools that the careless programmer  is supposed to use.
| 
| On the contrary, I was talking about a programmer who cares and wants
| to make her f(...) function safe and sound.

if they care, then they care about the arguments they get, especially
about the singular ones.  Their functions aren't going to be safe just
because they refuse to recognize that some arguments are singular.

-- Gaby