On Monday, 29 April 2013 at 09:38:10 UTC, David wrote:
> Null blows up your code, "" doesn't.

There's no difference between null and empty string in D.

> Also it's a SQL thing, "NotNull"
> must be initialized, when the row is filled.

Full SQL support is good to have. I question the decision to make it default.

On Monday, 6 May 2013 at 09:14:57 UTC, Kagamin wrote:
> On Monday, 29 April 2013 at 09:38:10 UTC, David wrote:
>> Null blows up your code, "" doesn't.
>
> There's no difference between null and empty string in D.
>
>> Also it's a SQL thing, "NotNull"
>> must be initialized, when the row is filled.
>
> Full SQL support is good to have. I question the decision to make it default.

Current implementation looks consistent:

D type -- SQL type
int    -- INT NOT NULL
Int    -- INT NULL
long   -- BIGINT NOT NULL
Long   -- BIGINT NULL
string -- VARCHAR(255) NOT NULL
String -- VARCHAR(255) NULL

Uppercase types are nullable.

@NotNull/@Null annotations are not required. Nullability can be inferred from type name.

On Mon, 06 May 2013 11:14:56 +0200
"Kagamin" <spam@here.lot> wrote:

> On Monday, 29 April 2013 at 09:38:10 UTC, David wrote:
> > Null blows up your code, "" doesn't.
> 
> There's no difference between null and empty string in D.
> 

That's not true:

    assert("" !is null); // Passes

Or did I misunderstand what you meant?

On Tuesday, 21 May 2013 at 22:24:06 UTC, Nick Sabalausky wrote:
> On Mon, 06 May 2013 11:14:56 +0200
> "Kagamin" <spam@here.lot> wrote:
>
>> On Monday, 29 April 2013 at 09:38:10 UTC, David wrote:
>> > Null blows up your code, "" doesn't.
>> 
>> There's no difference between null and empty string in D.
>> 
>
> That's not true:
>
>     assert("" !is null); // Passes
>
> Or did I misunderstand what you meant?

Strings are slices which are a pointer and a length. I think a slice compares equal to null only if the pointer part is null. However, a slice with a null pointer and a length of zero is still a valid empty slice, which is slightly odd behaviour compared to other languages...

An empty string literal initialises the pointer to non-null because string literals are null terminated, so the memory block actually has a length of one, even though the slice has length zero.

On Wed, 22 May 2013 02:10:52 +0200
"Diggory" <diggsey@googlemail.com> wrote:

> On Tuesday, 21 May 2013 at 22:24:06 UTC, Nick Sabalausky wrote:
> > On Mon, 06 May 2013 11:14:56 +0200
> > "Kagamin" <spam@here.lot> wrote:
> >
> >> On Monday, 29 April 2013 at 09:38:10 UTC, David wrote:
> >> > Null blows up your code, "" doesn't.
> >> 
> >> There's no difference between null and empty string in D.
> >> 
> >
> > That's not true:
> >
> >     assert("" !is null); // Passes
> >
> > Or did I misunderstand what you meant?
> 
> Strings are slices which are a pointer and a length. I think a slice compares equal to null only if the pointer part is null. However, a slice with a null pointer and a length of zero is still a valid empty slice, which is slightly odd behaviour compared to other languages...
> 
> An empty string literal initialises the pointer to non-null because string literals are null terminated, so the memory block actually has a length of one, even though the slice has length zero.

Right, exactly. In other words, there *is* a difference between null and empty string in D (even though it's sometimes a clouded issue since '==' counts them as equal).

New version: v0.2.9
Fixed HQL parser bug with underscores in field and table names
Fixed compilation error with undefined identifiers in metadata.d
Merged pull request to use standard mysql-native library

On Wednesday, 22 May 2013 at 00:19:04 UTC, Nick Sabalausky wrote:
> Right, exactly. In other words, there *is* a difference between null and
> empty string in D (even though it's sometimes a clouded issue since
> '==' counts them as equal).

I meant there's no difference between them in meaningful scenarios (where empty string and null behave the same). Your example is synthetic (its also unclear wheter you compare strings or void*'s). I wouldn't expect this behavior to be reliable as D doesn't need distinction between null and empty string.

On Wednesday, 22 May 2013 at 00:19:04 UTC, Nick Sabalausky wrote:
> Right, exactly. In other words, there *is* a difference between null and
> empty string in D (even though it's sometimes a clouded issue since
> '==' counts them as equal).

== compare value, is compare identity. Everything is fine on that one.