> What it does not do/is missing:
> - exception safety / multithread safety
> - out-of-class/struct serialization methods (is it possible to check
> whether a specific overload exists at compile time?)
> - static arrays need to be serialized with describe_staticarray (static
> arrays can't be inout, so the general-purpose template method doesn't
> work... is there a way around the problem?)
> - things I forgot right now

Hi, what about associative arrays ?
serialization\basicarchive.d(100): static assert  (0) is false, "describe called  with unsupported type"

> Hi, what about associative arrays ?

They're not supported at the moment as I forgot about them.

Hm, is there a way to check if a template parameter is an associative array and to get the key and value type without relying on .keys and .values?

Cheers,
Christian

Christian Kamm wrote:
>> Hi, what about associative arrays ?
> 
> 
> They're not supported at the moment as I forgot about them.
> 
> Hm, is there a way to check if a template parameter is an associative  array and to get the key and value type without relying on .keys and  ..values?
> 
> Cheers,
> Christian


static if(is(A in B))

??? IIRC in only works with AAs

BCS wrote:

> Christian Kamm wrote:
>>> Hi, what about associative arrays ?
>> 
>> 
>> They're not supported at the moment as I forgot about them.
>> 
>> Hm, is there a way to check if a template parameter is an associative array and to get the key and value type without relying on .keys and ..values?
>> 
>> Cheers,
>> Christian
> 
> 
> static if(is(A in B))
> 
> ??? IIRC in only works with AAs

opIn is overloadable.

-- 
Lars Ivar Igesund
blog at http://larsivi.net
DSource & #D: larsivi

> Hm, is there a way to check if a template parameter is an associative array and to get the key and value type without relying on .keys and .values?

What I'm planning on using is:

static if(
  !is(T == struct) &&
  !is(T == class) &&
  is(typeof(T.keys) KEY == KEY[]) &&
  is(typeof(T.values) VALUE == VALUE[])
)
{
  // static assert( is(T == VALUE[KEY]) );
}

It seems to work, but is rather clumsy.

Christian

Christian Kamm schrieb am 2006-12-21:
>> Hm, is there a way to check if a template parameter is an associative array and to get the key and value type without relying on .keys and .values?
>
> What I'm planning on using is:
>
> static if(
>    !is(T == struct) &&
>    !is(T == class) &&
>    is(typeof(T.keys) KEY == KEY[]) &&
>    is(typeof(T.values) VALUE == VALUE[])
> )
> {
>    // static assert( is(T == VALUE[KEY]) );
> }
>
> It seems to work, but is rather clumsy.

static if(is(T == typeof(T.values[$])[typeof(T.keys[$])]){
	// it is an AA, now do your KEY and VALUE magic
}

Thomas

> static if(is(T == typeof(T.values[$])[typeof(T.keys[$])]){
> 	// it is an AA, now do your KEY and VALUE magic
> }

Good idea! However, it doesn't compile for T = int, for example (unless that was changed between 0.173 and 0.177):
- no property 'keys' for type 'int'

Why is T.values[$] allowed, though - and is the [$] neccessary? I've so far relied on typeof returning the type of the return value for functions, but couldn't find it in the spec. Maybe it would be best to use is(T.values VALUE == return)?

Christian

Christian Kamm schrieb am 2006-12-21:
>> static if(is(T == typeof(T.values[$])[typeof(T.keys[$])]){
>> 	// it is an AA, now do your KEY and VALUE magic
>> }
>
> Good idea! However, it doesn't compile for T = int, for example (unless
> that was changed between 0.173 and 0.177):
> - no property 'keys' for type 'int'

Wrong order..., this should work:

static if(is(typeof(T.values[0])[typeof(T.keys[0])] == T)){
 	// it is an AA, now do your KEY and VALUE magic
}


> Why is T.values[$] allowed, though - and is the [$] neccessary?
$ is used to get an element of the array. You can use any index for that purpose.

Thomas

> Hi, what about associative arrays ?

Added support for associative arrays:
http://www.math.tu-berlin.de/~kamm/d/serialization.zip

I only did a very small unittest for testing the new functionality, so please report further problems and bugs to kamm 'replace with at' math.tu-berlin.de instead of the newsgroup.

I've also asked Brad for a project on dsource; once it's set up, further updates will go into the repository there.

Christian