Re: [Alephmodular-devel] Random rumblings on files and errors

[Br'fin <br...@ma...>]

On Thursday, January 30, 2003, at 06:56  AM, Michael Adams wrote:

> --- Br'fin <br...@ma...> wrote:
>> By calling conventions I was thinking of calls like
>> this:
>>
>> 	virtual bool get_named_child(const char *name,
>> IFileSystemDesc*
>> &out_desc) const = 0;
>>
>
> I thought you had mentioned useing STL strings instead
> of char* in future code, but maybe I'm remembering
> wrong.  Also did you really mean IFileSystemDesc*
> &out_sesc?  That * with the & confuses me.  Oh I think
> I get it now.  So you would call it like this:
>
> IFileSystemDesc *f = NULL;
> a->get_named_child("name", f);
> And the pointer f would now point to something new?

Yes, that was my idea when writing it out that way.

> I haven't been paying much attention to this whole
> discussion, but why not do this?
>
> ... get_named_child(..., IFileSystemDesc &out_desc);
> ...
> IFileSystemDesc  f;
> a->get_named_child("x", f);
>
> It would make one less pointer floating around which
> reduces chance of seg faults, but like I said I
> haven't been paying close enough attention so maybe
> there is some reason for using a pointer that I'm
> missing.
>
The biggest problem with that is this is C++. So IFileSystemDesc is an 
interface. An Abstract Base Class.

'IFileSystemDesc f;' would result in a compile time error (Trying to 
instantiate an incomplete class) But even if it were a full class, it 
would be set in its ways. You can't really cram a larger subclass into 
IFileSystemDesc as it was allocated purely and directly as a 
IFileSystemDesc.

-Jeremy Parsons