Re: [7/10] [try 2] comctl32: Add basic structure for IImageList interface

[Henri Verbeet]

2009/8/11 Owen Rudge <orudge(a)codeweavers.com>:
> +static const IImageListVtbl ImageListImpl_Vtbl = {
> +    ImageListImpl_QueryInterface,
> +    ImageListImpl_AddRef,
> +    ImageListImpl_Release
> +};
> You can't do that, you need to at least implement stubs for the other methods.

[7/10] [try 2] comctl32: Add basic structure for IImageList interface

[Owen Rudge]

> You can't do that, you need to at least implement stubs for the other methods.

This is something I did in the following patch, mainly to reduce the 
size of the individual patches (based on the assumption all the patches 
would be committed at once anyway). I will recombine the patches into a 
single patch though if necessary.

Cheers,

-- 
Owen Rudge
http://www.owenrudge.net/

[7/10] [try 2] comctl32: Add basic structure for IImageList interface

[Henri Verbeet]

2009/8/11 Owen Rudge <owen(a)owenrudge.net>:
> > You can't do that, you need to at least implement stubs for the other
> > methods.
>
> This is something I did in the following patch, mainly to reduce the size of
> the individual patches (based on the assumption all the patches would be
> committed at once anyway). I will recombine the patches into a single patch
> though if necessary.

Patch series aren't atomic, especially not when bisecting. A better
way to split things might be separating the classfactory and
IImageList implementation.

Also:
> +    classinfo               *info;
> This should probably be const.
> +    if (!ppv) return E_INVALIDARG;
> Any special reason for this check? MSDN claims it should return
> E_POINTER, but most implementations don't bother checking at all.

As a personal preference, I don't think adding typedefs for structs
adds much. Arguably, the only thing it does is hide that something is
a structure.