Re: [2/3] ntdll: Introduce a helper function to check for existance of server handles.

[Alexandre Julliard]

Sebastian Lackner  writes:
> @@ -1417,6 +1417,7 @@
>  # Server interface
>  @ cdecl -norelay wine_server_call(ptr)
>  @ cdecl wine_server_fd_to_handle(long long long ptr)
> +@ cdecl wine_server_handle_exists(long long)
>  @ cdecl wine_server_handle_to_fd(long long ptr ptr)
>  @ cdecl wine_server_release_fd(long long)
>  @ cdecl wine_server_send_fd(long)

Why do you need a new export for this?  Is it just to avoid a dup() call
or is there another reason?

[2/3] ntdll: Introduce a helper function to check for existance of server handles.

[Sebastian Lackner]

On 04.05.2015 09:25, Alexandre Julliard wrote:
> Sebastian Lackner  writes:
> > @@ -1417,6 +1417,7 @@
> >  # Server interface
> >  @ cdecl -norelay wine_server_call(ptr)
> >  @ cdecl wine_server_fd_to_handle(long long long ptr)
> > +@ cdecl wine_server_handle_exists(long long)
> >  @ cdecl wine_server_handle_to_fd(long long ptr ptr)
> >  @ cdecl wine_server_release_fd(long long)
> >  @ cdecl wine_server_send_fd(long)
>
> Why do you need a new export for this?  Is it just to avoid a dup() call
> or is there another reason?

The main idea was to avoid the unnecessary dup()/close() in situations where
no file descriptor is required.

BTW: Even if patch 2 is accepted, I'd like to resend patch 3 in order to replace FILE_READ_DATA -> 0.
I'll wait for your opinion on patch 1 and 2 first before resending.

[2/3] ntdll: Introduce a helper function to check for existance of server handles.

[Alexandre Julliard]

Sebastian Lackner  writes:
> On 04.05.2015 09:25, Alexandre Julliard wrote:
> > Sebastian Lackner  writes:
> > > @@ -1417,6 +1417,7 @@
> > >  # Server interface
> > >  @ cdecl -norelay wine_server_call(ptr)
> > >  @ cdecl wine_server_fd_to_handle(long long long ptr)
> > > +@ cdecl wine_server_handle_exists(long long)
> > >  @ cdecl wine_server_handle_to_fd(long long ptr ptr)
> > >  @ cdecl wine_server_release_fd(long long)
> > >  @ cdecl wine_server_send_fd(long)
> >
> > Why do you need a new export for this?  Is it just to avoid a dup() call
> > or is there another reason?
>
> The main idea was to avoid the unnecessary dup()/close() in situations where
> no file descriptor is required.

I think it's better to keep the interface clean, which is also the main
reason for having the dup() in the first place.