pidl facillitates illicit data sharing between idl files

[Sam Liddicott]

* tridge wrote, On 29/09/09 22:13:
> Hi Sam,
> 
>  > If I add a definition such as this, to my pidl file:
>  > 
>  > 	typedef struct DATA_BLOB {
>  > 		uint8 *data;
>  > 		uint32 length;
>  > 	} DATA_BLOB;
> 
> There are two problems with this:
> 
>  1) DATA_BLOB is a base type in pidl. It is not defined in any idl
>  file, but instead is a built-in type like uint32 and other base
>  types. Trying to re-define it is a bad idea.

good tip. I'm also working with MIDL where it isn't a base type, but I 
can #ifdef that

>  2) the normal form for structure typedefs in pidl is:
> 
>    typedef struct {
>    	   el1type el1;
>    	   el2type el2;
>    } NAME;
> 
>   Notice that the first line does not contain the structure name.

Aye, but:
http://blogs.msdn.com/oldnewthing/archive/2008/03/27/8338530.aspx

(MIDL again) but I can have a macro to nullify the struct name when not 
running under midl.

>  > For my own benefit I just renamed by struct to avoid the collision, but 
>  > I believe it is not intentional that one IDL file should cross-infect 
>  > another in this way - or have I mis-understood?
> 
> you've re-defined a base type. All bets are off after that :-)


No doubt - and thanks for the tips.
I wonder if, if I had also re-defined one of the private types used in 
winsrepl.idl if it would have had a similar effect.

It was the cross-infection that only occurs if all the IDL need 
compiling that I was thinking about...

anwyay, thanks for the tips.

Sam