Re: Implementation of convn

[David Bateman]

John W. Eaton wrote:
> On 27-Mar-2008, David Bateman wrote:
> 
> | However private directories can't stop the user copying the file from
> | the private directory to their home directory and using it in any
> | case.
> 
> Yes, but at that point, they "own" the file.  What I was thinking
> about is that someone uses a private function and they don't even know
> they are doing it, so then when the interface changes or disappears,
> they complain.  

How can they not know.. They have to themselves run the "sharedprivate"
function to get access to the files. In my opinion at that point they
take responsibility for the files imported into the function space of
the directory the "sharedprivate" function is run in, even though the
ownership rests with the installer.. I really don't see that the
situation is different.

> Of course that's the situation we have now (I think
> the naming scheme is not much of a clue for most) so yes it probably
> would be better to have something that allow us to have private
> .oct/.mex file functions and still maintain the separation of the
> architecture independent and dependent files.  Perhaps give it a name
> like __merge_private__ and have it emit a warning if not called from a
> PKG_ADD file?

Sure..

D.

> 
> jwe
>