[gmpi] Re: low level API - Abstract Factory summary
- From: "B.J. Buchalter" <bj@xxxxxxxxxx>
- To: "gmpi@xxxxxxxxxxxxx" <gmpi@xxxxxxxxxxxxx>
- Date: Fri, 11 Feb 2005 23:24:38 -0500
on 2/11/05 7:04 PM, Jeff McClintock at jeffmcc@xxxxxxxxxx wrote:
Jeff,
First off the two quotes in your message are from two different people.
Can we try to keep attribution in quotes/replies so that the thought threads
are not confused?
>> What about the object pointer thunking for multiple disjoint
> interfaces on the same object.
I asked this.
> I don't recall anyone suggesting "multiple disjoint interfaces on the
> same object".
Not correct. This has definitely been suggested by at least one person in
this discussion.
> I take that to be an object with multiple inheritance from two other
> COM-enabled objects.
Pretty much. Basically multiple Interfaces exposed by one object.
> Did anyone propose GMPI_Plugin needed multiple inheritance? No.
I asked about this because this definitely WAS suggested in this discussion
(Mike B, maybe Ron too), and I originally asked if multiple disjoint
interfaces were being contemplated, and the reply was yes. I also suggested
if the concept of multiple interfaces was implemented as factory methods for
independent interfaces/objects (much like Tim layed out in his 6:53 message)
than this issue goes away; but the point is that if Multiple interfaces on
one object are being contemplated than this issue needs to be hashed out.
If, on the other hand, we are not contemplating multiple interfaces on the
GMPI_Plugin object, than everyone needs to understand that and think about
it correctly -- these are fundamental implementation issues and need to be
worked out and understood.
> Why is some rare, complex corner case been presented as an argument
> against COM, as if we have to 'solve' the non-existant problem by
> ditching COM and inventing something ourselves.
I am not arguing against COM. I am asking pointed questions to people who
have more experience with it about things that I perceive to be real issues
*depending on what style of implementation is chosen*. In particular,
personally I think that COM is a great solution as long as it is constrained
properly so that implementing it is clean and safe.
Seems to me that we are triangulating pretty quickly, and that there are
other non-COM folks that are arriving at a better understanding of the COM
interface as well. If this is what we are going to use, it is critical for
everyone to understand the implications of these decisions.
>> Personally, I prefer to concentrate on a simple C interface.
This was *not* my comment.
> How will your "simple C interface" cope with "object pointer thunking
> for multiple disjoint interfaces"?
You wouldn't have multiple disjoint interfaces in the simple C case; we may
not have multiple disjoint interfaces in the COM case either -- but it has
to be by design, because that is certainly one design style that would meet
the requirements for optional interfaces.
Best regards,
B.J. Buchalter
****************************************************************************
NOTE: The contents of this message is METRIC HALO CONFIDENTIAL, and is for
the intended recipient ONLY. The contents of this message is governed by the
terms of the NDA that you have with METRIC HALO. DO NOT REDISTRIBUTE THE
CONTENTS OF THIS MESSAGE or any information contained HEREIN.
****************************************************************************
----------------------------------------------------------------------
Generalized Music Plugin Interface (GMPI) public discussion list
Participation in this list is contingent upon your abiding by the
following rules: Please stay on topic. You are responsible for your own
words. Please respect your fellow subscribers. Please do not
redistribute anyone else's words without their permission.
Archive: //www.freelists.org/archives/gmpi
Email gmpi-request@xxxxxxxxxxxxx w/ subject "unsubscribe" to unsubscribe
Other related posts:
