device filtering support

Michael Plante michael.plante at gmail.com
Sat Feb 4 19:34:25 EST 2012


Pete Batard wrote:
>> With the current Windows enum process, this results in
>> hubs being queried and potential packet dropped for app 1.

Ok, I understand, but I think that that is tolerable and most users need to
understand that can happen anyway with USB.  Don't go doing other things
when you expect time-sensitive operations to complete successfully (such as
watching video).  And if you do, try again.  Give people credit.  They'll
learn.  This is less bad than what I thought you were saying, so I still
think threads are the way to go.


>> Using pipes for messaging

That sounds better than shared memory.


>> The way I see the libusbx process should live for as long ad possible.

I don't see any benefit wrt your isochronous scenario above.  Leaving
processes running that don't die by any means other than task manager is
even worse than allowing someone to shut them down through the services
management console (which most users may not even know about).

There's one other issue:  versioning.  I doubt I need to elaborate here
much, other than to say programs would probably need either the newest
version running, or maybe one of each version.  After all, the pipe protocol
could change, and enumeration/hotplug bugs can be fixed.

Regards,
Michael




More information about the libusbx mailing list