device filtering support

[Pete Batard]

OK, since I was already well into my reply before I saw Kusti's one, 
this will be my last reply on the subject, and then I will shut up.

On 2012.02.06 12:53, Michael Plante wrote:
>>> you can end up with 2 processes servicing 2 different versions
>>> of the inter process API
>
> Oh, now you say it.  Well, if you go that way, that DOES solve it.  I'm
> surprised it took this long, though.

Because it was obvious to me, and it seems to me that you guys are not 
exercising objective criticism on the reasons you think it cannot fly. 
Especially you don't appear to be trying to think about simple counter 
arguments that may diminish the points you make.

You seem to be surprised that I appear to have thought about this 
solution for more than 5 minutes and may _also_ have tried to find 
arguments that would deconstruct my proposal and prove it either 
unfeasible or utterly unrealistic on my own.

Well, I don't really have a problems with you guys trying to expose 
stuff that I may not have considered. On the contrary, since there's 
plenty of time where I expect to get things wrong, and you can bring 
experience in areas where I don't, this is what I prefer. But please 
also try to exert objective criticism on your counter-arguments before 
proposing them and at least see if maybe there's not some aspect that 
would make them not as relevant as you think they are.

I'll be frank, I'm getting the impression that whatever I propose, and 
that goes against the tide, is immediately dismissed as "surely, he 
hasn't thought this through". That's usually not the case. And I find it 
seriously annoying, because then I have to go into yet another crusade 
as to why I think this is an option we should consider, and try to 
explain details that you may have yet to see, instead of simply waiting 
to be in a position to submit an explicit proposal of what I have in 
mind, one that doesn't leave room to interpretation (or realize, while I 
try to implement such a proposal, that it actually cannot work, and 
leave the matter closed, which is also all benefit for everyone).

And that's why I think this matter should be left for after we have 
hotplug and are in a position to _actually_ deal with the issue we 
perceive need to be addressed. This is what I said to Xiaofan and Travis 
at the time, and this is what I repeated right where this whole thread 
started.

> (Not to suggest I like it, but this
> is, I think, a prerequisite in the design of a process-based solution.)

I think I mentioned that before. And it's not that difficult, at least 
on Windows:
   CreateMutex(NULL, TRUE, "Global/libcdio-enum_API1.0");

No process with the same mutex? Spawn away (may result in 2 processes, 
one with API1.0 another with API1.1). Mutex? Has the same API so it can 
service us.

Regards,

/Pete