Dag wrote :
>
> Have you actually read your own answer? You are telling me that 127
> simultanous connections on top of Ultra is not enough! I remember once
> I had 7 "normal" connections at the same time ;-)
Wrong analogy, the PID are not dynamic. Once one is allocated
for an application, it's for life.
> Ultra _will_ be used since you can already find it in a couple of mobile
> phones which supports OBEX over Ultra. OBEX over Ultra uses PID 0x01, so
> that give you 126 other PID's to play with. Nobody else are using them ;-)
> If you need more for some special experiment, then why don't you multiplex
> on top of one Ultra connection (or use your own patch)?
Because ultra is suppose to do that for me, and I don't want
to add another layer of protocol when I don't need one.
> BTW. I really cannot see that limiting the number of PID bytes to 4 is more
> conformant to the spec than limiting it to 1 byte.
It's not. But what I meant is that this feature was 100%
compliant, or a least not less compliant than yours, so removing it
was stupid.
> > I just hope you didn't hardcoded 0x70 all over the place. My
> > code was careful to strike a balance between compliance to the spec
> > and future extensibility.
>
> Yes I did ;-) In section 3.2.2 (page 21) in IrLMP they say:
>
> "All Data LM-PDUs delivered via IrLAP_Unitdata.indication primitived (UI
> frames sent outside an IrLAP connection) except for those addressed between
> Connectionless LSAPs (DLSAP-SEL=SLAP-SEL=0x70) are discarded."
>
> Now why did they use the PID header if they planned to allow
> for more connectionless LSAP's? You wanted almost 4M "connections" over one
> SAP yourself!!! So now you want even more "connections"?
Because when I can code clean and leave flexibility, I do
it. I don't introduce stupid limitiations in my code just for
pleasure, especially when those limitations have no drawback.
Now, I will explain you why : all the PID above 0x70 are
reserved by IrDA. So, in theory is I've got an app sitting above 0x70,
I must ask IrDA to give me a PID. IrDA will say "show me the money"
and make me wait a few years to give me a number.
Above any other LSAP (for ex 0x71), IrDA has not defined
anything yet, so I'me free to do whatever I want. Basically, I can
developp my application, ship it in products, tell other people about
it, and IrDA can't complain about it and doesn't come in my way.
> > Currently, I feel that's it's me doing most of the debugging
> > and correcting your mistakes in the code.
>
> Sorry for giving you the Ultra patch in the first place! I told you this
> was work in progress! You said your were interested in Ultra, so I made a
> patch out of my code. Silly me! Why don't you make your own patches it you
> don't like mine. When you make patches for my patches, then you should
> accept it if I choose to do things differently.
I'm not talking about the Ultra code. Ultra is the least of my
troubles. With each new patch, some of the basic functionality is
broken. Look at my last patch and tell me how much is related to
Ultra.
>
> > Try to read a bit more carefully patches, comments and so on,
>
> Did that with the actisys dongle patch didn't I?
Ha, ha, ha ! And you introduced this stupid typo preventing
the module to compile ! I would not be so smuck. At least, my version
was working.
> BTW: I'm having problems with the auto-connect patch because it has
> problems when you put 2 or more IrDA devices in front of your machine
> (suddenly you don't care about making things so general after all). This is
> a policy decision which should be made in user-space and the user must
> probably be involved as well (as Win98 deals with it). That this hasn't
> been properly impl. in user-space yet in Linux, is not an argument for
> moving it into the kernel.
You didn't read into the detail of the patch, and it was much
more clever than that. Anyway, this is not a policy decision, because
the old way of doing things is working, you can still perform
discovery and set the desired daddr. It's just an optional feature.
If you think about it, most of the time only one device has a
specific service activated, so it connect to the right device
automagically. But I will probably do better...
Jean
_______________________________________________
Linux-IrDA mailing list - [EMAIL PROTECTED]
http://www4.pasta.cs.UiT.No/mailman/listinfo/linux-irda
