On Mon, 28 Aug 2006 15:31:55 +1200
Christopher Sawtell wrote:

> On Monday 28 August 2006 12:27, Nick Rout wrote:
> > It is very unclear to me whether it is dying because of failed
> > authentication, or because it fails to start ppp after successfully
> > authenticating
> I think it only manages to dial the number, and make an attempt to create a 
> data connection, but fails completely. 

no, IIRC it establishes a connection, or else 

(1) it wouldn'tbe saying CONNECT
(2) it wouldn't be sending LCP config requests
(3) I'm not sure, bnut I don't think it produces this line:

 Connect: ppp0 <--> /dev/ttyLTM0

Until the modem has connected (don't quote me on number 3)


>Would not the fact that Andrew had it 
> going with an external modem, and no other changes, mean that the 
> authentication setup is working correctly?

Possibly, but don't forget, as my experience yesterday showed (failed
first time, succeeded thereafter) the problem may not always be the
same.

> 
> My own opinion is that there is probably a kernel / modem / gcc versions 
> inconsistency.

IMHO that is wrong, it is all built with the same system, its all part
of the same distro.

> 
> At this point I'd be very tempted to build a recent stable kernel and then 
> the 
> open source module and interface shim for the closed modem module using a gcc 
> from the 3.4 series.
> 
> Whether or not this is feasible under the version of Mepis I know not, but I 
> do know that doing it would be a 'somewhat time-consuming exercise' to put it 
> very mildly. 

a waste of time IMHO, but about 30 minutes compiling plus 30 minutes
futzling - say 2 hours. I would have thought, however, given that the
driver and the kernel are all compiled for the same release of the same
distro, that it shouldn't make a difference.

> 
> If this whole schemozzel fails to prove that binary kernels, modules, and 
> distributions of Linux are totally undesirable, I don't know what does.
> 

No it doesn't, because it is known that this driver from this distro
works. (By the way I agree with you that binary drivers are a pain, but
not that this episode confirms it. What would you change if you had the
source? We don't know what is failing.

> btw, Do we have another Gentoo installfest on the horizon?
> 

Nah, its soo easy now, no fun in that!

-- 
Nick Rout <[EMAIL PROTECTED]>

Reply via email to