On Tue, Jan 08, 2002 at 10:31:21AM +1100, Simon Wong wrote:

> > > You could have saved some money if it's a Lucent based Winmodem as they
> > > work well with the linmodem drivers developed by the people at
> > > http://www.linmodems.org/ 
> > 
> > Please don't go around promoting OS-dependent hardware :).  They
> 
> Since he already had one, and it could work, he could have saved himself
> some money.  Why pay more than you have to?

I think I missed something, but I didn't read anywhere in his
original message that he was going to buy a new modem ... ?

> I would like to put pressure on h/w manufacturers to support open
> sourcing drivers for their hardware.

I agree, but promoting downloads of their driver's binary modules
aren't much of an incentive for them to do this.

> > maintain, keep up-to-date, and you have sometimes have to wait
> > for quite a while to get them working with the newer kernels.
> 
> The source code of the ltmodem driver has worked with both 2.2 and 2.4
> kernels for me.

I've seen it OOPS on me many a time.  It's working acceptably
right now, but it certainly wasn't for a while; iirc it was
around 2.4.1[23] where it started breaking.  I also couldn't get
it to work with some kernel patches that I _needed_ to run at the
time (NeTraverse's win4lin kernel patches), and while I probably
could have solved the problem, I really had other things to do at
the time.  (Like ski ;).

> > They also work via source code wrappers around manufacturers'
> > binary modules, which can cause more problems in the future.
> 
> I don't believe that this is the case with the ltmodem driver.

Kernel revisions tend to change many of the API/interface
functions at a scary rate, and the drivers have to keep up with
them.  It's not easy to do this.  Even 'minor' kernel revisions
can break things, and changes like 2.2 -> 2.4 (not to mention
2.4.9 to 2.4.10) breaks lots of things.  Keep in mind that you
also have no control over binary modules, and things that may go
into the kernel in the future (e.g. the preempt patch, VM) _will_
have disastrous effects until the rest of the kernel catches up.

In summary, I'm trying to say that while the binary modules work,
recommending such hardware for Linux isn't a good idea unless you
like spending lots of time getting it working.


-- 
#ozone/algorithm <[EMAIL PROTECTED]>          - trust.in.love.to.save
-- 
SLUG - Sydney Linux User Group Mailing List - http://slug.org.au/
More Info: http://lists.slug.org.au/listinfo/slug

Reply via email to