Discontinuation of Linux 2.2.x and old wireless extensions support

[Mark Glines]

Jouni, my own opinion is that you should do whatever you need to in
order to minimize development overhead.  Porting PCMCIA drivers has
been a pain in my ass at times, so I can understand why you'd want
want to deprecate v2.2 support.  Especially with v2.6 coming up.
Since the existing driver is already quite useable for v2.2, I'd say
go ahead and drop it.  It seems nowadays like most of the new stuff
is going into the userspace hostapd anyways...


[the following is offtopic]
On Sun, Apr 13, 2003 at 12:07:57PM +0100, gARetH baBB wrote:
> On Sat, 12 Apr 2003, Sergio M. Ammirata wrote:
> 
> > What about the openap (now linuxap, OpenAP-ct and openap-ng) projects?
> > They have a 2.4 system that has only 1MB of HD space and 4MB memory.
> 
> Maybe, perhaps I've overstating the case - but there is still some
> hardware which doesn't like 2.4, and there is still alot of problems with
> various parts of 2.4, viz. the number of patches people seem to apply
> almost by "default" to get a reasonably stable kernel.

This is an interesting statement.  I should mention that I've found
the 2.4 series to be perfectly stable nowadays, and all my boxes are
currently running unpatched 2.4.19 and 2.4.20 with no problems.  The
v2.4 series of kernels has better support for my hardware than v2.2,
too.  It works better both on my SMP machines at home, and on the
AMD SC520 microcontroller-based embedded machines at work.

At times it seems like 2.4 has a bad reputation just because it had
a rough start.  But this was true of v2.2 and v2.0, as well, and like
v2.2 and v2.0, v2.4 has become quite mature as it progressed.
Actually, a few my boxes ran 2.4.5 for more than 1.5 years until the
power went out...

So, I'm curious whether your opinion of the stability of v2.4 is based
on fact or conjecture.  What patches are these that you refer to?


> And it certainly is more bloated per feature set - 2.5 even more so, which
> has been a recent-ish thread in lkml.

I haven't noticed this problem, personally, but I'm sure they would
accept patches that address this.  It is open source, after all.

Mark