Re: axp patches for 2.2.9

[abel bfr co il (Alexander L. Belikoff)]

Richard Henderson <rth@twiddle.net> writes:

> On Fri, Jun 11, 1999 at 03:24:24PM -0600, Michal Jaegermann wrote:
> > I tried today 2.2.9 kernel with d-axp-229-4.gz patch on LX.
> > After fixing minor typos in Takara code everything compiled but
> > this kernel did not boot.  It printed a bunch of hash marks
> > and the machine froze.
> > 
> > Identically configured 2.2.9 kernel but patched with d-jay-229-2
> > booted without any problems.
> > 
> > Hope that this gives some information,
> 
> Not really.  Have another go with
> 
>   ftp://ftp.twiddle.net/pub/rth/d-axp-2210.gz
> 
> applied to pre-patch-2.2.10-3.gz.

Does it imply, the stock kernel is not Alpha ready? Oh, man, the
2.0.3x horrors come back. One major area, IMHO, wher Linux *sucks*
(yes, it is s.u.c.k.s) is the stable kernel release policy. Caching
patches for weeks in order to release a mega-patch is asking for an
horde of new bugs. I'm really missing the good old days of 1.0.x and
1.2.x when patches were released every two-three days. Let the last 
stable version will go up to 2.2.1024, but a known patch for a known
bug should should not wait for more than 3 days for incorporation.

But the worst thing is the orthogonal patch spaces: patches for memory
management from Andrea, patches with drivers' fixes from Alan, patches
for AXP from you, assorted patches from other people. In order to set
up a decent production system (not a cheesy desktop box) I have to
keep in mind about dozen of patches for various things and all kinds
of precious hints and tips I see in various places (yours, Jay, are
more than precious).

Sorry for all that bragging - I'm probably just tired. It is extremely 
embarrassing to handle monthly reboots of my Alphas just because
2.0.35 sucks so badly and it sometimes spits alignment errors in
ip_rcv(), and it sometimes hangs the net interface, so I have to
bounce it and that is because nobody (including Linus) cared about at
least keeping the kernel Alpha-compatible and the patches from Jay
never made it in, and yesterday we all saw a magnificent release of
2.0.37 - almost half a year if not more after 2.0.36 (despite the fact 
the patches were floating around all that time) and now we are
probably going to see the same story as with 2.0.30.

Sorry,

-- 
Alexander L. Belikoff
Bloomberg L.P. / BFM Financial Research Ltd.
abel@vallinor4.com, abel@bfr.co.il