Re: [PATCH] 2.0.35: updated Jumbo -8 patch

Peter T. Breuer (
Mon, 27 Jul 1998 12:06:24 +0200 (MET DST)

"A month of sundays ago Andrew Derrick Balsa wrote:"
> >>
> >Thanks .. I am looking for sources of patches that may allow me to stop
> >maintaining my old 2.0.25 sources! I have looked at yours in the past. Very
> >nice.
> Thanks :-)

> [list of 19 features snipped]
> >
> >How does that compare with your list?
> Sounds like a wish list for 2.2.x to me. :-) The people who worked on the Jumbo
> patch (myself included) are much less ambitious. We are just trying to get some
> of the already present features in 2.0.x to work with more hardware, and fix
> one or two bugs, while keeping 2.0.x stable and clean.

I see. But I would be interested in knowing (or persuading) if you have
any of the patches I listed included, so that I could base my maintenance
on your 2.0.3?++ stuff.

My central problem at the moment with adopting the 2.0.3? series is
lack of support for atm, e2compr and dmsdos. Maybe paride is already
dealt with. The trouble is that I can't add support for these myself to
2.0.35 without risking that 2.0.36 will break interfaces that I depended
on, thus invalidating the work I put in. I'll analyze this more in
terms of risk analysis for my own satisfaction later, but at the moment
it appears easier for me to backport changes in later kernels to my own
older kernel than to try and keep patching extra drivers forward onto
newer kernels. That way I only have one variable to control. The patch.

> >our mastery of multimedia technologies. That means newest
> >whoodangledinks installed in every machine when the money comes in every
> >year, and either linux makes it work, or the machine is lost to
> >windows95 mode forever.
> Hmmm, I can't help but comment on that last paragraph. There is a
> common misconception which consists in comparing OS's based on the feature set.
> This is a special case of the "Comparing Apples and Oranges" logical mistake
> (or sophism, depending on the original intentions of the author).

Unfortunately nobody is comparing OSs. If they were, linux would win.

The simple fact is that if linux doesn't support the newest woodinkle,
then we have to go into an OS that does support the drivers provided by
the manufacturer in order to run the wodinkle and give classes on it and
demonstrate it to visiting chancellors and so on ...

> I use GNU/Linux because it is Free (in the GNU/GPL sense), not because it has
> features a, b or c.

Me too. That's important to me. Unfortunately, the cost to me of
providing the apples and/or oranges to sample is immaterial to the
customer, in this case a visiting chancellor or class of no-hopers.
I can't say "hang on a moment while I write a driver, user interface
and demonstration application" for my morally wonderful OS!

> >distribute more than one basic kernel. The only exception I make is for
> >scsi-based file servers, which have a separate kernel.
> Then again I suggest you check 2.2.x, whenever it comes out. :-)

Of course. But from experience, I won't be moving to 2.2.* until it is at
least at 2.2.28 :-).

> Cheers,
> --
> Andrew D. Balsa

I simply put up my list of 19 "essentials" as a checklist for alan cox and
yourself to look at and think about. I'd love it if more of those 19 points
could make it into both your work.


To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
Please read the FAQ at