Date: Wed, 10 Jul 1996 23:17:28 -0500 (CDT) From: Joe Greco <jgreco@brasil.moneng.mei.com> To: jkh@time.cdrom.com (Jordan K. Hubbard) Cc: hackers@FreeBSD.ORG Subject: Re: Some recent changes to GENERIC Message-ID: <199607110417.XAA26888@brasil.moneng.mei.com> In-Reply-To: <6096.837048463@time.cdrom.com> from "Jordan K. Hubbard" at Jul 10, 96 06:27:43 pm
next in thread | previous in thread | raw e-mail | index | archive | help
>
> > However, read my lips: NEWBIES ARE FRIGHTENED BY THE PROSPECT OF
> > RECOMPILING KERNELS AND OTHER GURUISH CRAP. It needs to work cleanly
> > out of the box if we are to have any market appeal. GENERIC probably
> > runs on 80% of all FreeBSD boxes out there, I would bet. Hey, I even
> > run GENERIC on some of my systems. It's convenient. It's flexible.
> > It's the default, too.
>
> Ok, Joe, I'm going to put back sio2 and sio3 for you, OK?
I don't appreciate your tone.
How about you do it for US, the people who will have to deal with your
arbitrary implementation decisions on a daily basis?
> But I'll also tell you that your argument above is more or less total
> crapola. Doing their own kernel build is HARDLY something that I'm
> even going to spare even a recognisable fraction of those users who
> will be forced in 2.1.5 to build their own kernels, whether or not
> sio2 and sio3 are back (you never owned a sound card, Joe?).
No, I never have, although at one point somebody did give me a broken
SoundBlaster, if you really want to be picky. Several locals I have
talked to run GENERIC. Several run custom kernels. Oh.. wait..
what's this.. my home machine.
FreeBSD 2.1.0-RELEASE #0: Thu Nov 16 10:47:14 1995
jkh@westhill.cdrom.com:/usr/src/sys/compile/GENERIC
Wow, I'm running GENERIC. Why? Because it works. I can run X. I can run
SLIP or PPP to the office. It does all the things I need it to do. It's
not something I'll have to remember how I customized if I ever have a
problem and lose my disk. But that's right, Jordan says everyone else in
the Universe builds their own custom kernels.
> I
> furthermore seriously doubt that most people will have 4 serial ports
> enabled for general use and your claims that the world is a 4-port
> place are entirely inconsistent with my own experience - I think
> you've fabricated facts rather wildly here in an attempt to support
> your case.
Gee, Jordan, I count a lot of agreement in the replies I've seen. The big
winner appears to be the case of the Packard-Hell class computer with its
two built in serial ports and an internal modem. That's a combination I
have personally seen on a LOT of PC's, and I wasn't even the one to think of
that obvious scenario. So I'm _obviously_ fabricating facts.
> In other words, I don't think it's going to make any truly TANGIBLE
> difference in the amount of trouble people go through with custom
> kernels, and your arguments miss the mark entirely if you think you've
> somehow saved the world by getting me to bring back two lousy serial
> devices.
I agree, it will not make a bit of difference to the people who go through
the trouble of custom kernels. However, you're still missing the boat!
Many people choose _NOT_ to go through the trouble of building a custom
kernel. That's the situation we are discussing. For the people who choose
- for whatever reason - to stick with GENERIC, it offers more instant
flexibility in what they can do.
And I can picture several reasons for sticking with GENERIC:
Newbie, afraid to change the heart of his system.
Slow machine (build a kernel on a 386sx/16 once and see what I mean).
Low memory machine (build a kernel on a 4MB machine and see what I mean).
A quick, temporary install which will be deleted as soon as a project is
done with.
It's known to work.
It supports all common devices by default.
Not everyone wants to invest the time to set up and configure a box in an
"ideal" fashion.
> However, you seem to have a lot of passionate feelings where this is
> concerned so I'll tell you what: Do something genuinely meaningful
> about it rather than thinking that you've done your side of the
> argument proud simply by getting me to put two silly serial devices
> back. What would be meaningful?
>
> Write a kernel configuration front-end that does it all via
> configuration menus and popup dialog boxes, or set about finishing the
> infrastructure that Julian started in /usr/src/release to allow
> independant folks to easily build custom boot floppies for users with
> sound cards, frame grabbers, what have you. In short, direct all of
> those strong feelings in a more productive direction that genuinely
> moves us FORWARD on this issue and I'll say to myself "My my, so Joe
> Greco actually *does* put his money where his mouth is! I guess I was
> totally wrong about him being just yet another whining pedant!" :-)
Jordan, shove it where the sun don't shine.
I've visited many subsystems, made many improvements, done much
experimentation, submitted lots of patches, and had most of them
ignored or postponed for very long periods.. over the years, I've
submitted patches for all sorts of things, including the netboot
system, disk performance accounting, Kerberos, Kerberized telnet,
user-mode PPP, startslip (several times), some obscure C library
errors, getty, TCP/IP keepalive stuff, and numerous other things
that I've long since forgotten.
Of the above list, the only patch that I'm aware of that anybody
bothered to do anything with was my initial 2.0A startslip patch,
which you applied. I may be forgetting some, but the rate at which
patches that I've submitted have been committed has been abysmal.
When I have seen something I didn't like, I've fixed it, and submitted
it for inclusion so that the rest of the world can benefit too.
I call that a "productive direction that genuinely moves us FORWARD"
but obviously nobody's been paying attention.
Now that I've gotten tired of being ignored, I'm less enthusiastic
about submitting what I consider to be "neat feature" patches.. like
the PPP autodetecting getty code that I tailored so neatly and easily.
Maybe when you learn to take notice of the work a person has done,
and perhaps commit various portions of it, that person would be more
willing to work on larger projects. Personally I don't have time to
piss away on the project you suggested, particularly if it's going to
sit in some mouldy old bin someplace after submission, waiting to be
processed at some point when it is totally out of date.
It's like pulling teeth.
I don't particularly see why I should have to go write a whole new piece of
code to validate myself in your eyes anyways, since I am simply arguing to
maintain what most people would consider to be the status quo. If you asked
me to write code to allow config to handle a "disabled" parameter, that
would be quite appropriate.
As for you, I'll just say this: as president of this organization, you are
out of touch with what those of us who are in the real world are doing.
That is a massive liability for the FreeBSD Project. Lately, I've seen so
much disregard on your part for the rest of the world. You are bullheaded.
You go ahead and make arbitrary changes without appropriate discussion
beforehand. Three days before -RELEASE, in this case. I don't understand
why you think you should be able to do that.
There has clearly been a noticeable percentage of people besides me who have
objected to this arbitrary change. There have been several people who have
indicated that they believe you are misinterpreting the meaning of GENERIC.
If you wish to consider me a "whining pedant", that's fine, I really could
care less, but I don't seem to be alone.
... Joe
-------------------------------------------------------------------------------
Joe Greco - Systems Administrator jgreco@ns.sol.net
Solaria Public Access UNIX - Milwaukee, WI 414/546-7968
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199607110417.XAA26888>