Subject: Linux-Development Digest #786
From: Digestifier <Linux-Development-Request@senator-bedfellow.MIT.EDU>
To: Linux-Development@senator-bedfellow.MIT.EDU
Reply-To: Linux-Development@senator-bedfellow.MIT.EDU
Date:     Fri, 3 Jun 94 05:13:05 EDT

Linux-Development Digest #786, Volume #1          Fri, 3 Jun 94 05:13:05 EDT

Contents:
  Kernel change summary 1.1.16 -> 1.1.17 (Russell Nelson)
  Frustrated with new kernels (Elaine Walton)
  Re: Handwriting recognition in Linux? (yes (Elaine Walton)
  Appletalk interface options? CAP? UAR? (Gasparian)
  Re: 1.1.15 breaks SCSI (Eric Youngdale)
  gas, gcc, and i386's (Gordon William Matzigkeit)
  Re: Frustrated with new k (Elaine Walton)
  Can't compile latest ncurses for Linux (Sam A Gentile)
  Re: Linux game development (Was Re: Why [DOS, W (StO)
  Re: Frustrated with new k (Janne Sinkkonen)
  Re: Frustrated with new kernels (David Dwayne Wiaizenegger)
  Re: Frustrated with new kernels (Bjorn Ekwall)
  Re: Frustrated with new kernels (Lee J. Silverman)

----------------------------------------------------------------------------

From: nelson@crynwr.crynwr.com (Russell Nelson)
Subject: Kernel change summary 1.1.16 -> 1.1.17
Date: 02 Jun 1994 00:13:37 GMT

TCP/IP changes only:

Added support for a dummy net driver, to be used when you're disconnected from
        the net.
Added support for unused ethX slots, so that Ethernet devices could be
        removed and *added back into the same slot* (for modules).
WARNING: The PRE 1.1.16 plip will NOT work with the 1.1.16+ PLIP driver.
Fragmentation now works.
Missing byte order conversion was added for detecting loopback.
Fixed another RST bug (avoid unnecessary RST after FIN).
BSD UDP semantics *removed* to comply with RFC1122.
--
-russ <nelson@crynwr.com>      ftp.msen.com:pub/vendor/crynwr/crynwr.wav
Crynwr Software   | Crynwr Software sells packet driver support | ask4 PGP key
11 Grant St.      | +1 315 268 1925 (9201 FAX)    | Quakers do it in the light
Potsdam, NY 13676 | LPF member - ask me about the harm software patents do.

------------------------------

From: ewalton@magnus.acs.ohio-state.edu (Elaine Walton)
Subject: Frustrated with new kernels
Date: 2 Jun 1994 01:54:19 GMT

I have tried to compile the new kernels from off of tsx-11--no luck.
I am really confused how others are able to compile, yet I am not.
Maybe it's because I'm trying to go from 1.0.0 directly to 1.1.13-6?
Either way, I have pretty much given up..
-Sean

------------------------------

From: ewalton@magnus.acs.ohio-state.edu (Elaine Walton)
Subject: Re: Handwriting recognition in Linux? (yes
Date: 2 Jun 1994 02:01:11 GMT

In article <Cqq1yM.A2H@pts.mot.com>,
Kevin Burtch X8534 Ppppp <kburtch@pts.mot.com> wrote:
>In article 10876@cs.rit.edu, kxg7283@cs.rit.edu (Kiriakos X Georgiou) writes:
>> Yes, I have done work on this, but I guess it's not in any way limited to
>> Linux, since it's C source code. (It works on Linux and SunOS so far)
>
>Great! I'm glad someone has already started work on it! Now if only there was
a way
>to generate more interest in it...   Any hackers with NotePads out there???  :


I'd be willing to test the work--I have a pen-top, but I do not have the time
to develop (though I'd love to do so).
-Sean

------------------------------

Crossposted-To: comp.os.linux.misc
From: gaspo@netcom.com (Gasparian)
Subject: Appletalk interface options? CAP? UAR?
Date: Thu, 2 Jun 1994 16:13:27 GMT

Greetings all!
    I am currently trying to integrate a mac-appletalk-TP network to our
linux server, but have run into that money barrier.  If we had an extra
$1500 in the budget, I would just get a Shiva or Gatorbox to take care
of the AP<->ether connection, and CAP and UAR to handle the routing.
But, since I _don't_ have that extra cash, does anybody know of an
alternative solutions to the hardware link?  Has anybody got CAP to work
with an old PC-TOPS card or other cheaper Appletalk interfaces?

    Basically, I want to have mac-tcp running on our old macplus (i.e.,
no ethernet), and my 486 linux box doing the Fastpath buisness.

--gaspo.

-             Scott "gaspo" Gasparian                - Umwelt    -
- Instructor/SysAdmin     - Product R+D Director     -  Schutzen,-
- Lick Wilmerding H.S.    - Gaspo's Gadgets Pty Ltd. - Rad       -
- 755 Ocean Ave, S.F.,CA. - gaspo@netcom.com         -  Benutzen.-



------------------------------

From: ericy@cais.com (Eric Youngdale)
Subject: Re: 1.1.15 breaks SCSI
Date: 2 Jun 1994 02:20:49 GMT

>> >The kernel 1.1.15 does not work at all with my Adaptec 1540B.
>> >I get "Unable to reset SCSI host 0, probably a SCSI bus hang."

        OK, I think I understand what is going on.  I have come up with 
some patches which should correct this and a few other minor nits that 
have recently been introduced, and I have uploaded this to tsx-11 in 
pub/linux/ALPHA/scsi/scsifix-1.1.17.diff.gz.   If you experienced any 
trouble with 1.1.16/1.1.17, please try this and see if it helps.

        I have not sent this to Linus yet - I would like some direct 
confirmation that it actually helps first :-).

-Eric

-- 
"The woods are lovely, dark and deep.  But I have promises to keep,
And lines to code before I sleep, And lines to code before I sleep."

------------------------------

Crossposted-To: gnu.gcc.help
From: gwmatzig@acs.ucalgary.ca (Gordon William Matzigkeit)
Subject: gas, gcc, and i386's
Date: Fri, 3 Jun 1994 02:01:18 GMT

Hi, fellow GNUsers!

I was wondering if there is a good tutorial out there for i386 gas
programming.  I am a MS-DOG a86/TC programmer, and I am migrating to
Linux/NetBSD.  Any pointers/tips would be much appreciated.

My project is as follows:

I have implemented a DOS program which fools with the DOS boot sector
and hard drive partition in order to install an interrupt handler
before DOS loads.  The idea is that this handler hides in the top of
memory while DOS is running, and filters disk write/format requests to
the hard drive.

Net result: DOS can be run on a machine without having to worry about
users who might accidentally write to other OS partitions, or to vital
system files.  This allows sysadmins to make machines with, say, a
Linux partition next to a DOS partition, turn the users loose on the
system, and not worry about the consequences of an accidental "format
c:" or "fdisk."  My implementation is transparent to DOS programs
(heck... it even runs under Windoze!), to the best of my knowledge.

Due to the restrictive and proprietary nature of the DOS development
environment, I'd like to port my system to i386 *nix, with GNU tools,
so that the GPL that I put on my software can be put to good use.
i.e. everybody with access to an i386 and GNU tools can help
contribute to and debug the system.

What I need to know are such *nix fundamentals as:

1) How do I write a stand-alone gas program?  I'm currently using Eric
Isaacson's (happily registered) A86/D86 package, and I want to
translate my skills into gas.

2) How do I link OS-independant code under Linux gas?  i.e. in A86, I
can produce a .bin file, copy it to the boot sector, and the machine
executes it just peachy.  Is there a gas equivalent?

3) How do I interface gcc and gas routines?  Yes, I know it's a FAQ,
but I need to know if it's any different than the "byte in AL, word in
AX, args on stack" method.

Thank you very much for your time and bandwidth.  I hope to release
the system soon, so any prompt info is appreciated.

--Gordon Matzigkeit
  gwmatzig@acs.ucalgary.ca

------------------------------

From: ewalton@magnus.acs.ohio-state.edu (Elaine Walton)
Subject: Re: Frustrated with new k
Date: 3 Jun 1994 01:09:16 GMT

I've done that, and it had the reverse effect of what I expected: no one
used it.  I wrote a tool which will automatically calculate the ModeDB
table for a multisync monitor.  It works, but one has to be careful.

I think that tsx-11 has some corrupted kernels.  I was directed to
download a 1.1.17 kernel (prepatched) from linuxftp.caltech.edu.  I did
so, and it's working great.  Who's at fault?  I don't care.  I just
wanted to play with a Linux-sabre.... ;)

-Sean

------------------------------

From: sgentile@world.std.com (Sam A Gentile)
Subject: Can't compile latest ncurses for Linux
Date: Fri, 3 Jun 1994 02:51:05 GMT

I can't compile the latest ncurses package for Linux. I run Configure, select
Linux and then run make on the resulting makefile. I get a good way's thru it and then:
gcc -I. -Wall -O6    -DBROKEN_TCDRAIN -DSRCDIR='"/usr/lib/terminfo"'    -c lib_move.c -o lib_move.o
lib_move.c:15: parse error before `x'
lib_move.c: In function `wmove':
lib_move.c:16: number of arguments doesn't match prototype
curses.h:251: prototype declaration
lib_move.c:22: `x' undeclared (first use this function)
lib_move.c:22: (Each undeclared identifier is reported only once
lib_move.c:22: for each function it appears in.)
lib_move.c:22: `win' undeclared (first use this function)
lib_move.c:23: `y' undeclared (first use this function)
make: *** [lib_move.o] Error 1

Has anyone gotten this to compile and how?

Thanks,
Sam


------------------------------

From: sto@myhost.subdomain.domain (StO)
Subject: Re: Linux game development (Was Re: Why [DOS, W
Date: 2 Jun 1994 02:06:58 GMT

: 1. Develop a version of LILO that runs as a DOS executable and can boot
:    a Linux kernal image from a DOS file.

It's been around .. Called loadlin. It can load a kernel from a DOS prompt.

- Brian Stoler
- sto2@netaxs.com
- Having problems with talk over slip
- Having problems with POP3 over slip

------------------------------

From: janne@avocado.pc.helsinki.fi (Janne Sinkkonen)
Subject: Re: Frustrated with new k
Date: 2 Jun 1994 22:57:31 +0300

In article <1.8643.2382.0N27CD41@dscmail.com>,
Rick Emerson <rick.emerson@dscmail.com> wrote:

>Anyway, the original post points out an ongoing problem: Where is the
>latest *reliable* kernel?  Aside from the 1.0 announcement, I can't
>recall another notice saying, in effect, "OK, gang, this kernel works
>pretty well and won't surprise you."  

I have seen an announcement by L Torvalds which says that v1.1.x are
leading edge kernels, while v1.0.y are stable. From that, one could
conclude that the most stable kernel is the v1.0 version with the
highest patch level (9, is it?).

>So it *is* frustrating to think there's a later version of the kernel
>that might be a little quicker or cleaner but not which one it is.

What is frustrating here is that you _know_ that there exists a kernel
with more features than yours, not necessarily cleaner
(whatever that means) or more stable. If linux were commercial, you
would not have that temptation to get the latest alpha/beta version,
because it would simply not be available.

A good strategy is maybe to stick with one distribution, maybe with a
commercial one (so that the distributor has something to lose), and
update the system when new versions of that distribution becomes
available. I don't know what is the state of present distributions
considering updating the system without destroying everything.  I have
used this strategy since linux 0.2 (believe me things are easy
nowadays... :-)) with a lot of all kind of piecewise updating as long
as I'm able to. Now we have 1.1.17 in all three machines because it
has better SLIP support & better buffer caching. It is stable enough.

This still takes a lot of hacking time, and sometimes I think it is
not worth it (maybe addiction is the most real thing here :)). But you
surely learn something, too.

--
Janne


------------------------------

From: Dave.Waizenegger@launchpad.unc.edu (David Dwayne Wiaizenegger)
Subject: Re: Frustrated with new kernels
Date: 3 Jun 1994 01:16:26 GMT


Guys, the entire 1.1.x tree is for development/ALPHA testing!  The 1.0.x
tree is nearly perfectly stable, and won't cause you any problems of the
sort you refer to.  The 1.1.x tree is constantly evolving, so if you want
stability stay away from it.  Nobody is forcing you to get those patches
and recompile, so quit complaining anddevelop/test or use what you shoulld
be using.  
                                Cheers,
                                        Dave

--
==============================================================================
 \ The above does not represent OIT, UNC-CH, laUNChpad, or its other users. /
   ------------------------------------------------------------------------

------------------------------

From: bj0rn@blox.se (Bjorn Ekwall)
Subject: Re: Frustrated with new kernels
Date: 3 Jun 94 01:38:12 GMT

I'm making this post, not because I'm picking on anyone in particular,
but because I think it is important to refresh some posters about the
different objectives with the 1.0.X vs. 1.1.X.

The 1.0.X versions target those who want a "working and stable"
system.  The 1.1.X versions are explicitly named "hackers paradise",
by none other than our beloved Linus himself!
The wisdom behind this decision is greater than what you might think...

I consider myself a "hacker", which means you will find me
somewhere near the latest 1.1 patchlevels, since that is where
I like to be.  There are others who find their "fullfillment" in 
making the 1.0 series solid, and also some very impressive documenters
that make the whole thing understandable, and so on, and so on...
And yes, if you decide to be "just a user", I do find that to be a
very respectable position too... Especially if you make an honest
effort to learn about the things you want to use...
But that is your decision and responsibility, not mine or anyone else's!

What some people seem to forget, or perhaps never even reflected upon,
is the fact that Linux development is made in the open, with input
from _anyone_ who finds this interesting.  There is _no_ central authority
that decides _what_ should be developed, by whom, and within what deadline.
This means that the patchlevels we see "daily" should not be seen as the
final fixes to the last remaining bugs.
Far from it!
If you are mean-spirited, you could say that the idea behind these patches
is to create _new_ bugs :-) :-) :-)
The real reason is of course that _no-one_ knows what Linux will look
like when it is finished, and the reason for this is that no-one has
the authority to decide _when_ Linux is finished!!!

What differs from what most of us has seen before (including me) is that
new and/or alternative ideas and suggested solutions are presented "to the
public" at a _much_ earlier stage than what one traditionally expects.
One other unusual circumstance is that the Linux developers are
contributing their time and effort _only_ because they enjoy it,
and because it gives them a platform for trying out new ideas,
with early input from their "peers", wherever they are.

This means that Linux is _not_ developed primarily for the benefit of
the "users", but because the developers actually enjoy it, immensely!
The wonderful side-effect of this is that the "users" will get
a continously evolving "product" that, as time goes by, has been more
thoroughly "scrutinised" than most other "products" ever have been.

So, the "product" you receive and enjoy is really a side-effect of
the "process" that the developers enjoy...
You can actually _see_ the enormous flow of ideas. Some of them will
be included in the "main-line" releases, and some will live their own
life as potential powerful additions for anyone who is interested.
Just take a look in some of the mailing lists, or under the ALPHA and BETA
directories at your favourite ftp provider!  All those files are "Linux" too!

We are fortunate enough to have some volountary coordinators, "elected"
by virtue of their deeds, with Linus at the central point of focus.
His results with the herculean task of keeping this constantly moving
beast in an amazingly sane and coherent condition is _very_ impressive.
The same goes for the other coordinators. The success of a coordinator
depends both on his/her expertise and on the willingness to listen and
to be receptive to the ideas of other people.
It is possible to try to develop your own Linux all by yourself, but you
will _never_ be able to stay ahead of, or even keep up with, the results
that the combined developers/coordinators create.
Some attempts in this direction has been seen, but the power of the
"open development" has won every time (so far :-) ).


Well, it seems that Linux developers of all shades have all the fun...
Perhaps there is some constructive way for the "users" to join in
with their own "fun" too?
Whines from users who haven't made an honest attempt to learn by finding
answers in the available doc's is not considered constructive in my book
of rules; _substantiated_ bug-reports, whishes and ideas, OTOH, are.

One of my collegues (in another company, though) once had the practice
to have a clause in his contracts that allowed him to charge his customers
_double_ if he could show them that the answer to their question could be
found in the user documentation...
Sometimes I whish that practice could be used here too :-) :-)

Well, why not?  From now on, everyone who asks a question that is
answered in the FAQ or HOWTOs will be charged with an extra amount,
equal to the amount he already had to pay the Linux developers!

Virtual invoices fits Linux just as well as virtual beer, right? :-)


Bjorn Ekwall == bj0rn@blox.se

------------------------------

From: ljs@cs.brown.edu (Lee J. Silverman)
Subject: Re: Frustrated with new kernels
Date: 2 Jun 94 13:20:51


        Someone posts complaining about the fact that Linux Kernels
and versions of Libc change very often, often breaking existing
programs or not working at all.

        My first response is to tell you to look at the rest of the
industry: If you want a stable, backwards compatible configuration,
spend several million dollars and buy an IBM Mainframe running VM/ESA;
they're backwards compatible to 1965 (maybe earlier).  

        If you want Unix, spend several thousand dollars and get a
Sparcstation.  Keep in mind that the upgrade from SunOS 4.x to Solaris
broke almost every program in existence.  Or you could spend several
tens of thousands of dollars and get an RS/6000.  Keep in mind,
though, that the upgrade from AIX 3.1.x to 3.2 broke a lot of things
also.  Maybe you'd find Ultrix more stable (last I knew, Ultrix
upgrades didn't break lots of things) but keep in mind that many
analysts think that DEC is going down the tubes, so you may be stuck
with no upgrades at all.

        In short, you'd be hard pressed to find a stable version of
Unix from a commercial vendor.  HPUX strikes me as rather stable, as
does Irix, but you pay through the nose for SGI's or HP's.  Then
there's SCO...

        Second, you get what you pay for.  Linux Kernels and software
are developed by very skilled people who have donated their spare time
to the concept of free software.  They are doing their best, with
limited resources and limited time and zero budget, to create a really
good package.  If you want a stable version of Linux, find a kernel
version and a distribution you like, install them, make sure all your
binaries work, and then don't upgrade.

        Lastly, Linux is still very young and very much in the
development stage.  When you install Linux on your machine, the
documentation alone should be indication enough that there is still
much development going on.  Not all of this development will be
backwards compatible.  Not all of it will be filtered down to a level
where the non-developers can understand it.  I certainly don't
understand the point of a number of the Kernel mods, or why Libc 4.5
couldn't be backwards compatible.  I have faith in the people
developing Linux that they know what they're doing.  If you don't have
that kind of faith, buy a time-tested commercial package and then take
your chances with the vendor.
--
Lee Silverman, Brown class of '94, Brown GeoPhysics ScM '95
Email to: Lee_Silverman@brown.edu
Phish-Net Archivist: phish-archives@phish.net
"Nonsense - you only say it's impossible because nobody's ever done it."

------------------------------


** FOR YOUR REFERENCE **

The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:

    Internet: Linux-Development-Request@NEWS-DIGESTS.MIT.EDU

You can send mail to the entire list (and comp.os.linux.development) via:

    Internet: Linux-Development@NEWS-DIGESTS.MIT.EDU

Linux may be obtained via one of these FTP sites:
    nic.funet.fi				pub/OS/Linux
    tsx-11.mit.edu				pub/linux
    sunsite.unc.edu				pub/Linux

End of Linux-Development Digest
******************************
