Subject: Linux-Misc Digest #611
From: Digestifier <Linux-Misc-Request@senator-bedfellow.MIT.EDU>
To: Linux-Misc@senator-bedfellow.MIT.EDU
Reply-To: Linux-Misc@senator-bedfellow.MIT.EDU
Date:     Mon, 15 Aug 94 10:13:36 EDT

Linux-Misc Digest #611, Volume #2                Mon, 15 Aug 94 10:13:36 EDT

Contents:
  Re: WABI vs. SoftWindows? (Neil Dunbar)
  IDEA: termfs - NFS-like fs for Term (Sverre H. Huseby)
  Re: XON/XOFF (again) and "efax" HELP!!!! (Rob Janssen)
  Re: starting X automatically on installing linux distribution (Rob Janssen)
  Re: MCA SCSI Adapters for Linux? (John R. Campbell)
  Re: Where did "Yggdrasil" come from? (John R. Campbell)
  Re: Multi-threaded linux-kernel (Bouwmeester L.)
  Re: WABI vs. SoftWindows? (Dan Pop)
  "Ftp by email" server for Linux files wanted (B.A. Doehner)
  Re: Suggest:SCSI Tape File System (Christopher Cason)
  Re: WABI vs. SoftWindows? (Brandon S. Allbery)
  Re: WABI vs. SoftWindows? (Mark A. Davis)
  Video Cards Supported by XFree (Rich Larsen)
  DCE on Linux (David Jones)
  Re: WABI vs. SoftWindows? (Tom Krotchko)
  Re: WABI vs. SoftWindows? (Tom Krotchko)

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

Crossposted-To: comp.unix.unixware
From: neil@riscot.demon.co.uk (Neil Dunbar)
Subject: Re: WABI vs. SoftWindows?
Date: Mon, 15 Aug 1994 10:40:39 +0000

tomk@access.digex.net (Tom Krotchko) writes:

>In article <32fvne$50n@hawk.ee.port.ac.uk>, mike@hobbit.hum.port.ac.uk (Mike Meredith) says:

>[snip] A
>64 user license of Sybase is somewhere between $6000-9000.  The cost of the
>hardware to run it is another $15,000 (not counting the physical network).

>Whether an OS costs $0.00 or $1,000 doesn't seem all that important.

I'm not so sure, Tom. That's a question of paying $21,000 or
$22,000. If a customer is presented with two choices which are
equivalent OS's, which one would he choose? After all, that's $1000
better off in his pocket than his system suppliers. In certain
industries (like the company I work for), that $1000 can make the
difference between getting and not getting a contract. Of course, the
all important assumption here is whether the OS's are *really*
equivalent, but that's already been done to death in this thread.

Of course, using Linux instead of SCO, Unixware et al, removes the
uncertainty of what happens if SCO, Novell and Sun were all to go
belly up .... ;-)

Neil
-- 
\ Neil Dunbar [Alleged Software Engineer]/   Rochester Instrument Systems, Ltd.
 \ E-Mail : neil@riscot.demon.co.uk     /            Hillend Industrial Estate,
  \ CI$: 100066,603                    /                Dunfermline, Fife, U.K.
   \ "I made the blue car go away"    /                    Tel : +44 383 822911

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

From: sverrehu@ifi.uio.no (Sverre H. Huseby)
Subject: IDEA: termfs - NFS-like fs for Term
Date: 15 Aug 1994 13:33:09 +0200


How about making a new filesystem called termfs, that supports
mounting a directory on a computer accessed by term?

[Please let me know if that kind of mounting is already avail.]

If this was the start of the summer rather than the end, I would
have looked at it myself. Maybe for x-mas...


Sverre.

===========================================================
Sverre H. Huseby                                    Student
sverrehu@ifi.uio.no              University of Oslo, Norway
http://www.ifi.uio.no/~sverrehu

My employer (that's me) is not responsible for my opinions.



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

Crossposted-To: comp.os.linux.help,comp.os.linux.admin
From: rob@pe1chl.ampr.org (Rob Janssen)
Subject: Re: XON/XOFF (again) and "efax" HELP!!!!
Reply-To: pe1chl@rabo.nl
Date: Mon, 15 Aug 1994 08:32:29 GMT

In <32m5je$qmn@inferno.mpx.com.au> jhonan@jolt.mpx.com.au (Jamie Honan) writes:

>Rob Janssen (rob@pe1chl.ampr.org) wrote:
>: In <32ea0d$jm8@inferno.mpx.com.au> jhonan@jolt.mpx.com.au (Jamie Honan) writes:

>: >BTW, 1.1.42 (for me) excercises a limitation in efax. My modem
>: >has large buffer sizes, and I suspect more data is being buffered
>: >in the serial driver than previously. The author of efax is on holidays, so
>: >I am taking the liberty of posting a small diff here, in case others
>: >face a smilar problem. It is more noticable when connecting to fax devices
>: >operating at low speeds (2400 and 4800 bps devices).

>: >The author has assumed a maximum fixed time for the flushing of output data.
>: >My patches are to simply calculate what that time should be more accurately.

>: Indeed the new drivers buffer more data on output, I have also seen this
>: with another application.
>: However, the fix for this should IMHO not involve tricky timing.  You can
>: wait for all data being physically sent to the serial port using a
>: tcdrain(fd), and you will not be dependent on tricky timing issues.

>I'd wholeheartedly agree, if you could be sure of the buffer size
>of the modem.

>The transmission timing in fax is based on the negotiated bit rate, 
>independant of the interface rate between the modem and the serial port.

>The author took a punt on 4K as being the buffer size. A modem could
>equally have 20K. (most don't, simply because WinFax makes the same
>assumptions)

Of course there are two issues here:

1 - buffering in Linux

2 - buffering in the MODEM

The first is easily overcome by the tcdrain(fd) call.  I think it was the
problem you had, because you described it occurring after upgrading the
kernel.  And indeed the size of the output buffer has increased in Linux
with the introduction of the new tty drivers.

The second is more difficult, of course.  Any reasonable protocol for
FAXing would incorporate some way of knowing when the entire buffer has
been sent and the MODEM starts waiting on the remote acknowledgement,
but apparently Class-2 FAX is not that reasonable...

On the other hand, it should be safe to use no timeout at all, and just
monitor the DCD line and watch for error codes coming back.  The MODEM
should always at least drop DCD when the transmission has irrecoverably
failed or deadlocked.
However, I have read before that some modems have firmware bugs, and
can hang on to the line forever with no indication of error.  That is
probably the reason why the timeout was built in...

Rob
-- 
=========================================================================
| Rob Janssen                | AMPRnet:   rob@pe1chl.ampr.org           |
| e-mail: pe1chl@rabo.nl     | AX.25 BBS: PE1CHL@PI8UTR.#UTR.NLD.EU     |
=========================================================================

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

From: rob@pe1chl.ampr.org (Rob Janssen)
Subject: Re: starting X automatically on installing linux distribution
Reply-To: pe1chl@rabo.nl
Date: Mon, 15 Aug 1994 08:56:25 GMT

In <$oiJkukh1iK0069yn@shore.net> bjb@shore.net (Beverly J. Brown) writes:

>X is much more pwerful than Windows or OS/2. But try convincing a Linux/X 
>novice who is dismayed that he can't put xclock in the upper right corner of 
>the screen and expects it to be there the next time he starts X.

Actually, fvwm can save the state of the desktop in a file that you could
merge with your .xinit
It should be possible to extend this scheme to actually use the saved file,
and to have similar mechanisms to set and save other parts of the desktop
setup....
I agree this is something that is needed to make X more newbie-friendly!

Rob
-- 
=========================================================================
| Rob Janssen                | AMPRnet:   rob@pe1chl.ampr.org           |
| e-mail: pe1chl@rabo.nl     | AX.25 BBS: PE1CHL@PI8UTR.#UTR.NLD.EU     |
=========================================================================

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

Crossposted-To: comp.sys.ibm.ps2.hardware,de.comp.os.linux
From: soup@penrij.UUCP (John R. Campbell)
Subject: Re: MCA SCSI Adapters for Linux?
Date: Sun, 14 Aug 1994 12:43:07 GMT

jesusr@vnet.ibm.com writes:

>In <327nnj$ma1@news-feed-1.PeachNet.EDU> wmiller@vyonous.kennesaw.edu (Wade T. Miller) writes:
>>Klaus Kudielka (kkudielk@cacofonix.nt.tuwien.ac.at) wrote:
>>: I have already found two adapters which MIGHT be compatible, but I am not
>>: sure:
>>
>>: 1) Adaptec AHA-1640. Is it 100% compatible to the AHA-1540? Does anybody
>>:    use it with Linux? 
>>
>>: 2) Future Domain MCS-600. It has the TMC-1800 chip. Is it 100% compatible
>>:    with the Future Domain 16x0? Does anybody use it with Linux?
>>
>>It is doubtful that either MCA controller will work with linux now or
>>ever.  MCA is a proprietary bus architecture and IBM is not giving out
>>the specs without a non-disclosure agreement.  Hence the source for the 
>>driver cannot be released without violating that agreement.  Invest in a 
>>more compatible motherboard.

>WRONG!  The problem has nothing to do with MCA.  The problem is exactly what
>Klaus said, he either needs a new SCSI adapter or write a driver for the IBM
>SCSI adapter.  I had Linux boot on my 57SLC2 (MCA machine) and it found my
>diskette drives, my SB PRO Micro Channel Version, but it failed on the SCSI
>probe.  So I assume it's just a SCSI driver problem.  I've been trying
>to gather the information to write the SCSI driver but can't find the TIME
>to do it.

The AHA-1640 works just fine in an NCR MCA machine;  this how I got
LynxOS to run on an NCR system.  LynxOS's migration required me to
snatch the AHA-1540 driver and modify it to link into the MCA version
of LynxOS (which they've since dropped as a product).  I am unsure
whether the IBM will try to boot from the 1640, but (I think) it should.

My biggest peeve is that some drives need to be opened *twice* before
they'll work properly, but these are all 540MB and up.  We even had a
run of SCSI drives that wouldn't boot at all due to a firmware change
on the drives (They were either Maxtors or Conners, I cannot recall
which).

>So I think blaming MCA is no longer a valid excuse.

Agreed, but I suspect that Lynx Real-Time Systems has dropped MCA since
IBM is no longer pushing it.  Of course, I have not yet heard the
*real* reason this has occurred.  *SIGH*

MicroChannel was actually a pretty competent architecture;  I suspect,
however, that it does not compete well against PCI (that being a newer
standard).  If IBM hadn't tried to tighten the box (they over-reacted
after the PC/AT bus got turned into ISA) this might have been seen as
the next bus to migrate to.  See?  Technical excellence has *nothing*
to do with success (consider Motorola vs Intel's bug lists on CPUs
and glue chips).
-- 
 John R. Campbell                                              soup@penrij.UUCP
 Speaker to Machines                                           Resident Heckler
           Guns don't kill people.  GNUs, however, are Not Unix.
        When cryptography is outlawed, bayl bhgynjf jvyy unir cevinpl.

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

From: soup@penrij.UUCP (John R. Campbell)
Subject: Re: Where did "Yggdrasil" come from?
Date: Sun, 14 Aug 1994 18:34:20 GMT

kenbrody@cloud9.net (Kenneth Brody) writes:

>Subject says it all.  Does anybody know where the name "Yggdrasil" comes
>from?

I think it's from some "norse"???  mythology.

As a planetary name it crops up in H. Beam Piper's work (as in the
Yggdrasil Kooghra).

-- 
 John R. Campbell                                              soup@penrij.UUCP
 Speaker to Machines                                           Resident Heckler
           Guns don't kill people.  GNUs, however, are Not Unix.
        When cryptography is outlawed, bayl bhgynjf jvyy unir cevinpl.

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

From: leonb@tyr.research.ptt.nl (Bouwmeester L.)
Subject: Re: Multi-threaded linux-kernel
Date: Mon, 15 Aug 1994 11:46:28 GMT

bsa@kf8nh.wariat.org (Brandon S. Allbery) writes:

>In article <730.2E4E9E79@purplet.demon.co.uk>, jaggy@purplet.demon.co.uk (Mike Jagdis) says:
>+---------------
>| I think that at the moment only the user level is preemptable. Once you've 
>| entered the kernel you are safe until you either return or voluntarily wait 
>| on something. So the kernel level is safe unless you go to either a real 
>| time preemptable kernel or SMP - in which case processes have the same 
>| conflict problems as threads would anyway.
>+------------->8

>Again, a pre-emptable kernel with kernel-level threads is what I thought he
>was talking about.  Kernel support for user-level threads doesn't seem like it
>would need the rewrite he's claiming for Viper, after all.

To remove any confusion about this: two thread-levels are introduced (a la
Solaris): user-level threads and kernel-threads (or lwp's). Actually,
Solaris has three levels as lwp's are not kernel-threads but a data
structure running on top of kernel-threads. 

Anyway, the reason for these two-levels are: Most programmers need logical
concurrency (for instance, think of a window-system), and it may be
necessary to have several thousands (!!) around. Context-switching can be
done in a library without entering the kernel for each invocation. In other
words, the kernel don't know anything about user-level threads and vice-versa.

However, for some applications it may be necessary to exploit the underlying
hardware. Lwp's are introduced to support this and hence, they are heavier
than a user-level thread. The user-level threads are multiplexed on top
of these lwp's, and the low-level scheduler schedules the highest priority
lwp to run on the processor.

Of course, strictly speaking the above is not really necessary, but Viper 1.0
delivers the basis for multi-processor support and/or real-time support.
The first version of the multi-threaded kernel is not pre-emtible yet: we
just provide kernel-threads and user-level threads. Then, depending a bit
on the demand and/or our wishes/capabilities we continue with Viper 2.0:
a pre-emptible kernel. We deliberately have chosen for this approach so
we can develop a real-time kernel (for instance) step-by-step. This is
hard enough as it is without walking through (+-) 1.3 Mb source code
to find all the places where pre-emption should be introduced (miss 
one lock in the filesystem, and the filesystem crashes, miss one in the virtual
memory, and... well you know what can happen :-)).

Leon.

PS. Design of the kernel is near completion. Restructuring of the kernel
data structures has more or less started. Watch the announcements!


-- 
+----------------------------------------------------------------------------+
|Ir. L.H.A. Bouwmeester      PTT Research, Dr Neher Laboratorium             | 
|Phone : +31-(0)70-3325864   Network Service and Control Department, rm E120 |
|Fax   : +31-(0)15-3326477   St. Paulusstraat 4, 2264XZ, Leidschendam        |

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

Crossposted-To: comp.unix.unixware
From: danpop@cernapo.cern.ch (Dan Pop)
Subject: Re: WABI vs. SoftWindows?
Date: Mon, 15 Aug 1994 11:57:48 GMT

In <32hacj$pq1@news1.digex.net> tomk@access.digex.net (Tom Krotchko) writes:

>Not to be flip...I think Linux is great, myself...but the cost of the OS
>on a box that's running Sybase, for instance, is negligable.  A
>64 user license of Sybase is somewhere between $6000-9000.  The cost of the
>hardware to run it is another $15,000 (not counting the physical network).
>
>Whether an OS costs $0.00 or $1,000 doesn't seem all that important.

It's certainly not important for _that_ installation, but if you want
to install the same OS on your home machine, the difference between
$0 and $1000 is not exactly insignificant. Not to mention that the one
which costs $0 comes with complete sources. For some users this is
irrelevant, for others it makes a lot of difference. 

Choose the one which matches best your needs and enjoy it!

Dan
--
Dan Pop 
CERN, CN Division
Email: danpop@cernapo.cern.ch
Mail:  CERN - PPE, Bat. 31 R-004, CH-1211 Geneve 23, Switzerland

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

From: bad1679@ultb.isc.rit.edu (B.A. Doehner)
Subject: "Ftp by email" server for Linux files wanted
Reply-To: bad1679@ultb.rit.edu ()
Date: Mon, 15 Aug 1994 11:22:37 GMT


I am going to be at a site with only email access to the internet for
the next couple of months. Does anyone know of a good ftp mail server
for linux files that I could use from there?

Thanks...  

Bernie Doehner bad1679@ultb.isc.rit.edu

-- 
Electrical Engineering, Rochester Institute of Technology, Rochester,NY
Ax25 Packet: NU1S @ N0ARY.CA.USA.NA  (normal packet route) 
Internet (school): bad1679@ultb.isc.rit.edu

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

From: cjcason@yarrow.wt.uwa.edu.au (Christopher Cason)
Crossposted-To: comp.os.linux.development,comp.os.linux
Subject: Re: Suggest:SCSI Tape File System
Date: 15 Aug 1994 12:24:50 GMT

Chris Burke (clb@prism.mindware.brisnet.org.au) wrote:
: I have been pondering a bit and have worked out a reasonable way to allow
: a mountable file system for high speed tapes. 
[snip]

I think this is an excellent idea. Anyone else have any comments ???

regards,

-- Chris

==============================================================================
| Chris Cason via Univ. of Western Australia : cjcason@yarrow.wt.uwa.edu.au  |
|  Disclaimer : I don't work for/study at UWA. This is a commercial account  |
==============================================================================
|  POV by EMAIL : mail povray@yarrow.wt.uwa.edu.au with word HELP in body    |
|  POV by FTP   : FTP to ftp.uwa.edu.au and cd to pub/povray                 |
|  POV-Ray is a FREE raytracer for DOS, UNIX, VAX, Mac, Amiga, OS/2, etc.    |
|   - check out the images in our HALL_OF_FAME/ and Images_of_the_month/ ! - |
==============================================================================


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

Crossposted-To: comp.unix.unixware
From: bsa@kf8nh.wariat.org (Brandon S. Allbery)
Subject: Re: WABI vs. SoftWindows?
Date: Mon, 15 Aug 1994 12:15:11 GMT

In article <1994Aug15.104039.1877@riscot.demon.co.uk>, neil@riscot.demon.co.uk (Neil Dunbar) says:
+---------------
| Of course, using Linux instead of SCO, Unixware et al, removes the
| uncertainty of what happens if SCO, Novell and Sun were all to go
| belly up .... ;-)
+------------->8

SCO's management problems over the past year are enough to make one nervous...

++Brandon
-- 
Brandon S. Allbery KF8NH         [44.70.4.88]             bsa@kf8nh.wariat.org
Linux development:  iBCS2, JNOS, MH

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

From: mark@taylor.infi.net (Mark A. Davis)
Subject: Re: WABI vs. SoftWindows?
Date: Mon, 15 Aug 1994 12:19:30 GMT

rmk@rmkhome.com (Rick Kelly) writes:

>Brandon S. Allbery (bsa@kf8nh.wariat.org) wrote:
>: Let me tell you folks a little story.

>: I've got the SCO system stabilized for now, but I'm continuing with Linux
>: development in my off hours.  The next time an SCO bug traps us I may have a
>: *viable* escape route.

>I take it that the app is not using X11 on SCO.  While SCO ODT does come
>with X11R5 and Motif, the integrated IXI desktop probably causes some
>incompatabilities in the binaries.

I find all the X stuff works great, but, then, I threw away all the
X.Desktop stuff :)

-- 
  /--------------------------------------------------------------------------\
  | Mark A. Davis     | Lake Taylor Hospital | Norfolk,VA (804)-461-5001x431 |
  | Director/SysAdmin | Information Systems  |    mark@taylor.infi.net       |
  \--------------------------------------------------------------------------/

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

From: Rich Larsen <Rich.Larsen@DaytonOH.NCR.COM>
Subject: Video Cards Supported by XFree
Reply-To: Rich.Larsen@DaytonOH.NCR.COM
Date: Mon, 15 Aug 1994 12:35:16 GMT

I am looking to purchase a new VLB video card and one of the factors
in the decision is the ability to use it with Xfree.

The two I am looking at are:

1. ATI Graphic Expression which uses ATI's Mach 64 chip. I understand
the Mach 32 is supported but not sure about the 64.

2. Alaris Sky Eagle which uses the NCR 77C32BLT chip.

If anyone has insights on these, I would really appreciate it.






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

From: jones@bock.endicott.ibm.com (David Jones)
Subject: DCE on Linux
Date: 15 Aug 1994 13:02:54 GMT
Reply-To: dgj@vnet.ibm.com

Anyone know of an implementaion of DCE for Linux ? Thanks.

--
David Jones
Loral FS Owego

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

From: tomk@access.digex.net (Tom Krotchko)
Crossposted-To: comp.unix.unixware
Subject: Re: WABI vs. SoftWindows?
Date: 15 Aug 1994 13:31:25 GMT

In article <1994Aug15.104039.1877@riscot.demon.co.uk>, neil@riscot.demon.co.uk (Neil Dunbar) says:

>>Whether an OS costs $0.00 or $1,000 doesn't seem all that important.
>
>I'm not so sure, Tom. That's a question of paying $21,000 or
>$22,000. If a customer is presented with two choices which are
>equivalent OS's, which one would he choose?

Honestly?  The one he's most comfortable with.  People buy RS/6000's
with AIX because they feel warm and fuzzy with it in many cases when
a far cheaper box will do.

Most companies are so conservative (and rightly so) that any type
of switch is questioned.  $1000 seems like a lot of money to you
or me, but if a company has more downtime, or it costs them more in
consultant fees to keep running, that $1000 savings looks like
a bad deal.

Mind you, its not that Linux is worse than Unixware, but its the
perception of the customer.  To give you an idea of the length
these businesses will go through:

I was doing some purchase orders for a customer 2 weeks ago.  I
needed something with a fair amount of horsepower, so I chose a
Dell Computer with P90, a fairly complete system with lots of
everything.  The customer was worried that I'd chose a Dell
computer because they'd always used NEC in the past, and "won't
this really be a support problem" (you know, DELL, one of the
biggest PC manufacturers in the world), and they really need to
talk this over.  Now imagine telling them you're going to be
using a freeware OS that isn't actually supported by anyone.

>Of course, using Linux instead of SCO, Unixware et al, removes the
>uncertainty of what happens if SCO, Novell and Sun were all to go
>belly up .... ;-)

<G>.


Tom Krotchko
<tomk@access.digex.net>

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

From: tomk@access.digex.net (Tom Krotchko)
Crossposted-To: comp.unix.unixware
Subject: Re: WABI vs. SoftWindows?
Date: 15 Aug 1994 13:32:38 GMT

In article <CuKrwC.147@news.cern.ch>, danpop@cernapo.cern.ch (Dan Pop) says:

>It's certainly not important for _that_ installation, but if you want
>to install the same OS on your home machine, the difference between
>$0 and $1000 is not exactly insignificant. Not to mention that the one
>which costs $0 comes with complete sources. For some users this is
>irrelevant, for others it makes a lot of difference. 
>
>Choose the one which matches best your needs and enjoy it!

I think you're absolutely right.  At the same customer, Linux might
be appropriate if I set up a WWW server for them (I'll let you know
the address if I can convince them to give me the funding <G>).

Tom Krotchko
<tomk@access.digex.net>

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


** 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-Misc-Request@NEWS-DIGESTS.MIT.EDU

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

    Internet: Linux-Misc@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-Misc Digest
******************************
