Subject: Linux-Misc Digest #815
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:     Tue, 15 Mar 94 11:13:16 EST

Linux-Misc Digest #815, Volume #1                Tue, 15 Mar 94 11:13:16 EST

Contents:
  Re: DOOM for X (Sakari Jalovaara)
  Re: outbytes.c helps backup Linux (Alex Ramos)
  Re: Linux and Hayes 28,800's - anyone have them working? (Tom Webster)
  Re: Linux and Hayes 28,800's - anyone have them working? (Frank Lofaro)
  Re: Don't want to make floppys! (Andreas Helke)
  Re: *Please* comment on Gateway P4D-66  (486/PCI) (David Marples)
  PCI IDE is *not* just IDE on another bus... (David Marples)
  Re: Network support changes? (Alan Cox)
  Re: Does zless exist? (Greg Patten)
  Re: BSD vs. Linux (Sakari Jalovaara)
  Re: compiler quality (was "Reverse-engineering") (Joachim Schrod)

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

From: sja@snakemail.hut.fi (Sakari Jalovaara)
Crossposted-To: comp.os.386bsd.apps
Subject: Re: DOOM for X
Date: 15 Mar 94 10:26:13 GMT

>]To get high speed graphics on X you just need to re-write the server:
>]  [have the usual Xlib calls access the hardware directly]

> There are (apparently) several misconceptions here.
>
> (1)   You can't necessarily map memory from the display adapter
>       directly.

Indeed.  If the hardware doesn't support high-speed graphics it doesn't.
In that case you can't do high-speed graphics.

If the hardware _does_ support high-speed graphics but the operating
system doesn't give sufficient access to hardware, you need to have
the OS fixed.

Since these are 386BSD and Linux newsgroups I'd expect the OS not to be
a problem.

> (2)   Not all OS's support the large segment changes, where the

If the display hardware supports memory-mapping but the OS doesn't,
you need to have the OS fixed.  Or get another OS.  Or make do with
somewhat slower graphics.

> (3)   The client/server distinctions are fictional; I have a lab
>       full of NCR X terminals (now an AT&T X terminals I guess)
>       that support developement and download of clients to run in
>       the terminal (server) address space.

EXACTLY!  This is what I was talking about: compile (or link) the
X clients in such a way that X library calls are hardware-aware.
You call it "download the client to the terminal", I call it
"link with a library that accesses hardware when DISPLAY=:0.0".

All of this is academic, though, in that modifying the MIT X server
to be a shared library is a decidedly non-trivial task...
                                                                        ++sja

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

From: ramos@engr.latech.edu (Alex Ramos)
Subject: Re: outbytes.c helps backup Linux
Date: 15 Mar 1994 06:20:06 GMT

On article to comp.os.linux.misc, 29399@wubios.wustl.edu wrote: 
> Here is a program called outbytes.c.  It will allow you to backup your
> Linux systems to diskette.  It requires flop and unflop, also included
> below.  The command to use this is:

>     cd /
>     tar cvf - . | outbytes 1457152 /a/backup.tar

How is this any better than "tar -Mcvf /def/fd0 /" ?

(Your program does have a use though: tar won't support -M with -z !!)

--
Alex Ramos <ramos@engr.latech.edu> * This message is copyrighted material!
Louisiana Tech University BSEE/Sr  * All rights reserved. No warranty, etc

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

From: webster@kaiwan.com (Tom Webster)
Subject: Re: Linux and Hayes 28,800's - anyone have them working?
Date: 14 Mar 1994 22:50:56 -0800

Brian Kramer (bjkramer@remus.rutgers.edu) wrote:
: Does anyone have linux working with hayes 28.8's?  I may try
: it for a slip connection if possible.

Please be careful here!  Do you want to connect to another Hayes
28.8?  All of the 28.8 modems being sold today are V.FC (meaning
V.fast class).  THe V.fast standard has not been ratified, and
there are several different implementations and semi-compatible
chipsets on the market.  Most of the vendors (all?) are offering
to retro-fit their V.FC modems to V.Fast (for a fee), but in the
meantime, it could be difficult to connect various vendor's 
modems together at top speeds.

Connections to V.FC modems can be problematic also, Microcom's
DeskPorte uses the parallel port and for top speeds the Hayes
Optima 288 wants to connect with the serial port at 230.4Kbps.
Is this a reliable speed without exotic H/W like the Hayes
EPS-II serial card?  And can the Linux drivers be tweeked up
to this speed?

If you are gung-ho to get a 28.8 modem, PC Mag seemed to like 
the Hayes.  It will run with a DTE of 115.2 (but slower than
its max.), has fax, and the V.fast upgrade is $100.  [Ref:
PCMag (Network Ed), 3/29/94, pg NE15-NE18.]

Could the Linux serial drivers be tweeked to handle 230.4Kbps?
I keep hearing a little voice saying: "I dunno think she 'kin 
take much more 'o this cap'in!...."  :->

Tom
--
+--------------------------------+------------------------------+
| Tom Webster                    | "Funny, I've never seen it   |
| webster@kaiwan.com (home)      | do THAT before...."          |
| webster@ssdgwy.mdc.com  (work) | - Any user support person    |
+--------------------------------+------------------------------+
| finger -l webster@kaiwan.com to get my PGP Public Key.        |
+---------------------------------------------------------------+

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

From: ftlofaro@unlv.edu (Frank Lofaro)
Subject: Re: Linux and Hayes 28,800's - anyone have them working?
Date: Mon, 14 Mar 94 20:30:15 GMT

In article <Mar11.220605.33112@acs.ucalgary.ca> jcsokolo@acs.ucalgary.ca (Jason Conrad Sokolosky) writes:
>Brian Kramer (bjkramer@remus.rutgers.edu) wrote:
>: Does anyone have linux working with hayes 28.8's?  I may try
>: it for a slip connection if possible.
>
>I didn't know they had 28.8's out yet.  Who sells them besides Hayes????

I heard Practical Periphrials has one for $300 (or less).


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

From: andreas@orion.mgen.uni-heidelberg.de (Andreas Helke)
Subject: Re: Don't want to make floppys!
Date: Mon, 14 Mar 94 21:22:29 GMT

: grante@hydro.rosemount.com (Grant Edwards) writes:

: >I used the Yggdrasil CD for the first couple machines upon which I
: >installed Linux. This time, the machine doesn't have a SCSI
: >controller.

: >So, I decided to install a base system + network from floppies, then
: >install everything else via NFS.  My first choice was debian -- but it
: >won't install packages from a fs that's already mounted. As far as I
: >can tell, you have to have floppy disks.  I've got all of the packages
: >I want already on the Linux machine's hard disk, but I can't install
: >them from there, and I refuse to make a bunch of floppies containing
: >stuff that's already on the hard disk!

In the case of debian this is a little bug in the install script. I do
not know wether the fixed version is already out or if it comes with
debian 0.92. I also heared that debian supports nfs installation. I
did not try this, I copied the packages to the dos partition and 
unmounted that one.

: >Next, I thought I would try Slackware.  According to the README files
: >I have to make a bunch of floppy disks with MS-DOS file systems from
: >which to install the system.  Yikes!  Why bring MS-DOS into the
: >picture?

I never installed a complete slackware system. But the individual
packages can be installed from the current directory. NFS installation
is probably also supported.

Andreas
-

Andreas Helke

Institut fuer molekulare Genetik, Universitaet Heidelberg
Im Neuenheimer Feld 230 
69122 Heidelberg

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

From: dmarples@voyager.eee.strath.ac.uk (David Marples)
Subject: Re: *Please* comment on Gateway P4D-66  (486/PCI)
Date: 15 Mar 94 11:17:50


In article <GISLI.94Mar14114046@timoshenko.eecs.umich.edu> gisli@timoshenko.eecs.umich.edu (Gisli Ottarsson) writes:

   I have just placed and order for a Gateway P2D-66 and
   would sleep a little better if I could resolve a few
   iffy areas.  I understand that Gateway is a moving
   target so what works this week may not work next week.

   Here are some questions:

     o  Are there any problems with the PCI mother-
           board used in this machine?  Does it work
           with Linux?  Would you buy (another) one?

Mine's working fine.


     o  The 2Mb MACH32 video card that comes with the
           PCIs from Gateway uses DRAM.  This had been
           solved for the DRAM VLB-version called ATI CLX 
           a while ago and is supposedly OK in Xfree86 2.1.
           Can I expect to use this card for 1280x1024NI
           @ > 60Hz which is as high as the 17" CrystalScan
           will go?  Does it work at all with Xfree86 2.0?

I've got an Xconfig for 17" Crystalscan and ATI GUP DRAM.  It
works fine at up to 1152x910x256.  It will go higher (I think) when
Xfree 2.1 comes out and allows faster dot clocks.  If you need it
e-mail me.

     o  I've saw one report of problems with the 540Mb
           IDE hard disk available in this machine.  It 
           had something to do with Linux not seeing it.
           Can someone refute or substantiate this?  Is
           the 424Mb a better choice?

Don't worry about this either.  The problem is that G/W set up the
BIOS to auto-detect the size of the drive.  Simply hard setting the
parameters in the BIOS rather than letting it get them itself makes
this problem go away.  fdisk (and a few other bits) complain about a
drive with >1023 cylinders, but they still work perfectly.  

(When is someone going to take out those warnings, BTW?  Are they
relevent under *any* conditions any more?).

     o  Has someone ordered this machine with the Colorado
           Jumbo tape system?  Does it work with Linux and
           if so, at what transfer rate?

You need to use ftape which doesn't come bundled with 1.1.2 Slackware.
Having said that, I can't get it working, but it is supposedly
explicitly supported by this driver.  If anyone has got it working,
would they please talk to me so I can figure out my problems....

     o  My system will have a CD-ROM.  Some kind soul 
           posted info about which CD-ROMs work (according
           to this source Gateway uses three kinds).  I
           have lost this posting.  Any comments here?
           Should I order the Gateway sound card?  Is it
           any good under Linux?

The CD-ROM I received with mine was the Mitsumi FX001D.  Works fine.
Raw data transfer rate is 385K/Sec.

   Thanks for any comments.  Needless to say I will post a
   summary if any of these things become sufficiently clear.

Machines are good, G/W tech. support is unavailable ->  Its getting
worse in the UK too!.

                                           Gisli


hope this helps - email for more
DAVE

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

From: dmarples@voyager.eee.strath.ac.uk (David Marples)
Subject: PCI IDE is *not* just IDE on another bus...
Date: 15 Mar 94 11:26:14


.. leastways, thats my understanding.  Although it will behave
perfectly well if its just used that way it also supports fast DMA
transfer modes.  Under DOS this operating mode is accessed using a
special device driver.  If you don't load it, you just get normal IDE
performance, well, across a faster bus, admittedly.

To get the real boost (and, under Linux, much lower processor
utilisation) you need to turn on the DMA mode.  This should then beat
the performace of SCSI drives (especially if you only have one drive
in your system) in terms of busy time etc.

It seems to me that theres quite some performance there to be had if
the appropriate drivers are built.  This post is intended to open up
this subject for discussion so *please* don't blatently flame me
*unless* I'm so far off the ball as to be on a different pitch.  If
anyone has the 'specs for this transfer mode I'd appreciate a pointer
so that I can look into the subject in a bit more detail.

DAVE

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

From: iiitac@uk.ac.swan.pyr (Alan Cox)
Subject: Re: Network support changes?
Date: Tue, 15 Mar 1994 11:33:12 GMT

In article <1994Mar14.012446.6377@compiler.tdcnet.nl> jvoosten@compiler.tdcnet.nl (J.S. van Oosten) writes:
>Hmm. Networking is unfortunately still the most unreliable, vague, and
>errorphrone area in Linux... Despite a lot of HOWTO's and manuals it can
>take hours to set things up properly... (how about a comp.os.linux.networks
>?)
In some ways not a bad idea. The problem is network configuration is complex
at times. Almost every Linux networking bug report I get now is a user setup
error. There are still a few lurkers in 1.0 woodwork but they are slowly coming
out. 

I do however consider the difficulty in setting up Unix networking to be
in many ways a bug in itself. IP doesn't help as it was designed to be setup
by someone who knows what they are doing, as opposed to the turn on and boot
mentality of netware.

Alan


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

From: greg@loose.apana.org.au (Greg Patten)
Subject: Re: Does zless exist?
Date: 15 Mar 1994 20:15:29 +1100

andreas@orion.mgen.uni-heidelberg.de (Andreas Helke) writes:

>Charles W. Blumreich (cwblumre@major.cs.mtu.edu) wrote:
>: Phillip Hardy (phillip@mserve.kiwi.gen.nz) wrote:
>: :->i have been playing with my system and i realy like compressed
>: :->Faq files now to read them all i have to do is.
>: :->zcat LINUX-FAQ.gz | more
>: :->and Vola i can read/search the faqs without takeing 
>: :->up late-amounts of hdd space :)

>: not to be critical, but what's wrong with zmore?

>This is one of the less commonly known programs. I did not realize that
>I have it until now. But I do not like more because of it's one way nature.
>Therefore I have to use zcat * | less. I could probably write a shellsript
>with the name zless which automates this process. 

under tcsh I have:

alias zless     "zcat \!* | less"

Cheers,
-- 
Greg Patten                                              Melbourne, Australia
greg@loose.apana.org.au       For information on APANA mail info@apana.org.au

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

From: sja@snakemail.hut.fi (Sakari Jalovaara)
Crossposted-To: comp.unix.bsd
Subject: Re: BSD vs. Linux
Date: 15 Mar 94 11:24:31 GMT

> > Unused hardware can be usually configured out of the kernel.
> > I seriously doubt that whatever unused hardware support remains
> > is going to make a big difference. 
>
> That was just an example. There are all sorts of areas where excess
> junk floats in.

Yes, there can be excess junk in software.  In this specific case:
where, exactly, is there excess junk in BSD?

There are a lot of very smart people working on NetBSD, FreeBSD,
BSD 4.4 and Linux. If you have specific knowledge of excess junk
in any of these systems, I'm sure you'll find at least someone
who'll want to hear about it!

So, instead of anecdotal stories about software having unnecessary
code: where exactly in NetBSD-current, FreeBSD 1.0 or BSD 4.4 is that
unnecessary code?  What operations do you know to be slow because of
unnecessary code?  What parts of the kernel or utilities are
"accumulated cruft" that can't already be configured out?

Really, tell us, I wouldn't mind putting
        #ifdef OLD_DHZ11_CODE   ...   #endif
in the BSD source if significant amounts of such code can be found.

> ----------   1 kernel   kernel     215556 Mar  4 23:03 /vmlinux
> Of course, my kernel is compressed. Maybe yours isn't.

Oh, as we were discussing kernel bloat due to unnecessary code,
I didn't think it might be even remotely relevant to look at
compressed kernel sizes.  Unless the CPU can execute such code
without unreasonable overhead.

The point remains: with Linux and BSD kernels being fairly close
in size, I can but wonder: where's the bloat?

> Code size is only part of it. Speed is another part. The memory
> requirements of the basic utilities, too. 

Yup, that matters.  As (I think) we were discussing code bloat
due to old unnecessary code in the system, where exactly is that
old code in either BSD or Linux utilities?  Really, tell me, I
wouldn't mind putting a few "#ifdef notdef"s in there.  The source
is available, so finding and fixing stuff like this should not be
a problem.

I've used NetBSD on a 486. A largish application runs 10-20%
slower than on twice-the-price workstations.  I've read some of
the kernel's and utilities' source and haven't yet come across
mountains of old unnecessary code.  Is it really there somewhere?

> > > Are you going to convince me that BSD
> > > doesn't suffer from the same sorts of problems?
> > Probably not.
> Oops. I meant to say "are you going to TRY TO convince me..." :-)

Not really even that.  Probably no net article I can write is going to
convince you beyond any doubt.

If you _want_ to be convinced, ftp the BSD code and read it.
All the source is available.  It'll eat up a lot of your time,
though.
                                                                        ++sja

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

From: schrod@iti.informatik.th-darmstadt.de (Joachim Schrod)
Crossposted-To: gnu.misc.discuss
Subject: Re: compiler quality (was "Reverse-engineering")
Date: 15 Mar 1994 12:43:38 GMT

In article <1994Mar10.033415.10872@rpp386>, jfh@rpp386 (John F. Haugh II) writes:
> In article <2linvj$7ue@fitz.TC.Cornell.EDU> elan@tasha.cheme.cornell.edu (Elan Feingold) writes:
> >In contrast, getting a patch from a commercial compiler vendor is like getting
> >teeth pulled, or something similar.  Even if they send you a patch, it takes
> >a long time, and what about all the other users out there who have the buggy
> >version?  They don't see that patch.
> 
> I don't know which companies you've been dealing with, but if you call
> IBM and the bug is known, you can get a set of update floppies in your
> hand in under 24 hours.

That is plain false.

To wit:

        IBM support
        
        CREATED=92/03/17 OPENED=92/03/27 CLOSED=92/07/06
        ERROR DESCRIPTION:
            the function prototype for fchmod() in the include file
            /usr/include/sys/stat.h is incorrect. the first parameter should
            be a file descriptor (int) not a string (char *).
        LOCAL FIX:
        PROBLEM SUMMARY:
            our current update scheme doesn't allow us to update header
            files.  therefore,  we will not ship this change.
        PROBLEM CONCLUSION:
            changes will not be shipped.
        TEMPORARY FIX:
            customer can edit the file themselves.  change the first
            parameter of the fchmod call to (int) instead of (char *).
        
A well known bug for IBM -- try to get a fix into the next AIX release...


On the topic that they work hard on fixing the error: Not so long
ago, IBM introduced an error in the regex library functions. (Around
3.2.2, if my memory is correct; but I can dig more information if
necessary.) That meant that awk, sed, expr and a lot of other
programs simply did *not* work any more correctly. They needed almost
half a year to fix that; they weren't able to just use the old
version -- which was working, after all -- as an intermediate
solution. And that's laughable.

Oh yes, we got an early workaround proposal: They told us ``why don't
you use the GNU tools?'' Nice, isn't it? So much about the IBM support...

        Joachim

--
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Joachim Schrod                  Email: schrod@iti.informatik.th-darmstadt.de
Computer Science Department
Technical University of Darmstadt, Germany

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


** 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
******************************
