Subject: Linux-Misc Digest #578
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:     Fri, 21 Jan 94 04:13:17 EST

Linux-Misc Digest #578, Volume #1                Fri, 21 Jan 94 04:13:17 EST

Contents:
  Re: uucp 1.04 - looking for tester, _complete_ (better)version (Andreas Klemm)
  Re: WUARCHIVE LOST :-(  (David Way)
  Re: [Q] Pentium or 486DX2.... (Mark A. Bentley)
  Re: RFV: Linux International proposal (Tony Porczyk)
  Can you help with this SLIP/DIP problem? (Dennis Director)
  Re: uucp 1.04 - looking for tester, _complete_ (better)version (Vince Skahan)
  Re: infomagic CD Does it exis (Peter Z. Simpson)

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

From: andreas@knobel.knirsch.de (Andreas Klemm)
Subject: Re: uucp 1.04 - looking for tester, _complete_ (better)version
Date: 20 Jan 1994 20:56:42 +0100

vince@victrola.wa.com (Vince Skahan) writes:

>[...sigh...yet another one...let me preface by saying:
>       - I like Taylor UUCP
>       - I like Ian Taylor
>       - this is not a flame

Ok.

>       - I'm all for helping newbies

To be precise: my intention was to help everybody.

>       - I'm sure your release will be fine work...]

Hope so ...

>andreas@knobel.knirsch.de (Andreas Klemm) writes:
>>I'm looking for test sites for my taylor uucp package.
>>Compared with other uucp packages it's _complete_, has good default 
>>configuration and comes with the taylor uucp contributed software ...

>I just don't see what the value of all this work is.

;-) But I see.

>>o  I have made efforts to create *well documented* configuration
>>   files, to make uucp configuration easier for ``newbees'' 
>>   and ``oldbees'', too ;-)

>The example HDB ones I provided for SLS and that are in the UUCP HOWTO
>work just fine as is.  Drop in your system name, phone number, username,
>and password and they should work for 99% of the systems out there.

No !  ;-)

In all Slackware releases I saw, the example in the Systems file
for a site entry is wrong. Is that example from your package ?

This is one of the reasons I deceided to offer Patrick a new uucp package. 
Such a little bug can cost lost of hours configuring uucp. 
I don't have the original package online in the moment.

It was the string that was in front of the phone number in 
/usr/lib/uucp/Systems. This string normally matches with an entry in
the Dialcodes file. In the example it had the name of an Dialer entry 
"scout" ;-) And this doesn't work :-(

>I've heard from Ed Carp that the next LGX is supposed to be perfectly
>configured and install-time-configurable.  I've offered my newspak stuff to
>Patrick for Slackware and he (and anybody else) is of course able and
>encouraged to grab it from sunsite to have plug-n-play config files that
>they can put into place and type 'make' for their particular installation.

And I had the goal to offer Patrick a working uucp/inn solution.
Several weeks ago he was interested in getting inn from me.
But since he was busy last weeks and out of town, we didn't get it
managed and coordinated until now.

If I had known, that there is somebody who ``redesigns'' t-uucp for
Linux then I wouldn't had any reason to start an own version of an 
uucp binary package ...

But now I started the thing and want to come to a result.
There are only two possibilitiews:
        - people like my package
        - or not.

If not, then I'll throw it away, really. But if I read the replies that I got, 
then there are many people out there who want another uucp package. 
Maybe they are not satisfied with the current one ??? 

>HDB is fully documented in about every elementary unix communications
>book you'll ever find.  Please find me one that documents Taylor mode.

Ok I agree with you. HDB is a kind of standard. But ....

Taylor is fully documented, too, and the documantation costs _nothing_ !
What I missed in Slackware until now is the original taylor documentation !
This is another point I wanted to change !

With nicely explained config files and good examples I think there is no 
difference if you use taylor or HDB config style.
Additionally I think taylor is easier to understand, because its
configuration syntax is much more verbose !

>So what work is really needed ?

Give me some time and I will tell you the experiences that the about 20 testers
made with the package. If they say it's not better or even more worse, then I 
will try to help other people debug their work and thro my package away.


=========================================================================
BTW: OK IT'S ENOUGH NOW. I HAVE Tester ENOUGH  ;-) ;-) ;-)
=========================================================================

>Want to come up with a quickie install script that asks the questions
>that need to be asked and stream edits the Taylor config files into
>the user's choice of V2, HDB, or Taylor config files ?  Now that would
>add significant value.  Reinventing the wheel and typing 'make' to
>sources that require no patching adds no value that I can see.

Sorry. You're wrong. I didn't make modifications/patches to the sources.
But I made the experience, that defining

#define HAVE_UNBLOCKED_WRITES 1
and
#define SINGLE_WRITE 128

in policy.h gives a better performance than I got with the original package in
Slackware... (using the same proto and package sizes).
I'm interested in hearing, if other people make the same experiences....

>>o  My package offers faster data transfer rates ( >50 Bytes/sec.)
>>   in comparison to the original package (using G proto).
>>   I would be interested in hearing if YOU made the same
>>   experiences...

>Huh ?  Have you hacked the vanilla sources to the released Taylor 1.04
>or are you just turning the 'i' protocol on as a (poorly chosen) default ?

No. See above. I only make some reasonable defines in policy.h.

>If you've hacked the sources, I sure wouldn't use it.  The standard 
>released sources work just fine and I'd rather be able to drop Ian a
>line to understand what's going on in the remote possibility that there
>is a major problem.

No. See above.

>If you made something other than 'g' the default protocol, you've hurt
>portability and the ability to connect to non-Linux boxes that have
>something other than Taylor as their UUCP.

No. See above.

>If you just ignore the issue and let Taylor negotiate the protocol, it'll
>do the right thing.  Ian did a fine job with what's there now in the 1.04
                      ^^^^^^^^^^^^^^^^^^
>sources...

Yes of course. He did a fine job. And I only want to do a fine job, too ;-)

>>o  I used Taylor Configuration instead of BNU/HDB. I think this offers 
>>   more advantages since this way it's possible to make use of the 
>>   contributed software, that comes with taylor uucp 1.04:

>you're kidding right ?
>you're inflicting the extremely non-standard Taylor-mode config files
>on newbies to get them a few utilities that already exist for HDB mode ?

Wheres the difference if newbees learn taylor or HDB ?????!!!
In the beginning they don't know much about the different styles and
so on. And they surely didn't read a book about HDB uucp ;-)

So I think it's allowed to introduce them to Taylor Config mode, which is 
more verbose. And I made efforts to create well dokumented config files.

And if anything fails, they can READ the uucp.texi file, because I 
have included it to my package.

>I don't understand.

>You're delivering a non-standard setup to new uucp users and expecting
>them to read texinfo stuff online rather than having them drop a dime on
>the Linux networking guide (for free) or the O'Reilly books (small cost) ?

Take a look at that files. And then decide !

>And this is to permit use of a few utilities that have dozens of existing
>variants for HDB config files already ?

uutraf ? uurate ? Where are these and other utilities in slackware ?

>I guess I'm missing something...

>When I took Ed Carp's mailpak and stole liberally from it :-) [...hi Ed...]
>to integrate his uucp/mail with the news stuff I assembled and it made it
>into SLS1.01, I selected HDB mode for many, many reasons.

Everybody has his reasons for doing something.... Me too. I told about
the reasons in the announcement.

>I've taken the liberty to include the following mail message that was my
>reply to Peter regarding why I gave him uucp in HDB mode for SLS 1.01.  The
>questions in the attached was from a SLS user (not Peter) asking why not
>Taylor.  The reply is from me to both he and Peter.

>#============== begin excerpt ======================

>From vince Thu Jan 28 16:56:40 1993
>Subject: Re: why have uucp in BNU mode ?

>> My question is now, why do you compile the Taylor-UUCP with the BNU-CONFIG 
>> files. Wouldn't it be better to use the Taylor Config.
>> The reasons are:
>>      1 The documentation is only for the Taylor Config files.
>>              (uucp-info*)
>>      2 You can, if you want, define BNU Config files in the Taylor-CONFIG.
>>        So if someone would like it, he can use it.
>>      3 Like me, it wouldn't be so hard to understand what is happening.
>>        I have the O'reilly book (Managing UUCP and the Usenet). But it's
>>        not so easy to understand.
>>      4 In BNU you can only define the packet-size at compile time.
>>        WIth Taylor you can define it in the Config-Files.
>> 

>I think the uucp should be configured in Honey DanBer mode just like it is
>currently for that following reasons:

>       - Taylor UUCP is totally nonstandard, yet there are tens of 
>               thousands of HDB systems out there. It is *THE* standard.

But sometimes it is nicer to take a nonstandard package. AT&T is a standard, 
too. nroff is standard, too.  But I like Linux and groff more (ok, ok, bad 
example.). ;-) ;-)

>       - accordingly, there are also thousands of experienced admins
>               who can help debug it

I think the configuration of taylor is more verbose and easier
to handle. If you want, you can theoretically make all definitions
in one file. Again look into the files and then deceide.
I'd bet, that if you show a person who is new to that stuff
HDB and taylor files, that the taylor config files are better
readable.


>       - it's fully documented in quite a number of books including
>               (but not only) the O-Reilly and Associates series and
>               the how-to-administer manuals of every *nix I've seen
>               at work, which includes all the major vendors.

Ok, thats right. But the most people don't buy a book about uucp.
So, where is the advantage.

And: taylor uucp comes with doku and is very well documented.
No need to buy a book. So ..... ?!?!?

>       - no vendor I'm aware of provides Taylor UUCP anywhere. 

Is it necessary for using a free uucp on a free unix to have a vendor 
providing it ????

>       - C-news/smail/elm play together perfectly with the HDB mode,
>               yet they do not expect the strange Taylor format.

I think you are wrong. Cnews, smail and elm don't use uucp's
config files. And here they are playing together with taylor config
mode as well. BTW: works with inn-1.4 perfectly, too.

>       - defining non-default packet sizes for uucico is a very dangerous
>               thing that can result in your system crashing the remote
>               system, depending on which software each side runs.

But if the other side uses taylor uucp, too, then it can be an advantage.
In Ian Taylors Doku this point was clearly explained. I did it also
in my config files. No danger !

>I see absolutely zero value in switching from BNU (HDB) format to anything
>else and I also see considerable risk in doing so.  As far as the SLS

Let users come to own conclusions, ok ? 

>news/mail stuff I've provided, I do not plan on configuring them to support
>anything other than BNU...but of course the sources are available on

No, no, no !  What part of cnews or smail needs the HDB/BNU config files and 
for what purpose ? I think it's wrong what you say or I misunderstood you
completely ....

BTW: I get the certain feeling, that you are looking vehemently for arguments
to argue against my package ... arguments that I can not share with you .... 

And to say it again very clearly:

- I didn't hack the original source.
- But I get a better performance with my package.
- I like taylor configuration more. It's more clear.
- I replaced your uucp package with mine. I'm very happy with it now.
  && I had NO problems with pine, elm, smail, cnews and inn-1.4.

And I have nothing personally against you and your uucp package ;-)

Andreas ///
-- 
Andreas Klemm                 /\/\____ Wiechers & Partner Datentechnik GmbH 
andreas@knobel.knirsch.de ___/\/\/     andreas@wupmon.wup.de (Unix Support)

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

Crossposted-To: comp.sys.amiga.misc,demon.local,connect.chatter,connect.audit
From: david@ncc1701a.demon.co.uk (David Way)
Subject: Re: WUARCHIVE LOST :-( 
Reply-To: david@ncc1701a.demon.co.uk
Date: Fri, 21 Jan 1994 00:07:50 +0000

In article <2hn28o$9v5@hp-col.col.hp.com> terry@col.hp.com writes:

>    Sorry to hear about their loss. I have contributed several
>    pictures there myself.
>    Some of the stuff was:  WW1 stuff, WW2 aircraft, NoseArt,

What's NoseArt?  Stuff painted with yer nose?

-- 
David Way
==========================
david@ncc1701a.demon.co.uk

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

From: bentlema@nxsci173c.mrs.umn.edu (Mark A. Bentley)
Subject: Re: [Q] Pentium or 486DX2....
Reply-To: bentlema@cda.mrs.umn.edu
Date: Fri, 21 Jan 1994 00:38:16 GMT

In article <2hjci6$jkc@silver.jba.com> chris@jbasp.demon.co.uk (Chris Bradshaw)  
writes:
> Greetings Netters....
> I am on the (steep) verge of buying my first PC, and am trying to buy the
> best system for use with Linux and Xfree. I was wondering if anyone could
> give me some advice or opinions on the two machines I currently have in mind.
> 
> The first machine I am considering is:
> 
>         Gateway 2000 P5 Pentium (256k cache) system with PCI local bus, 
>         15" monitor, 
>         424 Mb IDE drive, 
>         16Mb RAM, 
>         PCI LB ATI Ultra Pro 2Mb accelerated graphics adapter.
>         Price=~#2,300 ex VAT
> 

Good system.  Wait for prices to drop more if you really want a Pentium.  (They  
will drop very soon!)
  
> 
> The other machine I am considering is:
> 
>         Vale 486DX2 66 Mhz (256k cache) with VESA Local Bus, 
>         15" monitor, 
>         450 Mb IDE drive with a 2 Mb VLB caching controller,
>         16Mb RAM, 
>         VLB Prolink Cirrus Logic 5428 1 Mb acclerated graphics adapter.
>         Price=~#2,000 ex VAT
> 
> 

> As far as I am aware, all of the above hardware is supported by Linux 
> and Xfree. The two points I am most concerned about are the disk and
> the display adapter. Are there PCI based local bus disk caching controllers
> on the horizon and if so, will they be supported by Linux? If the answer is
> no to either of these questions, then if I go for the 486, will the 2Mb VLB
> caching controller make a great difference when using Linux & Xfree?
> As regards the display adapters, is either of the above cards considered
> better/faster/more suitable for use with Linux or Xfree? Can Xfree support 
> the maximum resolutions of either of these cards? At 1024x768x256, is either
> one of these cards significantly better or faster under Xfree? 
> 

I wouldn't go with anything too exotic unless you are absolutely sure it will  
work with XFree as far as video controllers go.  The ATI Graphics Ultra Pro or  
the newer XLR (Local bus) boards are really nice and work great with linux.   
Also any of the S3-805's I believe will work too.  Save your money!  The most  
IMPORTANT thing is memory!  I've you want a really nice system get at least  
16MB or RAM.  I'm running 8MB with XFree now and get by fine running 2 or 3  
large applications, but when compiling (even with everything swapped out) I  
still thrash on the drive alot.  16MB!  16MB!  16MB!

> Finally, given that Pentium & PCI are relatively new technologies, in the 
> long term it is reasonable to expect that 486's and VLB might/will be 
> superceded by these new systems. Would it be worthwhile to get the Pentium,
> and add things like a hard disk caching controller later?

PCI is here to stay, but that doesn't mean VESA local bus will fade away.  
        (Heh, that rhymes!)
 
> 
> If I sound confused....its because I am. Apart from the above questions, 
> I would really appreciate anyones opinions or thoughts on the pro's and con's
> of going for either one of the above machines. If requested, or it seems 
> worthwhile, I will post a summary of my findings. Any and all replies are
> as always greatly appreciated. Thanx a million in advance.   
> 

I wouldn't buy a Pentium at all.   The technology is great, but the chip is hot  
because it has so many transisters crammed into a little space and using 5V  
technology instead of the 3V technology.  It is very likely that Intel will  
redesign the Pentium with 3V technology and built-in power saving features like  
they did for the 486, but there has been no confermation on this.

Also, Intel has a new 486 chip coming out:  the "486DX4".  It's a  
clock-trippled chip with has been reported by beta testers to out perform the  
Pentium in many areas.  I think they call it a DX4 since IBM already calles  
there's a DX3.  I'm not sure about this though...

According to Computerworld the I486DX4 will be coming out on the market this  
March, and should sell in systems for about $2000!!

--
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Mark Bentley                     bentlema@caa.mrs.umn.edu     (VAX/VMS)
University of Minnesota, Morris          @cda.mrs.umn.edu (UNIX/Ultrix)
UXEM Contact Person                      @nxsci173a.mrs.umn.edu  (NeXT)
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=


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

From: messina@netcom.com (Tony Porczyk)
Subject: Re: RFV: Linux International proposal
Date: Fri, 21 Jan 1994 05:12:43 GMT

pdcruze@ee.uwa.edu.au (Patrick D'Cruze) writes:

> [...]

This is an interesting idea.  However, what relevance has voting to
the creation of Linux International (LI)?  Since the power of creation
is surrendered to the voters, what role will they play in LI?  Is LI
supposed to be a USENET-style democratic anarchy, or is it supposed
to be a non-profit organization, with offices, etc.?  If the latter,
then voting makes little sense if any.

Anyone else has the same doubts?

t.


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

From: dennis@dehn.math.nwu.edu (Dennis Director)
Subject: Can you help with this SLIP/DIP problem?
Date: 20 Jan 1994 21:34:31 GMT

I need dome help.  Trying my first SLIP connection, I got
to the following "Please set HOSTNAME first." message.

I think I have set /etc/HOSTNAME and all other related files
as suggested in the HOWTO docs.  Any suggestions?  dennis@math.nwu.edu

### Begin logfile
elvex>slip
Entering SLIP mode.
Your IP address is 129.105.9.52.  MTU is 1500 bytes
Header compression will match your system.

[ Back to LOCAL mode. ]
DIP> local 129.105.9.52
DIP> mtu 1500
DIP> mode SLIP
Please set HOSTNAME first.
DIP> 
### End logile

Thanks for any help.


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

From: vince@victrola.wa.com (Vince Skahan)
Subject: Re: uucp 1.04 - looking for tester, _complete_ (better)version
Date: 20 Jan 1994 19:35:55 -0800

jbettis@cse.unl.edu (Jeremy Bettis) writes:
>Taylor UUCP can be compiled to use either HDB or Taylor at runtime.  This is
>what I have mine set up as.  First I started with the same old HDB config
>files I always had, and then as I wanted to use the new features or Taylor
>(Like canceling dialing if the line is BUSY instead of waiting 30 secs for
>timeout) I converted the files one by one to Taylor after I understood them..
>It works with some files as Taylor and some as HDB.  Why didn't you compile
>it with both enabled???

There is a bug/feature in the code where if you compile with anything more
than HDB enabled, HDB-mode anonymous uucp will not work.  Ian confirmed it
in some mail to me when I provided all-three-types-built-in to Peter for SLS
and I couldn't get anon uucp to work here in HDB mode.

Of course if you don't want to do anon-uucp in HDB mode, you can surely
compile with all three types of configs built-in (but you have to pick one
way to do spooling and logging I think) and run any type of configs you
want.

Other than that, it's a quickie to edit the policy.h file and reset things
to Taylor mode and type make if you are so inclined...

>I have had nothing but problems with the package you put together for
>SLS1.01, no docs, no sample config files nothing.  With the source I could
>just read the docs and make a Taylor config file, you expect me to go out and
>buy a book on UUCP just so I can set up my sys files.. I don't think so.

I provided what Ian provided.  Short of violating copyrights, I couldn't
do any better.  In particular, I would strongly recommend that any uucp admin
drop the dime for the O'Reilly book and/or print the Linux Networking Guide
from the doc project (free to get, not necessarily free to print).

-- 
     ---------- Vince Skahan --------- vince@victrola.wa.com -------------
     "When you can get your customer to tatoo your name on their chest,
      it is unlikely that they will change brands"
           - Indiana Univ. of PA professor about Harley Davidson owners

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

From: peter_simpson@3com.com (Peter Z. Simpson)
Subject: Re: infomagic CD Does it exis
Date: 20 Jan 94 20:16:39 GMT
Reply-To: peter_simpson@3com.com

In article 138D10D8@billteds.com, CALE_HOLLINGSWORTH@billteds.com (cale hollingsworth) writes:
> ->
> -> In early November I ordered a CD which contains a snapshot of the lin
> -> archives on tsx11 and sunsite. The distributor said that it was relea
> -> every two months and that the next one was due in a few days...
> ->
> -> I later heard that the snapshot was taken on December 16th. Has this
> -> happened?
> -> If so where can I get the CD (UK supplier preffered)?

For ordering info, see below.  I phoned them, gave them my credit card number and
received the disk in a few days.  I'm happy with what I received.

Peter

[Begin included text]

From Joel@infomagic.com (Joel Goldberger)
Newsgroups: comp.os.linux.announce
Subject: Update to the InfoMagic Linux CD-ROM
Date: 19 Dec 93 23:56:09 GMT

InfoMagic is pleased to announce the first update to our Linux snapshot CD.
This disc is a complete snapshot of the linux archives from tsx-11.mit.edu and
sunsite.unc.edu.  The disc contains a number of "distributions" including SLS,
TAMU, MCC, *and* slackware.  The snapshot was taken on 8 December 1993.

This disc is primarily intended as a developer resource unlike the other CD
products which include extensive installation support and scripting.

The disc is priced at $15/copy.  Shipping within the US is $5, international
airmail is $10.  This disc is updated every two months.  Subscriptions are
available beginning with this release for $75/year + shipping.  We accept
Mastercard and Visa.

For more information, or to order, contact InfoMagic at:
email:  info@infomagic.com
Tel:    800-800-6613 or 609-683-5501
Fax:    609-683-5502

[end included text]
---
=================================================================
 Peter Simpson, KA1AXY                    Peter_Simpson@3com.com
 3Com Corporation                                 (508) 836 1719
 71 Lyman Street                            Northboro, MA  01532
 #include <std_disclaimer.h>        Linux = *free!* Unix for 386         
=================================================================



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


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