Subject: Linux-Development Digest #744
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:     Sat, 21 May 94 15:13:05 EDT

Linux-Development Digest #744, Volume #1         Sat, 21 May 94 15:13:05 EDT

Contents:
  Re: Distributions considered harmful (Matt Welsh)
  Re: Two suggestions that might speed up Linux. (Steve Kann)
  Re: Anybody working on BSD dump porting? (Mark A. Horton)
  Re: Anybody working on BSD dump porting? (Gary Merinstein)
  Re: 32-bit Novell desktop OS combines Unix, (Rob Newberry)
  Re: 32-bit Novell desktop OS combines Unix, (Rob Newberry)
  Re: Anybody working on BSD dump porting? (Gary Merinstein)
  umsdos for 1.1.12? (Mark Swanson)
  Re: Gnu ASM docs wanted (Whistler)
  Need a little help with xmodmap (Christian Saucier)
  Linux for the masses? (WordProcessing again) (Byron A Jeff)
  Re: SIGHUP: We tried it! (Rocco Caputo)
  Re: 32-bit Novell desktop OS combines Unix, (Gareth Bult)

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

From: mdw@cs.cornell.edu (Matt Welsh)
Subject: Re: Distributions considered harmful
Date: Thu, 19 May 1994 06:07:25 GMT

In article <NELSON.94May17063549@crynwr.crynwr.com> nelson@crynwr.crynwr.com (Russell Nelson) writes:
>I don't think the current state of Linux distributions is good.

The Debian development project hopes to solve this problem.
We are working on a distribution of Linux that should resolve
all of the issues of incompatibility, version control, and
release stability shared by current distributions.

Note that this does not preclude the need for other distributions;
specifically, I hope that folks like Yggdrasil will continue to
make their distributions available with support. However, we hope
that Debian will emerge as the de facto standard. Things are already
moving forward by leaps and bounds.

Debian is currently in beta development, so I don't suggest it for
general use at this time. If you're a hacker and want to work on a 
new distribution, you should look into Debian, however.

If you have questions about the Debian development project,
read the various documents on sunsite.unc.edu in 
/pub/Linux/distributions/debian/info. I'll also field questions
about Debian, although the overall coordinator of the project
is Ian Murdock, aided by many others. But I won't solicit e-mail on 
their behalf.

>And Debian, while it has the support and name of the FSF, is just
>plain not needed.  Much, much better to fix the problems with
>Slackware, SLS, or Yggdrasil.

I disagree, Russ. If you would take a look at the work that
has been done on Debian I think you'd agree that this is the
best route to take. Debian takes a very different development
philosophy than Slackware and SLS---it is developed and maintained
in an open, distributed environment. This is an extremely vital
aspect of the project, with benefits that SLS and Slackware
simply can't partake in. 

Don't knock it until you've tried it, eh? If you're interested
in fixing the state of Linux distributions, please help us to 
develop Debian.

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

From: stevek@panix.com (Steve Kann)
Subject: Re: Two suggestions that might speed up Linux.
Date: 19 May 1994 03:26:19 -0400

Robert Sanders (gt8134b@prism.gatech.edu) wrote:
: stevek@panix.com (Steve Kann) writes:

: >William O Smith (wos@dcs.warwick.ac.uk) wrote:
: >: Suggestion 1:  Compressed buffers.  On fast machines, if there is idle
: >:       time, and spare buffer space, how about compressing the
: >:       disk buffers, storing both versions (compressed and un-
: >:       compressed) until memory gets tight, then just keeping
: >:       the compressed version (if it saves much space).  Presumably
: >:       if the machine becomes fast enough, decompressing memory
: >:       should be faster than loading off disk.

: >I think this is a really interesting idea...  And the basics of the

: And a really bad one.  It's kinda amusing how people will want
: to compress everything.  I imagine quite a few of them have tried
: to gzexe gzip...

: First off, if things aren't being used, there really isn't much
: use in keeping them in the buffer cache.  That's why old buffers get
: flushed.  To be used, buffers must get uncompressed.  Therefore,
: to stay compressed for any sensible length of time, a buffer would
: have to be unused.  But unused buffers have no place in-memory: 
: they stay on disk.

: Now, the reason why buffers can't just get uncompressed at read()
: or write() time: the kernel uses buffers for more than just disk
: I/O: any mmap()ed file shares buffer space.  So, if I mmap()
: /lib/libc.so.4.5.21 (which is what uselib() does), I have the
: disk buffers resident.  Even if noone read()s that file, they need to
: be resident and uncompressed so I can access libc from all the
: programs that mmap() it.

: People, if you don't have enough disk space, buy more.  Linux
: is great partly because it's fast; this sort of bad hackery will
: slow it to a backwards crawl.  Then the same people who wanted 
: compressed swap partitions will complain because their machine
: is slow when they swap.


: -- 
:  _g,  '96 --->>>>>>>>>>   gt8134b@prism.gatech.edu  <<<<<<<<<---  CompSci  ,g_
: W@@@W__        |-\      ^        | disclaimer:  <---> "Bow before ZOD!" __W@@@W
: W@@@@**~~~'  ro|-<ert s/_\ nders |   who am I???  ^  from Superman  '~~~**@@@@W
: `*MV' hi,ocie! |-/ad! /   \ss!!  | ooga ooga!!    |    II (cool)!         `VW*'

--
- Steve

stevek@panix.com
stevek@magnum.cooper.edu



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

From: mahmha@crl.com (Mark A. Horton)
Crossposted-To: comp.os.linux.help
Subject: Re: Anybody working on BSD dump porting?
Date: 21 May 1994 08:36:32 -0700

Gioacchino La_vecchia (gio@di.unipi.it) wrote:

: I'm trying to port bsd dump to linux.
: Many header files have not corrispondence on linux systems.

: Nobody working on it?


:                                       gio


Bleeechh!  I don't even use dump on my SPARCS anymore! (got bitten 
too many times!)  I now use regular, normal, cpio -- I like to know
what I'm really putting on tape!  It's no big deal to write a few
scripts to control incremental dumps, etc.  And now I backup the
SPARCS through the Linux box via NFS to an external EXB-8200, so it's
even easier.

Just my $.0250 (inflation) :)

-- Mark




--
Mark A. Horton      ka4ybr              mah@ka4ybr.atl.ga.us   mah@ka4ybr.com 
P.O.Box 747 Decatur GA US 30031-0747      ICBM: 33 45 N / 084 17 W
+1.404.371.0291                         Cruise: 33 45 31 N / 084 16 59 W
   "We may note that, for the purposes of these experiments, the symbol 
                "=" has the meaning "may be confused with."  

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

From: gmerin@panix.com (Gary Merinstein)
Crossposted-To: comp.os.linux.help
Subject: Re: Anybody working on BSD dump porting?
Date: 19 May 1994 19:23:23 -0400

In <Conrad_Nobili-180594213429@nobili.harvard.edu> Conrad_Nobili@Harvard.EDU (Conrad C. Nobili) writes:

>In article <2rdcfn$lu4@csugrad.cs.vt.edu>, mmead@csugrad.cs.vt.edu (matthew
>"pipeman" mead) wrote:

>>      What if you need to backup over the net?  There's rdump, but I assume
>> that won't work for the same reasons that dump won't ... how would dumping to a
>> disk mounted on a machine elsewhere on your local net be accomplished? :-)

>Is there some reason why you can't use tar or cpio in conjunction with rsh?
> Is there something I'm not getting here?  I haven't tried this with Linux
>yet, but I don't immediately see why it wouldn't "just work".

>I'm surprised, given your nickname, that you haven't already tried this
>yourself....  ;-)

rdump deals with raw data / raw disks; sort of a bit-by-bit image (you 
can boot & mount a dump tape as tho it were a disk) can you do these 
things with tar and/or cpio?


-- 
*** gmerin@panix.com  "..here pigs will fly, lightning will strike twice,  ***
***   mci:  489-6979       hell will freeze over, and eventually,          ***
***   ci$ 74035,1232         things will get really interesting..."        ***

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

From: rob@eats.com (Rob Newberry)
Subject: Re: 32-bit Novell desktop OS combines Unix,
Date: Sat, 21 May 1994 11:41:10 UNDEFINED

>>Sorry, again, but Novell doesn't support the open concept of Unix.  What 
they >>say is, if it's not ours, it isn't Unix -- that's not very open to me.

>Sorry and sorry again :) but the fact that Novell is involed in Unix AT ALL
>is a major positive point to me.  Even in the light that Novell is a business,
>and must make money, like the businesses we all work for....

Yes, yes, but you said they support the "open" concept.  As I'm well aware, 
Novell has the rights to the name "UNIX", but at FOSE, they would not admit 
that anything could be UNIX-like except what they made, or what they allowed 
to be made.   The feeling I got from those a**holes was, "We are Novell.  We 
are Unix."  Again, while that doesn't trouble linux or Unix or you or me, 
there is absolutely nothing "open" about it.  Whether or not you feel gooey 
for them because they are "involved in Unix AT ALL" is your business, but 
there's nothing open about Novell.

And I won't argue that we don't all try to make money, but IMHO, Novell makes 
money a lot like Microsoft does -- they use someone else's technology (USL 
then, linux now) and market it.  The only difference, at least in the USL 
stuff, is that they did pay someone for it.

Rob



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

From: rob@eats.com (Rob Newberry)
Subject: Re: 32-bit Novell desktop OS combines Unix,
Date: Sat, 21 May 1994 11:44:45 UNDEFINED

>In article <rob.150.02644292@eats.com>, rob@eats.com (Rob Newberry) says:
>+---------------
>| >Sorry, but Novell is much better, if in nothing else but that they support
>| >the open concept of Unix.  This says a whole lot to many of us...
>| 
>| Sorry, again, but Novell doesn't support the open concept of Unix.  What they
>| say is, if it's not ours, it isn't Unix -- that's not very open to me.  When
>+------------->8

>SOMEONE is still in the Dark Ages... The trademark "UNIX" is being/has been
>transferred to X/Open by Novell, and to be allowed to use it you need to pass
>a verification suite for "Spec 1170" compliance.  As yet, only Novell/USL-
>based systems can be "UNIX" because Spec 1170 is still under evaluation.  When
>released as a formal standard and when the verification suite is ready,
>non-Novell/USL systems will be testable for Spec 1170 compliance and will be
>brandable as UNIX.  ---And the grandfather clause will expire, meaning that if
>Novell UNIX doesn't pass the verification suite, *it* won't be UNIX any more!
>(In theory, at least...)

Well, to stick up for myself, I wasn't saying what I thought, but what Novell 
said to me at FOSE.  No, I wasn't really aware of the transfer to X/Open and 
all that (and BTW, thanks for the info in the rest of your post!), but then, 
neither we're the Novell jerks at FOSE.  Again, what they said to me was along 
the lines of "We are Novell.  We are Unix."  My post was simply to say that 
this attitude, given to me from Novell people, was not "open."

Rob



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

From: gmerin@panix.com (Gary Merinstein)
Crossposted-To: comp.os.linux.help
Subject: Re: Anybody working on BSD dump porting?
Date: 21 May 1994 12:16:53 -0400

In <2rjhfd$m7r@apollo.west.oic.com> dillon@apollo.west.oic.com (Matthew Dillon) writes:

>In article <2rgsdb$1c7@panix.com> gmerin@panix.com (Gary Merinstein) writes:
>:>I'm surprised, given your nickname, that you haven't already tried this
>:>yourself....  ;-)
>:
>:rdump deals with raw data / raw disks; sort of a bit-by-bit image (you 
>:can boot & mount a dump tape as tho it were a disk) can you do these 
>:things with tar and/or cpio?
>:
>:-- 
>    I did that once (not on a Linux system)... I never did it again.  I do not
>    believe it is even possible on a linux system.  No, I'll boot from a floppy
>    thank you!

>                                       -Matt
well, given that there isn't a dump/rdump/ufsdump etc. for linux yet, 
i'll agree that it probably isn't possible on a linux system. as for 
sunos (where even the idea of booting from a floppy, is silly) the tape 
works quite well - takes forever tho - but in a pinch, it works.

can you mount a tar or cpio tape as tho it was a disk? you can mount a
dump tape that way..you can dump an entire 5gb system to tape, and later
mount it inorder to search for anything that used to be on the original
disk; files, inodes, links, *anything*.

-gmm
-- 
*** gmerin@panix.com  "..here pigs will fly, lightning will strike twice,  ***
***   mci:  489-6979       hell will freeze over, and eventually,          ***
***   ci$ 74035,1232         things will get really interesting..."        ***

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

From: mswanson@hookup.net (Mark Swanson)
Subject: umsdos for 1.1.12?
Date: 21 May 1994 11:50:02 -0000

Has anyone patched the umsdos 0.2 patches to Linux 1.0 to work
with Linux 1.1.12?  I get the error on line 505 in the patch -
vm_task->min_flt stuff.  
Help!
(and thank you)

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

From: slouken@cs.ucdavis.edu (Whistler)
Subject: Re: Gnu ASM docs wanted
Date: Thu, 19 May 1994 23:22:09 GMT

Nicolaus Thirion (nthirion@rkw-risc.cs.up.ac.za) wrote:
: Hi,
: I am really in a dead end, anybody know of documentation 
: for Gnu assembly (GAS).

        I'm interested in this too.  Please email me any info
at slouken@cs.ucdavis.edu

Thanks!

        -Sam


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

Crossposted-To: comp.os.linux.help
From: saucc00@DMI.USherb.CA (Christian Saucier)
Subject: Need a little help with xmodmap
Date: Fri, 20 May 1994 23:01:33 GMT

After having unsuccessfully tried to find a canadian-french keyboard map
for Linux, I decided to make one myself.

I've been reading the xmodmap man pages and I've started modifying the
xmodmap.def file to suit my keyboard.

To assign a character to a key and a different one to the SHIFTED key
is pretty simple.  But how do I assign a character to the Controled or
Right_Alted key?

keycode 10 = 1 exclam

The first argument is for the unaltered key, the second for the shifted
key but how do I add a different one to the alt-key?

Looks like it has something to do with the modifiers but I'm a bit confused
with those up to now.

Can anybody give me a few pointers?

Thank you in advance,

Christian
 


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

From: byron@cc.gatech.edu (Byron A Jeff)
Crossposted-To: comp.os.linux.misc
Subject: Linux for the masses? (WordProcessing again)
Date: 21 May 1994 14:27:11 -0400

In article <1994May21.000155.9445@kf8nh.wariat.org>,
Brandon S. Allbery <bsa@kf8nh.wariat.org> wrote:
>In article <2rj4a6$9rq@nic.umass.edu>, cmay@titan.ucs.umass.edu (CHRISTOPHER M MAY) says:
>+---------------
>| Linux isn't for the feint of heart when it comes to computer knowledge.
>| It's really for those who want to push the envelope of what their PC
>| can do, under a multitasking Unix-like OS.
>+------------->8
>
>To make it short and unsweet:  if you want an operating system For Hackers
>Only, use *BSD.
>
>Does it *offend* you that someone might want to make a *ix usable without
>having to be a Unix God?  Why?  Why do you need to argue against *any* attempt
>to produce a non-hackernix, even if it can't/won't prevent you from using a
>hackernix if that's what you prefer?  (Not that I want to hear the answer; I'm
>not a psychoanalyst.)

This is an important point. Linux is quickly becoming bigger than all of us
and a lot of novice users who are open minded are starting to try it.
I at least offer it as an alternative to folks because the software is
free and Linux can wring many more uses out of your machine than DOS/Windows.

Where we're having problems breaking through is with the simple core software
base that many unix OS's don't have because they've never been in the
personal computer market with all the feature and the right price before.
(meaning that Unixware and Mark Williams didn't have the features, while
SCO and Intel SVR4 didn't have the price. Linux has both).

The last two installations I've done have been for novices. I'm batting
500 right now (one is still working with Linux, the other not). Let me
share some perceptions:

1) The lemming syndrome: novices need to be able to be part of a group
   that share similar software. Brand name is a monster in the PC software
   business. The best example of this is xspread: Unix product, runs on
   Linux, looks and acts like lotus 1 2 3. A novice linux user can go
   and get basic commands and whatnot from Lotus users and books. It gives
   them a sense of comfort to them.

2) Documentation: We're doing pretty good here. The LDP has excellent books
   available. Also all the "Unix for Dummies" type books on the market are
   starting to fill the gap. The one thing that's yet to be written is
   the type of 15-20 printable manual for each application that in tutorial
   fashion shows users how to use an application without assuming that they
   already know how to use it. Most experts understand that an editor is an 
   editor is an editor, but we get that from experience. So having a tutorial
   document is a good thing.

3) Word Processing: I've finally figured out that outside of the expert class
   that folks really only use computers to write documents. I can name
   a dozen folks that I work with that only used their machines to write
   documents until I introduced them to E-mail, which is essentially a
   another document tool. We need a word processor that native to the
   platform. I know that it's been kicked around before but I have a new
   twist to add: It needn't be complicated. Most of the ideas that have kicked
   around have centered around complex issues like object based semantic 
   models and SGML layout architecture. But at heart there are only 5 or
   6 essential tasks that a WP needs to perform:

   1) Getting text into a document (obviously)
   2) Applying attributes to portions of the text (fonts, size, bold, italics)
   3) Printing
   4) The ability to include graphics
   5) WYSIWYG
   6) Simple interface for commands.
   7) On-line and offline documentation
   
   Before you add more (and there is probably more) ask youself how often
   any other feature is used by the average user. Remember also that we
   already have tools for doing complicated layouts. The key to a good
   WP is that it performs that average WP tasks and not much else.

   I light of this I have a new proposal for getting a simple WP off the
   ground: start with an existing product and extend it into the WP arena.
   I have not yet seen or run Harry's WP yet which may do the job but I'll
   start with an example I'm familiar with: JOE.

   JOE is a simple editor. JOE also has innovative features for an editor 
   (help that stays up while you edit, multiple windows), and has the start
   of rudimentary WP functions. And it's simple which is the most important
   thing. Most importantly JOE doesn't intimidate folks with hard to use
   interfaces or massive complexity. Type joe and start typing. Doesn't
   get much simpler than that. I'm partial to it because I can introduce it
   to novice users (like my 10 YO, or my parents) and they pick it up
   quickly.

   Right now JOE and fullfill items 1 and 7 on my list and these are the
   ideas that I have for the others:

   2) Attributes: Seems to me that SGML has the right procedural model
      for attributes: just tag them with text. So it you want something
      in bold Century 15 pt just tag it:

      [Cent-b-15]This is Century bold 15 text.[def] and some more text
      in regular format.

      Keep it in regular text so that simple editors can edit it and it 
      can be mailed or processed with text tools.

   3) Printing wise I think the tools already exist: simply convert to
      PostScript and Print it out. Or convert the above tags into something
      else. The important thing is that the ability to print should be
      able to happen from the app without leaving it.

   4) Included graphics: Encapulated postscript is an obvious choice. Include
      graphics just like xfig does. Supply a bounding box and the graphic
      will go right in. Question: is there a linux tool that is Like Xpaint
      for VC's. Seems it would be simple to create using the SVGA lib.

   5) WYSIWIG: I'm a believer in text based applications because I've seen
      many situations where one works from terminal, over modems, telnet
      links and the like. So both a text based and X based display model
      need be addressed:

      text: The most important thing is that the app computes the amount of
            space each character takes and only show an appropriate number
            of character for each line. Something in 30 point should only
            show 5 characters to the line if that's all that fits.
            
            The Second most important thing is to somehow highlight the
            attributes for the text. The color console model will be good
            for this. And I think that simply highlighting text that has
            attributes attached for mono terminals is sufficient.

         X: Obviously an X tool can show everything is it's proper form.
            And it should be used when at all possible. How difficult
            would it be to build a TCL/TK front end? I've only played with
            it briefly.

   6) Simple Interface for commands: solution is to use the mouse when at
      all possible. Attribute selection should be done by doing a selection
      type drag over the text, hitting a button, and choosing an attribute.
      the attribute will then be applied to the selected text and the 
      document will adjust to reflect any text changes.

      Also arrow and control key commands should be available so that
      terminal users can still work.

Anyway given these criteria (and I'll gladly accept any other suggestions)
how difficult would be be to either adapt JOE (or HWP) or generate a new
application?

I of course know the charge of "You dreamed it up. You write it!" But I'm
unfortunately too deep in my research to do anything else right now.

Anyway not having something simple to create documents is really hurting 
Linux expansion and the non native tools (like WordPerfect or Word for
DOS) can't do the complete job. Plus it always raises the question "If
I'm running WordPerfect anyway then waht difference does it make under which
OS?"

We need to somehow develop something that simple, different, and better.

I'm aware of the WORD mailing list folks but frankly we haven't heard
a whole lot from them lately. The Linux tide is being stemmed for the lack
of a basic yet essential tool. And this can be complementary to a richer
and more powerful tool. But I think this is something we need to do in
typical Linux fashion: Get it out quickly, let people bang on it, improve
it 'til it's acceptable, then distribute the hell out of it.

Comments accepted, flames ignored.

Later,

BAJ
-- 
---
Another random extraction from the mental bit stream of...
Byron A. Jeff - PhD student operating in parallel!
Georgia Tech, Atlanta GA 30332   Internet: byron@cc.gatech.edu

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

From: troc@loreli.ftl.fl.us (Rocco Caputo)
Subject: Re: SIGHUP: We tried it!
Date: Sat, 21 May 1994 16:41:13 GMT

Theodore Ts'o penned:
> The "ivory tower tribe", huh?  I'll excuse you, and assume you don't
> know how much work I've done working on the Linux kernel.  Some of us
> just have had enough experience doing Unix work that we *know* when some
> idea is half-baked, without having to try it.  :-)

You'll excuse him because you're an Alpha and he's not.  You're allowed
to make assumptions because you're an Alpha and he's not.  Some of us
have enough experience doing Unix work, and that makes them Alphas.

And some of us don't, so we get treated like garbage when we ask an
honest question.  And maybe if the Alphas would give some straight
answers from time to time, people wouldn't have to go around patching
their kernels to find out why SIGHUP shouldn't be propagated past a
session leader.

Three questions and zero-content answers:

  "Why is the sky blue?"
      "Because it's supposed to be that way."
  "Why is the earth round?"
      "Because it's supposed to be that way."
  "Why does SIGHUP only go to session leaders?"
      "Because POSIX says so."

Is that what folks around here consider a decent answer?

--
-><- Signature.

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

From: gareth@ftlinux.demon.co.uk (Gareth Bult)
Subject: Re: 32-bit Novell desktop OS combines Unix,
Date: Sat, 21 May 1994 19:01:58 +0000

Sorry, I missed most of this thread, but has anyone a realistic idea of then this OS is going to appear?

TIA
Gareth.

-- 

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


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