Subject: Linux-Development Digest #734
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:     Thu, 19 May 94 13:13:14 EDT

Linux-Development Digest #734, Volume #1         Thu, 19 May 94 13:13:14 EDT

Contents:
  Socketvar.h (Miles)
  Transputer port of LINUX (Baranski, A.S.)
  Re: Binary compatibility (Rob Janssen)
  Re: In defence of variety (was Re: Distributions considered harmful) (Wayne Schlitt)
  9246 (kempsonc@logica.co.uk)
  Re: 1.0.9 kernel bug (+ partial fix): when using non-loopback address (Elaine Walton)
  Re: Anybody working on BSD dump porting? (Bernd Anhaeupl)
  Re: Distributions considered harmful (Brandon S. Allbery)
  Re: Distributions considered harmful (Brandon S. Allbery)
  Re: Distributions considered harmful (Daniel Barlow)
  Re: The type of "speed_t" in <linux/types.h> (Bruce Evans)
  Re: Distributions considered harmful (Dave Safford)
  X Programming Example Wanted ... (kempsonc@logica.co.uk)
  strftime() bug (Olaf Brandes)

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

Subject: Socketvar.h
From: oursler@kenyon.edu (Miles)
Date: 16 May 94 22:14:38 -0500

Hullo,

        Forgive me if this is obvious, but why don't
        I have a sys/socketvar.h in my include files?

        I went to compile some source and then realized
        it was missing.  Did I forget something?


                                        - Miles -
-- 

        -------------------------------
        Miles Oursler
        Manager of Networks and Systems
        Kenyon College
        oursler@kenyon.edu
        -------------------------------

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

From: v922215@si.hhs.nl (Baranski, A.S.)
Subject: Transputer port of LINUX
Reply-To: v922215@si.hhs.nl
Date: Tue, 17 May 1994 06:57:59 GMT

Hi world,

  Two weeks ago I posted a message concerning a memory management system
idea for the transputer, due to my holliday I wasn't able to read the reply's on
the net. So If you have any idea on the subject or you have a reaction on my
posting please mail me at:


   Antoni.Baranski@si.hhs.nl


Thanx
================
Baranski, A. S.         | Haagse Hoge School  (The Hague Holland)
e-Mail:         | Sector informatica
Antoni.Baranski@si.hhs.nl       | Student Software Engineering
========================
P.S. I am sorry for not metioning this in my original posting.


SU



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

From: rob@pe1chl.ampr.org (Rob Janssen)
Subject: Re: Binary compatibility
Reply-To: pe1chl@rabo.nl
Date: Wed, 18 May 1994 17:05:41 GMT

In <2rcmsv$jil@omega.gmd.de> veit@borneo.gmd.de (Holger Veit) writes:

>The difficulty is with the customers. The systems the potential customers run
>differ as much as the versions. There are about a dozen, partly incompatible
>"commercial" or public collections like Yggdrasil, SLS, Slackware, MCC, Debian
>etc., all of them with different versions of kernel and utilities.
>Some people still run 0.99pl13 ("because it worked stable"), others 0.99pl14s,
>or 1.0.8, or 1.1.? or others. Some people do not want to buy the latest
>CD of a distribution when it comes out, and then have many problems with new
>software, because the shared lib version has recently changed, and further things.

>As a vendor: would you be willing to sell not one version or two or three
>(let's say: one for DOS5.0 and one for DOS6.0), or twenty, thirty, hundred?
>A vendor could provide a CD-ROM with all versions that have been built at the
>company, but this CD-ROM will become outdated before it goes into the production,
>because recently there was a change from kernel 1.X.Y to 1.X.Z.

>And the customer who paid some hundred $ for the fabulous "Word for Linux"
>is disappointed because he would have to downgrade his brandnew Linux 1.X.Z
>to the old and known buggy version 1.X.Y. I guess he won't try such an
>experiment again, and will stay with the, admittedly old and probably
>buggy, but stable commercial SVR4.

I think this is not a correct description of what is happening with
binary compatability in the Linux world.  People who do not run Linux
might thing "geesh.  everytime a new kernel is released I need a new copy
of all my software".

However, this is not the case, and I can't remember any example of this
ever happening in the Linux world since I joined it (at 0.96).

True, the libraries have changed, and there have been introduced new
binary formats that old kernels cannot run.  But an executable that runs
on 0.99pl13 should still run on 1.1.12, as long as you make sure that the
libraries it requests are available on your system.
Of course, this does not mean that every executable you can generate on
1.1.12 will run on 0.99pl13, but that is not what you are considering in
the scenario depicted above.

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

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

From: wayne@backbone.uucp (Wayne Schlitt)
Subject: Re: In defence of variety (was Re: Distributions considered harmful)
Date: Thu, 19 May 1994 05:29:35 GMT
Reply-To: wayne@cse.unl.edu

In article <1994May17.221224.1245@kf8nh.wariat.org> bsa@kf8nh.wariat.org (Brandon S. Allbery) writes:
> In any case, I'm no longer worrying about distributions that much because
> FSSTND (and Debian in particular) seem to be h*ll-bent on disenfranchising
> folks who need to be compatible with other systems for network mounting or use
> of iBCS software which wants iBCS standard paths.  

Pardon my complete ignorance on what FSSTND specifies, but in what way
does it cause problems with networks?  I thought it was supposed to be
very similar to SunOS/SVR4/Solaris.


-wayne

-- 
Vg'f _nznmvat_ ubj znal crbcyr jvyy eha fghss guebhtu ebg13 vs gur grkg
ybbxf tneoyrq, vfa'g vg?

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

From: kempsonc@logica.co.uk
Subject: 9246
Date: Thu, 19 May 1994 12:53:50 GMT

I have just started some X development on my new
Linux workstation and I have come across the following problem:

   Many of the Xt Calls I am used to are no longer available
   eg. XtVaAppInitialize ...  Where are these?  I note that
   XtAppInitialize is available though.

Could somebody who has done some development work under X using
Xt and/or the OpenView widget set possibly send me a code example
of the main program (eg setup and widget creation) plus an
example makefile (preferably including an Imakefile too?).  This
would help me to get started!

Thanks in advance,

Charlie K

==========================================================================
Charlie Kempson                       INTERNET: charlie@mpef4.logica.co.uk
                                                kempsonc@logica.co.uk
Logica Space & Communnications Ltd.,
68 Newman St.,
LONDON W1A 4SE                        TEL: 071-637-9111 ext. 4192

==========================================================================


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

From: ewalton@magnus.acs.ohio-state.edu (Elaine Walton)
Subject: Re: 1.0.9 kernel bug (+ partial fix): when using non-loopback address
Date: 17 May 1994 21:11:33 GMT

I have been working on computers for almost 20 years now.  I do not believe
in perfect code.  It's just not possible.  Consider, if we were to spend our
time on every bug (defect, anomaly, etc.), we would never have the tools
we have.  I feel very grateful to all those who have spent countless years
on this OS.

Back to the topic: "Defects and Acceptable Error".
    The purpose of testing is to locate defects and identify their severity.
Defects are differences between expectations and results.  Severity is
weighs between impact and probability.  The impact indicates "how bad can
bad get."  The probability notes the likelihood that a defect will be
encountered.
    Coding focuses on what we can do within acceptable limits.  If we
spend too much time on extreme corner cases (low probability), diminished
returns result.
    It all comes down to this: do you want the perfect sort routine or an
imperfect (yet functional) OS?  If the defect is too great, offer to help
fix it.

-Sean Walton

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

From: anhaeupl@late.e-technik.uni-erlangen.de (Bernd Anhaeupl)
Crossposted-To: comp.os.linux.help
Subject: Re: Anybody working on BSD dump porting?
Date: Thu, 19 May 1994 12:04:54 GMT

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

   In article <CpyxL3.AzH@pe1chl.ampr.org>, Rob Janssen <pe1chl@rabo.nl> wrote:
   >Because of the many filesystems that Linux supports, trying to port
   >something like "dump" is not simple, and IMHO not a good idea.
   >
   >You are better off making backups with "tar" or "cpio".

           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? :-)


You can make backups over the net with tar! If you have installed /etc/rmt on the
backuphost you can use commands like

        tar cf backuphost:/dev/tape files_to_backup

or
        tar xf backuphost:/dev/tape files_to_extract

with GNU tar.
--
Bernd Anhaeupl                  Tel.:  +49 9131 857787
LATE - Uni Erlangen                     
Cauerstr. 7                     Email: anhaeupl@late.e-technik.uni-erlangen.de
91058 Erlangen

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

From: bsa@kf8nh.wariat.org (Brandon S. Allbery)
Subject: Re: Distributions considered harmful
Date: Wed, 18 May 1994 22:11:07 GMT

In article <Cq0E2t.1Jq@cs690-3.erie.ge.com>, teffta@cs690-3.erie.ge.com says:
+---------------
| >| distribution. Why does Solaris still not have emacs, mosaic, archie,
| >
| >Your argument here falls flat with one counter-question:  Why doesn't SCO
| 
| Gee, the reason nobody includes this stuff is probably because they aren't
| willing to support it without ownership of the code, and the authors aren't
| about to give that away (or sell it).
+------------->8

Explain SCO's inclusion of MMDF, then.  Complete (in releases after 3.2.2, at
least) with UDEL documentation and a pointer of sorts to the original source
distribution.

++Brandon
-- 
Brandon S. Allbery         kf8nh@kf8nh.ampr.org          bsa@kf8nh.wariat.org
The FUDs at Microsoft are shouting "Kill The Wabi!"

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

From: bsa@kf8nh.wariat.org (Brandon S. Allbery)
Subject: Re: Distributions considered harmful
Date: Wed, 18 May 1994 22:15:28 GMT

In article <2rdp5t$5bf@hippo.shef.ac.uk>, ac3slh@sunc.sheffield.ac.uk (Stuart Herbert) says:
+---------------
| : feel a need to differentiate their product from the base System V
| : distribution from USL.  So when they fix a bug, or add a feature, they
| : *don't* send it back to USL.
| 
| And these virulent bouts of featuritis are the reason that 'standards' are
| such a mockery.
+------------->8

Actually, another reason is that USL has a long history of ignoring such
feedback.  Ask old (ex-?)Altos support folks about bugs in SVR3.1 and in sdb.
I think old Dell support folks could tell tales about about SVR4 fixes as
well.

++Brandon
-- 
Brandon S. Allbery         kf8nh@kf8nh.ampr.org          bsa@kf8nh.wariat.org
The FUDs at Microsoft are shouting "Kill The Wabi!"

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

Subject: Re: Distributions considered harmful
From: u92drb@ecs.ox.ac.uk (Daniel Barlow)
Date: 17 May 94 22:14:00 BST

This one will run and run!
In article <NELSON.94May17063549@crynwr.crynwr.com> nelson@crynwr.crynwr.com (Russell Nelson) writes:
>... multiple distributions are not good, because they tend to fragment
>the market.  The Linux market is small enough as it is -- we don't
>need to make it smaller!  If I'm running Slackware, I don't feel like
>I can purchase SLS's support package.  Or if I've bought Yggdrasil's
>CD-ROM, can I install an slackware package?  And not break everything?

Two points I'd make here:
  
(a) I can't see that it matters too much.  If the distributions are
only slightly different, why does it matter which you install?  If
you're running Slackware and SLS won't sell you support, buy it
somewhere else.  (One of) the point(s) of the GPL is after all that
you get your software where you like and pay whoever you want for
service (subject to their having offered support, of course :-) 

(b) What do you mean 'the market'?  All the distributions are
sufficiently similar that programs written for one will work on the
others - hell, probably even the *Novell* distribution will run
everyone else's binaries :-) 

>We should start with a merger between SLS and Slackware.  Send your
>requests for same to Peter and Patrick, imploring them to cooperate
>for the betterment of all.  Yes, they've had words, but they should
>still be able to work together.

We could start with a merger between SLS and the kernel.  I don't know
how much of their module stuff has been incorporated into the standard
kernel, but given that I had to recompile it the other day to get
parallel printer support, probably not much. 

Daniel
--
Daniel.Barlow@sjc.ox.ac.uk              dbarlow@teaching.physics.ox.ac.uk

"He'd never wanted much, except perhaps to be left alone and not woken up
 until midday"                          -- Moving Pictures, Terry Pratchett

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

From: bde@kralizec.zeta.org.au (Bruce Evans)
Subject: Re: The type of "speed_t" in <linux/types.h>
Date: 19 May 1994 21:34:35 +1000

In article <2rehcf$a71@bmerha64.bnr.ca>,
Hamish Macdonald <Hamish.Macdonald@bnr.ca> wrote:
>I'm having problems with some programs dumping core in m68k-linux
>(indirectly) as a result of speed_t being defined as "unsigned int" in
><linux/types.h>

POSIX specifies that speed_t shall be an unsigned integral type.

>Many, many, many applications hard code a "short ospeed" variable (or
>an "extern short ospeed").

ospeed is for the termcap nonstandard.  It's not clear that it is even
in the same units as speed_t's!

>The Linux libc termcap/termio libraries think that "ospeed" is of type
>"speed_t/unsigned int".

That is probably the bug.  The best documentation that I have seen for
termcap is in the gnu termcap package.  Only type `short' is mentioned.
The type probably has to be short for backwards compatibility, although
that is inadequate for holding unsigned shorts, unsigneds and unsigned
longs.  This would work under linux, since there the largest speed is
15.  It is unlikely to work if the speed is encoded more flexibily.
E.g., the speed might be encoded as 10 * (unsigned long)speed_in_bps.
The factor of 10 helps to represent fractional speeds like 134.5 and
the type has to be unsigned long to represent even the largest standard
speed (10 * 38400 = 384000 may be larger than UINT_MAX).
-- 
Bruce Evans  bde@kralizec.zeta.org.au

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

From: drs0587@net.tamu.edu (Dave Safford)
Subject: Re: Distributions considered harmful
Date: 17 May 1994 21:21:09 GMT
Reply-To: drs0587@net.tamu.edu

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

While I understand your point, and respect your frustration, let me 
offer a counter viewpoint.

I think any of the major linux distributions are outstanding efforts
that are as good as, if not superior to commercial distributions
such as Solaris. The fact that linux users have a choice among several
strong distributions is certainly better than having no choices.
On a Sun there is no operating system distribution competition, and
as a result, you see almost no movement toward including significant
features/tools that we have come to expect in any decent linux 
distribution. Why does Solaris still not have emacs, mosaic, archie,
ftptool, mtools, perl, tcl/tk, gnu tar, tn3270, gzip, and patch,
or at least put out a catalyst cdrom with these as binary and source packages?
 
<We've got at least four different package conventions (SLS, Slackware,
<Debian and Yggdrasil).  Why?  No good reason I can see.  FTP sites
<carry these different distributions, most of which differ only slightly.

There are substantial differences in the distributions, particularly in content,
organization of source code, installation method, and kernel module support.
The good news is that the FSSTND effort has been accepted by all major
releases, so that users will see a consistent file layout between
distributions, while still enjoying their differences.

<And multiple distributions are not good, because they tend to fragment
<the market.  The Linux market is small enough as it is -- we don't
<need to make it smaller! 

1. Multiple distributions increase the market. Different users have different
needs and preferences that are not currently (and may never) be met by any
one release.  If we don't offer what they want, they will go elsewhere.
If one distribution does offer everything to everyone, then it will
succeed, and the others disappear anyway.

2. Competition between the releases has been a tremendous force for
innovation and progress.  It is a story that keeps repeating -- one
distribution adds a new feature, and if it works, it soon shows up
in the others.  Do you think that Slackware or Debian would be where they
are now without learning from their predecessors?

<We should start with a merger between SLS and Slackware.  Send your
<requests for same to Peter and Patrick, imploring them to cooperate
<for the betterment of all.  Yes, they've had words, but they should
<still be able to work together.

I fear that a forced cooperation would serve only to slow down their
efforts, and reduce a user's choices.  

My vote is for free competition, rather than coerced standardization.

dave safford

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

From: kempsonc@logica.co.uk
Subject: X Programming Example Wanted ...
Date: Thu, 19 May 1994 15:38:00 GMT

I have just started some X development on my new                               
Linux workstation and I have come across the following problem:     
  
   Many of the Xt Calls I am used to are no longer available       
   eg. XtVaAppInitialize ...  Where are these?  I note that    
   XtAppInitialize is available though.         
  
Could somebody who has done some development work under X using  
Xt and/or the OpenView widget set possibly send me a code example             
of the main program (eg setup and widget creation) plus an     
example makefile (preferably including an Imakefile too?).  This         
would help me to get started!
  
Thanks in advance,     
  
Charlie K    
  
==========================================================================      
Charlie Kempson                       INTERNET: charlie@mpef4.logica.co.uk 
                                                kempsonc@logica.co.uk 
Logica Space & Communnications Ltd.,     
68 Newman St.,   
LONDON W1A 4SE                        TEL: 071-637-9111 ext. 4192               
  
==========================================================================    


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

From: brandes@cs.tu-berlin.de (Olaf Brandes)
Subject: strftime() bug
Date: 19 May 1994 13:04:14 GMT

Hi folks,

I've come across the phenomen that strftime()'s %j outputs the same value
as the struct tm member tm_yday.
According to POSIX, the %j output value should be in the range from 001
to 366. Actually, values 000 to 365 are returned.

I guess there aren't too many applications using %j. But if there are
some, I donn't know what's going to happen.

Gretings
  Olaf

--
 Olaf Brandes        "There are a finite number of jokes in the universe."
  TU  Berlin                                                   David Byrne

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


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