Linux-Misc Digest #240, Volume #19 Mon, 1 Mar 99 01:13:12 EST
Contents:
Re: Microkernels are an abstraction inversion (Francois-Rene Rideau)
Re: Can NT with NTFS coexist with RedHat Linux (Michel Catudal)
Re: Can NT with NTFS coexist with RedHat Linux (Michel Catudal)
Re: MS Explorer 4.0 for Unix (Michel Catudal)
Re: Will ORACLE8.05 install without kernel recompile (Scott Webb)
Re: can Linux see windows 95/98 network neighborhood?? (Tim Moore)
Re: Win95 vs. Win98 and Linux (Tim Moore)
Re: running executables from cdrom? (Tim Moore)
Re: sharing .netscape (Tim Moore)
Re: Customising Xdm (Dave Edick)
Re: installing components once linux has been installed (FusionGyro)
Re: Where is the config file that sets which librarys at strart up? (Greg White)
Re: Missing Screen Savers in RedHad v5.2? (Tim Moore)
Re: Best Free Unix? (why FreeBSD?) (Leslie Mikesell)
Re: Exporting Windows filesystem for Linux... (Perry Pip)
Re: Any JDK for Linux? ([EMAIL PROTECTED])
----------------------------------------------------------------------------
From: Francois-Rene Rideau <[EMAIL PROTECTED]>
Crossposted-To: gnu.misc.discuss,comp.os.linux.advocacy
Subject: Re: Microkernels are an abstraction inversion
Date: 01 Mar 1999 05:20:11 +0100
Emile van Bergen <[EMAIL PROTECTED]> writes:
> Francois-Rene Rideau wrote:
FR> [...] the only thing �K does is introduce stupid low-level barriers
FR> between services. [...]
> Among those 'stupid' low level barriers is address space separation.
Indeed.
> I wouldn't want to live without it, now would you?
I certainly would want to live without it!
Why can't I? Because I have to run intrinsically unreliable software.
Why is software so unreliable? Because people use shoddy low-level languages.
Why don't users care? Because they're trained to not see the sources anyway,
and can't make the difference.
Why don't corporate programmers don't use higher-level languages?
Because high-level languages depend on elaborate runtimes and libraries,
and corporate hoarders can only use whatever lies within
their license barriers, which gets worse when they need interoperability.
Happily, there is free software.
With free software, users care, and so do programmers.
With free software, programmers do not hesitate to develop elaborate
high-level languages: Elisp, Perl, Scheme, ML, Haskell, Mercury.
Of *course* when you have to face stupid code written by stupid people
in stupid languages, you have to use stupid low-level barriers.
But hopefully, that's as seldom the case as possible
(one day, that'll be only for emulating legacy Win/Lin/* software).
In particular, that doesn't have to be the case AT ALL
when developing a free software OS infrastructure!
> As to the requirement
> of data to be marshaled and un-marshaled: this has a rather nice
> side-effect, in that most developers will design the necessary links
> between modules as clean and stateless as possible. This in itself is
> beneficial, mainly for re-useability and debugability reasons.
By following your argument to its extreme point
(which is the one and only test for an argument),
the ideal system would require that between any two functions
(why stop to functions? let's make it instructions!)
all data has to be marshalled and un-marshalled,
which would lead to complete re-usability and debuggability!
This is utter rubbish.
People are free to design clean and stateless things
even without forced low-level barriers.
The barriers doesn't BRING *anything*, it only gets in the way,
for the (common) cases when there HAS to be shared state
(be it only locks, etc, for transactions).
And if writing clean and stateless modules was always the best,
then the bestest system programming language would be Haskell or Clean!
> Also, a modularised design with simple links that can be made
> network-transparent (message passing), will scale better towards large
> scaling distributed environments, without much re-design.
Mind you, concurrent programming languages like JOCAML not only
are fully network transparent, they also make re-implementation
(not just re-design) useless, and their compiler also optimize
the local case, so you don't pay the price of a low-level "modular" design.
> Formally, you're absolutely right. In the real world, you're absolutely
> not.
This couple sentence is despicable crap. It's an insult to intelligence.
Formal thinking is there to describe the world in useful ways.
It's a one world, and something is formally right iff it's informally right.
> Program functions are almost never formally/mathematically
> 'proven', so I want those protectional barriers so that fail-safety
> mechanisms are more easily implemented.
Certainly every single program written in LISP, ML, Perl, Modula-3, Haskell,
Mercury, Prolog, or otherwise high-level language, is formally proven
to never ever do an unauthorized memory access,
[least you explicitly do unsafe operations].
This is already much more than stupid low-level memory protection
will *ever* bring to you.
> Apart from that, there will
> never be clean and uniform inter-module interfaces unless 'forced' upon
> designers by the barriers (i.e. protocols) you mention.
Why wouldn't there be? Do you think programmers are stupid?
If they are possible (in whatever language, including C),
programmers will use them whereever useful.
If they are not possible, mandating them everywhere by force,
regardless of their utility, won't make them happen.
That's the typical fascist pig approach!
> As long as different components of a system come from different
> sources that one can never fully trust, i.e. as long as Utopia doesn't
> exist, I want process separation.
As long as people stupidly take the present state of the computer art
(an ever-fast evolving technology!) as an eternal truth, they say rubbish.
First, if you can't trust software, don't run it!
Second, if you partially trust software, you can sandbox it!
Third, software, security and sandboxes are high-level concepts,
whereas processes and address spaces are low-level concepts.
Directly mapping high-level concepts into low-level ones is called
by definition naive implementation.
Requiring people to do naive implementation by hand and then not trust
each other is as stupid as can get.
Instead, I propose that you open your mind to the concept of a compiler,
whereby the computer does the work of mapping higher-level programs
into lower-level instructions, in ways that preserve system invariants.
And in case you haven't heard about it despite the hype,
let me introduce you to the concept of compilers that work even for
stupid low-level code for pseudo-"portable" second-zone language bytecodes
(meant to lengthen the evil reign of binary compatibility
with proprietary code): even braindead languages can be reasonably efficiently
implemented in a way that is intrinsically secure,
without the need for stupid low-level barriers.
> I'm sorry mr. Rideau, but I think you should get both your feet on the
> ground.
I think you should get both your brain hemispheres up and running for once.
Well, at least either. See? ARGUMENTUM AD HOMINEM is sooooo easy!
> In the software world as it is
> now, more problems are caused by the absence of narrow interfaces and
> component separation than are caused by this 'abstraction inversion' you
> scream about.
So what? The topic was: "microkernels", I explained why they were evil.
Happily, they are mostly no more a problem today, since people,
even without understanding why, mostly left them for better designs
(and ANY design is better).
If you want to talk about interfaces, well, yes,
shoddy interfaces are a problem. And again, the problem is: C.
The C language doesn't allow to specify high-level interfaces
with rich semantics. No amount of low-level explicitation
through marshalling can help.
The problem is the expressive power of the interface definition language.
Richly typed languages such as ML or Modula-3 (as opposed to one-typed LISP)
may help quite a bit, here.
Even in C, metaprogramming as demonstrated in Tom Lord's ctool
can help a lot at maintaining elaborate system invariants throughout C code.
> I'm sorry if I'm not too polite to you, but in the world
> we live in I see no reason for such hot-tempered statements as those
> expressed by you.
I don't give a damn about your pseudo-politeness or lack thereof.
On USENET, the only relevant politeness is being up to the point,
and developing non-empty arguments, instead of wasting people's
bandwidth and time with rubbish and fallacies.
FR> Requiring the system programmer to emulate, by hand, in C,
FR> a strongly typed concurrent agent programming model,
FR> and shooting him if he makes the slightest mistakes,
FR> is not only a REAL STUPID design decision, it's also deeply EVIL.
FR> The solution is, again:
FR> to achieve a system that follows the above model,
FR> use a strongly typed concurrent agent programming language! [duh!]
>
> This is exactly the 'fascist' (your wording!) approach that you so
> strongly oppose.
Certainly not.
If you consider strong typing as a limitation,
explain to me how perl is a limitating language
(and forget the lack of efficiency of the perl compiler;
other strongly typed languages have quite good compilers).
> What is the formal difference in discipline enforced by
> the language and discipline enforced by the run-time environment?? I
> prefer the latter, personally, I like the freedom of C.
You FOOL! You should learn about the notion of a compiler,
and more generally, think about the fact that
computers are there to relieve man from automatable tasks.
A language that provides constructs for concurrent agent system
doesn't ENFORCE, it ENABLES; it relieves you from the details of execution.
Its runtime system is IMPLICIT. The programmer doesn't have to WORRY about it.
A design with a language that does not provide constructs instead
FORCES the details upon you without ENABLEing you;
it forces you to manually respect stringent consistency rules,
that is the invariants required by the runtime;
instead of relieving the programmer, it puts a new burden on him.
And that, regardless of whether he needs it.
As a side-effect, a high-level language can do compile-time optimizations
that are completely inaccessible to someone explicitly hand-emulating things.
Such optimizations are implicit, orthogonal and maintainable;
explicit hand-optimization would be expensive and very unstable
by its dependence on code evolution.
I suggest that you learn a few high-level programming languages first
(assuming you're not hopelessly closed-minded and/or stupid),
before you even try to continue this discussion.
I suggest LISP or Haskell as both very providing excellent insight,
and as a complete change from your deeply entrenched computing prejudices.
> To quote an old
> C reference manual: "With C, you've got enough rope to swing with, and
> enough rope to hang yourself with."
So what? The quote was as compared to Pascal,
a notoriously fascist language with roughly the same abstraction power.
The problem of C is its lack of abstraction power
(for better and worse, it's a clumsy restricted portable assembler),
that makes it unsuitable to specify high-level interfaces,
as needed between components of a concurrent agent system.
The fascism here is forcing people to use an unsuitable language
for their task at hand.
You wouldn't imagine requiring of any project that has been
written in LISP, Perl, Erlang, etc, that it had been written in C, would you?
C is just unsuited to high-level data manipulations.
Well, with the �K design in particular, and other all-C design,
that's precisely what you do.
[ "Far�" | VN: Уng-V� B�n | Join the TUNES project! http://www.tunes.org/ ]
[ FR: Fran�ois-Ren� Rideau | TUNES is a Useful, Nevertheless Expedient System ]
[ Reflection&Cybernethics | Project for a Free Reflective Computing System ]
A language that doesn't affect the way you think about programming,
is not worth knowing. -- Alan Perlis
------------------------------
From: Michel Catudal <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.linux,comp.os.linux.admin,comp.os.linux.networking,comp.os.linux.questions
Subject: Re: Can NT with NTFS coexist with RedHat Linux
Date: 28 Feb 1999 22:59:02 -0600
Jim Esparza wrote:
>
> Have been reading quite a few of the HOWTO's lately trying to get
> Redhat 5.2 installed on my machine. According to the Linux+NT-Loader
> document, standard Linux can't access NTFS.
This is nonsense. When I compile the kernel I have the option to install NTFS support.
I have RedHat 5.2 with kernel 2.2.2 and I can read my NTFS partition without any
problem.
--
Tired of Windows' rebootive multitasking?
then try Linux's preemptive multitasking
http://www.netonecom.net/~bbcat/
We have software, food, music, news, search,
history, electronics and genealogy pages.
------------------------------
From: Michel Catudal <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux,comp.os.linux.admin,comp.os.linux.networking
Subject: Re: Can NT with NTFS coexist with RedHat Linux
Date: 28 Feb 1999 23:02:06 -0600
[EMAIL PROTECTED] wrote:
>
> Actually the new 2.2.x kernels include support for writing to NTFS partitions
> as well but it is an experimental driver and they warn that it has the
> potential of corupting your NTFS partition.
It is more than having the potential, I have to reinstall NT after my partition
was hozed. I keep NT or read only for now. I use a fat partition to transfer data
back and forth. It is actually my DRDOS partition.
--
Tired of Windows' rebootive multitasking?
then try Linux's preemptive multitasking
http://www.netonecom.net/~bbcat/
We have software, food, music, news, search,
history, electronics and genealogy pages.
------------------------------
From: Michel Catudal <[EMAIL PROTECTED]>
Subject: Re: MS Explorer 4.0 for Unix
Date: 28 Feb 1999 23:04:21 -0600
ACE Alex wrote:
>
>
> Im a web developer and sadly I have to say that IE4 is far better with java
> script and Css. Netscape claims they have full suport for it but i dont
> agree with that!
>
> So yes,, as long as netscape cant show every page on the net i would say
> that there would be nice to have ie 4!
The stuff that Netcape can't handle is actually illegal crap setup by Microsoft.
In case you wouldn't know there is a court order for Microsoft to remove their shit
off the market and make it compatible to Java and they need Sun's blessing.
--
Tired of Windows' rebootive multitasking?
then try Linux's preemptive multitasking
http://www.netonecom.net/~bbcat/
We have software, food, music, news, search,
history, electronics and genealogy pages.
------------------------------
From: Scott Webb <[EMAIL PROTECTED]>
Reply-To: [EMAIL PROTECTED]
Crossposted-To: comp.os.linux.setup
Subject: Re: Will ORACLE8.05 install without kernel recompile
Date: Sun, 28 Feb 1999 23:13:43 -0500
[EMAIL PROTECTED] wrote:
>
> Hi mates!
>
> Fighting with recompiling kernels on Linux laptops is a real hessle.
>
> Does anybody know if ORACLE8.05 will install and run (not in big volumes, just
> for demo purpuses) on REDHAT Linux 5.2 (kernel 2.036) without kernel
> recompile and relink??
>
> I would appreciate your comments.
>
> Cheers
>
> Shay Tochner
> International Systems Support Specialist
>
> -----------== Posted via Deja News, The Discussion Network ==----------
> http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
I have been fighting the same battle. I have re-compiled the kernel
twice to set the appropriate variables. Now when I try to read the
CD-ROM drive, it says iso9660 fs is not supported. I recompiled
specifying Yes to iso9660 support, but it didn't help. Now I have to
figure out how to have the proper module loaded. Any help would be
appreciated. Thanks!
Scott Webb
[EMAIL PROTECTED]
------------------------------
Date: Sun, 28 Feb 1999 21:03:06 -0800
From: Tim Moore <[EMAIL PROTECTED]>
Subject: Re: can Linux see windows 95/98 network neighborhood??
ls /usr/doc/samba*
> as subject stated, can linux see it??
> if there is some HOWTO, or docs, please direct me.
--
[Replies: make the double y a single]
"Everything is permitted. Nothing is forbidden."
WS Burroughs.
------------------------------
Date: Sun, 28 Feb 1999 21:07:02 -0800
From: Tim Moore <[EMAIL PROTECTED]>
Subject: Re: Win95 vs. Win98 and Linux
> under win 3.1 (e.g. FreeIBComponents for use with Delphi & Interbase). So,
> considering all aspects - which is the least grief? Which occupies the least
> disk space? Which is easier to handle in a mixed environment (using Wine
> etc.)?
w95a, which I still use for video editing, is fairly stable if you don't
mess with it, and straight forward (e.g.- no IE crap).
w98 possesses all the characteristics of a virus *and* you pay $100 to get
it.
--
[Replies: make the double y a single]
"Everything is permitted. Nothing is forbidden."
WS Burroughs.
------------------------------
Date: Sun, 28 Feb 1999 20:56:53 -0800
From: Tim Moore <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup,comp.os.linux.hardware
Subject: Re: running executables from cdrom?
from mount(8):
user Allow an ordinary user to mount the file system.
This option implies the options noexec, nosuid, and
nodev (unless overridden by subsequent options, as in
the option line 'user,exec,dev,suid').
I use 'iso9660 defaults,noauto,ro,user,unhide,exec'
> shouldn't be a problem, but anyway, when I mount the cd and try to run a
> program which has these permissions: '-r-xr-xr-x 1 root root', I still get
> a Permission denied error. Does anyone know why this is the case? Thanks for
> any info. BTW, this is how I mount it (from my fstab):
>
> /dev/cdrom /cdrom iso9660 noauto,ro,user 0 0
--
[Replies: make the double y a single]
"Everything is permitted. Nothing is forbidden."
WS Burroughs.
------------------------------
Date: Sun, 28 Feb 1999 21:12:23 -0800
From: Tim Moore <[EMAIL PROTECTED]>
Subject: Re: sharing .netscape
Not a good thing to share.
> 1. user1 want to share .netscape with user2.
> 2. user2 made symlink like as 'ln -s /home/user1/.netscape'
> 3. user2 made setuid user1 script which just excute netscape.
--
[Replies: make the double y a single]
"Everything is permitted. Nothing is forbidden."
WS Burroughs.
------------------------------
From: [EMAIL PROTECTED] (Dave Edick)
Subject: Re: Customising Xdm
Date: Mon, 01 Mar 1999 05:21:00 GMT
James,
How do you change the background on the xdm login screen?
/Dave Edick/
On Thu, 25 Feb 1999 20:33:19 -0600, John Thompson
<[EMAIL PROTECTED]> wrote:
James Porritt wrote:
>
> Is there anyway to customise the Xdm login screen without hacking the
> source? I can get images to appear in the background, but have found no
> way of altering what gets displayed in the login window.
Change the "xlogin*greeting: " line in
/etc/X11/xdm/Xresources
I haven't found a way of making it more than one line,
though.
--
-John ([EMAIL PROTECTED])
--
/Dave Edick/ dedick at home dot com.
or remove the hates.spam part from the header
------------------------------
From: FusionGyro <[EMAIL PROTECTED]>
Crossposted-To: alt.linux,comp.os.linux,comp.os.linux.redhat,comp.os.linux.setup
Subject: Re: installing components once linux has been installed
Date: Sun, 28 Feb 1999 22:24:02 -0700
Ivo Sbalzarini wrote:
>
> this will install the component and update your systam database, so you will be
> able to uninstall it later on just doing a:
>
> rpm -u <package_name>
>
Actually, it's
rpm -e <package_name>
to remove it (think erase), and
rpm -U <package_name.rpm>
to upgrade a package.
FusionGyro
------------------------------
From: Greg White <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux.slackware
Subject: Re: Where is the config file that sets which librarys at strart up?
Date: Sun, 28 Feb 1999 21:26:09 -0800
NecX wrote:
>
> Simple.... Look in the the scripts in the /etc/rc.d directory. (Particularly
> rc.M, rc.S, and rc.modules) The command to active the removed packages is
> there. Just coment in out and reboot. the error and the process call are gone.
>
Actually, in this case, I don't think so...
> Nectioch
>
> Joe (theWordy) Philbrook wrote:
>
> > I.m running slackware 3.5... A ways back I tried <unsuccesfuly> to install
> > slrn, slrnpull and the slang library that slrn's docs said was required...
> >
> > I couldn't get it running in the time I had for this so I removed the
> > packages... I found the packages on the achive disk that came with my
> > cheepbytes slack 5 cd set. But I had to rpm2targz and installpkg to put
> > them in... to remove them I used removepkg... BUT ever since then I'm
> > getting this during the start up...
> >
> > -snip. . . . . . . . . . .sig
> >
> > Starting daemons: syslogd klogd portmap inetd lpd mountd nfsd
> > /sbin/ldconfig: warning: can't open /usr/lib/libslang.so.1 (No such file or
> > directory), skipping
> > Starting sendmail daemon (/usr/sbin/sendmail -bd -os)...
> > Running gpm...
> >
> > -snip. . . . . . . . . . .sig
> >
> > Would somebody please tell me how to stop my linux from looking for that
> > library... Please...
It will be somewhere in /etc/ld.so.conf, unless I'm sorely mistaken...
:)
> >
> > Thanking you in advance
> >
> > | ~^~ ~^~
> > | <?> <?> Joe (theWordy) Philbrook
> > | ^ J(tWdy)P
> > | \___/ <<[EMAIL PROTECTED]>>
------------------------------
Date: Sun, 28 Feb 1999 21:20:56 -0800
From: Tim Moore <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.questions,comp.os.linux.setup
Subject: Re: Missing Screen Savers in RedHad v5.2?
'rpm -qa | grep screen' or 'man xscreensaver'
> I just installed Red Hat v5.2 and XFree86. When I checked out the
> Screen Savers.... They don't seem to be anywhere.
> Can anyone tell me what happened to them?
> Previous installs had them all installed.
--
[Replies: make the double y a single]
"Everything is permitted. Nothing is forbidden."
WS Burroughs.
------------------------------
From: [EMAIL PROTECTED] (Leslie Mikesell)
Crossposted-To:
comp.unix.questions,comp.unix.advocacy,comp.unix.misc,comp.unix.bsd.freebsd.misc
Subject: Re: Best Free Unix? (why FreeBSD?)
Date: 28 Feb 1999 22:54:19 -0600
In article <7bcqvl$211$[EMAIL PROTECTED]>,
Tim Smith <[EMAIL PROTECTED]> wrote:
>david parsons <o r c @ p e l l . p o r t l a n d . o r . u s> wrote:
>> The GPL lets you modify the code it applies to, and Linus explicitly
>> allows binary-only modules.
>
>Linus does not have the power to make such an exception.
>
It isn't an exception about derived works being covered, it is
a definition of the module interface that says that such
modules do not become derived works of the kernel. While there
still might be a legal question here, I don't see how anyone
can consider Linux as anything but an expert on this topic.
Les Mikesell
[EMAIL PROTECTED]
------------------------------
From: [EMAIL PROTECTED] (Perry Pip)
Crossposted-To: comp.os.linux.setup,comp.os.linux.advocacy
Subject: Re: Exporting Windows filesystem for Linux...
Reply-To: [EMAIL PROTECTED]
Date: Mon, 01 Mar 1999 05:30:51 GMT
In comp.os.linux.setup, you wrote:
>Hi all,
>
>Is there a way to export MS-Windows filesystem and let Linux mount it?
>I have tried the following:
>
>mount -t msdos 123.123.123.2:/public /mnt/public
>However, it wasn't successful...
>This reason that I am asking is because I wanted to use WINE to run some
>of the Win95's applications on my Linux... but unsuccessfully... Would
>Samba work? I thought Samba is just like FTP... you can only transfer
>files but not share the same file system.
Samba will do both. You can use the samba server make your linux files
available to your windows box. But to mount shares from your
windows box onto your linux box you need smbfs. Try:
man smbmount
If it isn't there you'll need to install the smbfs mount command and
possibly smbfs support in your kernel.
Perry
>Thanks in advance...
>
>Brian
>
--
Show the code....or hit the road.
Perry Piplani www.open-systems.com
[EMAIL PROTECTED] perrypip.netservers.com
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To: comp.lang.java.setup,comp.lang.java.api
Subject: Re: Any JDK for Linux?
Date: 28 Feb 1999 17:34:24 GMT
Jet <[EMAIL PROTECTED]> sez:
--> You might want to write to solaris_jdk12_feedback at Sun dot COM.
--> J
--> James Tam wrote:
-->>
-->> Hi,
-->>
-->> I am looking for a JDK for Linux.
-->> Anyone know anything?
www.blackdown.org
They are not done with the 1.2 port but they do have a 1.1.7b port
-->>
-->> Thanks
-->> James
--
Hercules Huggins Federal Express Corporation
(407)916-3863 [EMAIL PROTECTED]
(888)935-0004 [EMAIL PROTECTED]
PGP Fingerprint: 9B 5C 2C 21 B4 77 C5 96 AD 99 44 B2 CA A8 45 14
Send email for my pgp key.
------------------------------
** 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: [EMAIL PROTECTED]
You can send mail to the entire list (and comp.os.linux.misc) via:
Internet: [EMAIL PROTECTED]
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Misc Digest
******************************
