Linux-Misc Digest #207, Volume #20 Fri, 14 May 99 18:13:08 EDT
Contents:
Re: Pro-Unix vs anti-WinTel (Paul D. Smith)
Re: kernel compile error ("Spud")
Re: Proper use of /usr/local (Re: The Best Linux distribution?) (Alexander Viro)
KDE 1.0 -> KDE 1.1.1. update, wie mu� ich vorgehen? (Stefan Drews)
How can I back *everything* up? (marco tephlant)
kernel compile error (Bruce Linton)
Re: Symbolic Link ("J�rgen Exner")
Re: 'Find'. what a strange command (Andre Kostur)
Re: Pro-Unix vs anti-WinTel ("G. Sumner Hayes")
Netware RPL Boot ("Dave Williams @ Home")
Shutdown (and X) errors. (Gavin McCord)
Re: Red Hat info needed. (Johan Kullstam)
Re: A bash question (NF Stevens)
Re: Food for thought (Mark Hahn)
Re: GNU reeks of Communism (Frank Sweetser)
Re: SuSE 6.1 for cheap... ([EMAIL PROTECTED])
Re: Proper use of /usr/local (Re: The Best Linux distribution?)
([EMAIL PROTECTED])
Re: What happened to fdformat ([EMAIL PROTECTED])
Re: Pro-Unix vs anti-WinTel (Richard Caley)
Linux Software Archive + more (Never spam a hacker)
Re: page faults (NF Stevens)
Re: Pro-Unix vs anti-WinTel (Doug DeJulio)
Re: Pro-Unix vs anti-WinTel ("G. Sumner Hayes")
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (Paul D. Smith)
Crossposted-To: comp.os.linux.development.apps,comp.unix.bsd.freebsd.misc
Subject: Re: Pro-Unix vs anti-WinTel
Date: 14 May 1999 16:27:04 -0400
Reply-To: [EMAIL PROTECTED]
%% [EMAIL PROTECTED] writes:
z> Ports is probably the only package system that *does* have a
z> central repository.
What exactly do you mean by that? Debian certainly does. Even RedHat
does, basically (although they have a lot of unofficial stuff on their
site, too, and it's less well-organized). I'm not following your
terminology.
z> access through CVSUP, which makes keeping an upto date ports tree
z> *trival*. So much so that many, myself included, simply stick the
z> job in a cron tab:
z> From my /etc/weekly.local:
z> /usr/local/bin/cvsup -E -g -Z /usr/share/examples/cvsup/ports-supfile
Unfortunately I don't have the disk space or 'Net bandwidth to download
all the code, much less keep it. However, I agree it would be nice to
have the option; you can obviously get the source for all Debian
packages but it's not this trivial (it's not hard, though).
But I can do this in my /etc/weekly.local (or the equivalent) on my
Debian system, and get the same effect without compiling anything:
apt-get update
apt-get upgrade
apt-get clean
(basically) and Viola [sic]!
z> One word, "libc". Linux above all else should know the pain of not
z> building from source.
With proper dependency checks, this isn't a problem. libc5->libc6 was
easy, once people thought about it for a minute, since the sonumber
changed; just keep both.
The changes between glibc 2.0 to 2.1 had a small enough impact that a
dependency scheme is perfectly appropriate. I upgraded my Debian system
to glibc 2.1.1 and everything works fine. I didn't have to build
anything locally.
z> Actually, for stability, consistency, ease of use, and legal
z> reasons, it *is* thus far the best system going.
I disagree on all those points, of course, when compared to my Debian
system, but that's largely a matter of personal preference.
What would be great is if the Debian stuff was upgraded to allow a
ports-like source code distribution mechanism, and/or ports was upgraded
to provide prebuilt binaries as easily and slickly as Debian.
Then we could have the best of both worlds!
--
===============================================================================
Paul D. Smith <[EMAIL PROTECTED]> Network Management Development
"Please remain calm...I may be mad, but I am a professional." --Mad Scientist
===============================================================================
These are my opinions---Nortel Networks takes no responsibility for them.
------------------------------
From: "Spud" <[EMAIL PROTECTED]>
Subject: Re: kernel compile error
Date: Fri, 14 May 1999 20:27:10 GMT
I just saw someone post a solve to this problem on one of the Linux
newsgroups not two days ago. The problem was that you didn't install a
program (it was xx86, maybe bc86?), and it is pretty necessary. So go back
and look for that proggie and install it. :)
>Can someone please help decipher this error message and point me in a
>direction to solve the problem? I'm upgrading from 2.0.7 via redhat
>rpms, have followed all the instructions to the letter, cleanly
>installed all the necessary packages, confirmed that the pkgs have
>correct md5sums, etc., etc., etc.
>
>...
>...
>gcc -D__KERNEL__ -I/usr/src/linux-2.2.5/include -E -traditional
>-DSVGA_MODE=NORMAL_VGA bootsect.S -o bootsect.s
>as86 -0 -a -o bootsect.o bootsect.s
>make[1]: as86: Command not found
>make[1]: *** [bootsect.o] Error 127
>make[1]: Leaving directory /usr/src/linux-2.2.5/arch/i386/boot
>make: *** [zImage] Error 2
>
>
>Thanks,
>Bruce
>
------------------------------
From: [EMAIL PROTECTED] (Alexander Viro)
Crossposted-To: comp.unix.bsd.freebsd.misc
Subject: Re: Proper use of /usr/local (Re: The Best Linux distribution?)
Date: 14 May 1999 16:05:36 -0400
In article <7hhst6$561$[EMAIL PROTECTED]>, Leslie Mikesell <[EMAIL PROTECTED]> wrote:
>In article <7hh690$[EMAIL PROTECTED]>,
>Alexander Viro <[EMAIL PROTECTED]> wrote:
>>In article <7hgkjt$nco$[EMAIL PROTECTED]>, <[EMAIL PROTECTED]> wrote:
>>>This makes everyone happy:
>>>
>>>/bin - system binaries
>>>/usr/bin - distribution binaries
>>>/usr/local/bin - custom binaries
>>>
>>>[just throw *sbin away, I guess]
>>>
>>>Whaddya think?
>>
>>Extremely bad. Big root filesystem is less than bright idea even with
>>big disks. The most fundamental thing about disks being: they fail. At
>>the most inconvenient moments.
>
>I haven't seen any relationship between size and failure. Actually
Yes, but if the disk started to develop bad blocks your chances are better
if you are a small target ;-)
>>have root small. Keeping everything on one filesystem... Well, good luck
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>>cleaning the mess after big mailbomb targeted to one of your lusers.
>
>None of that should hit / or /usr, regardless of them being on
>the same fillesystem. Given the system invasion into /usr, everyone
>moved users to /home long ago.
Fine, so you are agree to take /tmp, /var and /var/spool apart (I'ld cut
on /var/spool/mail and /var/spool/news, actually, but then it depends on
the site). That's better. As for the / vs. /usr - ever heard of NFS mounting?
--
"You're one of those condescending Unix computer users!"
"Here's a nickel, kid. Get yourself a better computer" - Dilbert.
------------------------------
From: [EMAIL PROTECTED] (Stefan Drews)
Subject: KDE 1.0 -> KDE 1.1.1. update, wie mu� ich vorgehen?
Date: Fri, 14 May 1999 19:22:46 +0200
Ich arbeite zur Zeit mit SuSE Linux 6.0. Nun m�chte ich mein KDE 1.0 auf die
neue Version 1.1.1 updaten, wie gehe ich vor???
Ich habe mir gestern den KDE1.1.1 (Slackware Edition) aus dem Netz gezogen.
Ich danke !!!
Stefan
PS.:
Gute Linux Homepage
http://www.rz.fh-heilbronn.de/mirrors/linuxberg/x11soft.html
------------------------------
From: [EMAIL PROTECTED] (marco tephlant)
Subject: How can I back *everything* up?
Date: Fri, 14 May 1999 21:39:05 +0100
As a real beginner I frequently mess up my whole linux installation,
whats getting to me is having to redo everything when I reinstall, ppp
settings, etc..
For windows I used tar to make a bzipped "image" of my C drive, so when I
want to reinstall I delete everything on C and untar from within linux,
it takes less than ten minutes and I dont have to go hunting round for
Cd's with drivers on, dont have to renter email settings and lilo remains
intact aswell! Is there anyway I can do this in linux? I can't think of
anyway of backing everything up and don't know what specific bits should
be backed up. Recently I have been mucking up my system trying to
install the latest version of Gnome and Enlightenment - who says RPM's
are easy?!?
Any suggestions?
Cheers
--
Marco
------------------------------
From: Bruce Linton <[EMAIL PROTECTED]>
Subject: kernel compile error
Date: Fri, 14 May 1999 11:25:50 -0700
Can someone please help decipher this error message and point me in a
direction to solve the problem? I'm upgrading from 2.0.7 via redhat
rpms, have followed all the instructions to the letter, cleanly
installed all the necessary packages, confirmed that the pkgs have
correct md5sums, etc., etc., etc.
...
...
gcc -D__KERNEL__ -I/usr/src/linux-2.2.5/include -E -traditional
-DSVGA_MODE=NORMAL_VGA bootsect.S -o bootsect.s
as86 -0 -a -o bootsect.o bootsect.s
make[1]: as86: Command not found
make[1]: *** [bootsect.o] Error 127
make[1]: Leaving directory /usr/src/linux-2.2.5/arch/i386/boot
make: *** [zImage] Error 2
Thanks,
Bruce
------------------------------
From: "J�rgen Exner" <[EMAIL PROTECTED]>
Subject: Re: Symbolic Link
Date: Fri, 14 May 1999 13:19:41 -0700
Reply-To: "J�rgen Exner" <[EMAIL PROTECTED]>
David Noseworthy <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> I have created an anonymous ftp server on my link box (Box A). The home
> directory for anonymous ftp is /home/ftp.
>
> Within this directory I created a symbolic link to the /usr directory.
> After creating the link I tested it, and it seemed to be ok.
>
> When I ftp to Box A and log in as anonymous, I get an error to the
> extent "No such file or directory." An ls -l shows the link is there,
> but I just can't change to that directory.
Of course you can't.
For security reasons anon FTP does a "chroot /home/ftp", i.e. /home/ftp
becomes the root directory while running anon ftp.
That means your symbolic link to /usr actually points to what would be
/home/ftp/usr in normal life.
jue
--
J�rgen Exner
------------------------------
From: [EMAIL PROTECTED] (Andre Kostur)
Subject: Re: 'Find'. what a strange command
Date: Fri, 14 May 1999 21:25:20 GMT
In article <[EMAIL PROTECTED]>, Mark Forsyth <[EMAIL PROTECTED]> wrote:
>Or even find / | grep -i netscape
>
>
>Mark F...
>
>Brad wrote:
>>
>> I was trying to find where netscape is installed on my newly installed
>> Linux setup. I figured there must be a 'find' command and I wasn't
>> dissapointed. I typed 'man find' and had a read.
>> There was the traditional user friendly syntax guide.......
>>
>> find [path....] [expression]
>>
>> so I typed
>>
>> find / netscape
>>
>> As the man pages poorly written man pages have no examples, I figured
>> that this would search from the root and show me all the occurences of
>> 'netscape' on my machine.
>> Wrong. This, for a reason I hope someone can explain, listed, I think,
>> every file on my machine.
>>
>> I got around the problem by typing 'find / netscape >> find.txt' and
>> then using the wonderful grep command to display all the lines with
>> 'netscape' in them. Exactly what 'find' should have done in the first
>> place.
find is very powerful, and it doesn't know what it's supposed to match
'netscape' to. The command I use is:
find / -name netscape -print | more
------------------------------
From: "G. Sumner Hayes" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps,comp.unix.bsd.freebsd.misc
Subject: Re: Pro-Unix vs anti-WinTel
Date: Fri, 14 May 1999 17:07:08 -0400
Doug DeJulio wrote:
> If the answer to #2 is "no", then you *know* every single package
> installed via "Ports" comes from the central repository. You do not
> know this with RPM, and you do not know this with Debian.
rpm -K whatever.rpm
will verify that the PGP signature on whatever.rpm is correct and
will tell you who the signer is. In particular, you can check to make
sure that it's from Red Hat.
Debian offers similar facilities.
So you can install non-centralized packages if you want to, but you
don't have to. Flexibility is a feature, especially when you can
get the more limited (and more secure) behavior easily.
Ports is cool, but not for this reason. It saves a ton of time and
effort when tracking source changes on a regular basis.
Does Ports offer any sort of cryptographic security? Either signed
sources or authenticated connections when downloading? If not, it's
far less secure than the current package solutions.
--Sumner
------------------------------
From: "Dave Williams @ Home" <[EMAIL PROTECTED]>
Subject: Netware RPL Boot
Date: Fri, 14 May 1999 21:29:19 +0100
I have some PC motherboards with network boot proms using RPL (Novell
Netware) protocol. Anyone know if it is possible to make a Linux box act
as a server for these
--
Dave Williams @ Home
------------------------------
From: Gavin McCord <[EMAIL PROTECTED]>
Subject: Shutdown (and X) errors.
Date: Fri, 14 May 1999 21:23:15 GMT
I get the following messages every time I shutdown, sometimes before the
"System halted" message, sometimes after. Occasionally, it hangs without doing
a proper shutdown, so I have to do a hard reboot and then fsck.
Unable to handle kernel NULL pointer dereference at virtual address c0000000
current->tss.cr3 = 00101000, %cr3 = 00101000
*pde = 00102067
*pte = 00000000
Oops = 0000
CPU: 0
EIP: 0010:[<02884957>]
EFLAGS: 00010246
eax: 0288495f ebx: 01531818 ecx: 00000000 edx: 001b1444
esi: 02884944 edi: 00000034 ebp: 001afa2c esp: 001afa10
ds: 0018 es: 0018 fs: 002b gs: 0018 ss:0018
Process swapper (pid: 0, process nr:0, stackpage = 001adaf8)
Stack {three rows of eight eight-digit numbers}
Call Trace {nine eight-digit numbers enclosed in [<...>] bracket
sequence}
Code: ff 41 00 0f 8e e1 5b 91 fd c3 83 ec 38 55 57 56 53 8b 5c 24
Aiee, killing interrupt handler
kfree of non-kmalloced memory:001afb40, next = 00000000, order=0
x 3
idle task may not sleep
x 5
System halted
My system includes:
Cyrix P166 on VX motherboard
Matrox Millenium
Awe 64 Value
Netvin NV5000C - NE2000 clone
Hauppauge WinTV
Slackware 3.6
I've a feeling these errors are related to my sound or video hardware as I also
have crashes with X and SVGAlib, though I still got the shutdown problem with
my S3 Virge DX videocard.
Any ideas, folks?
--
"I'm Keyser Soze. No, I'm Keyser Soze. I'm Keyser Soze and so's
my wife..."
-Monty Python plays The Usual Suspects
------------------------------
From: Johan Kullstam <[EMAIL PROTECTED]>
Subject: Re: Red Hat info needed.
Date: 14 May 1999 15:19:40 -0400
"Peter F. DeMos" <[EMAIL PROTECTED]> writes:
> Randy wrote, after Carlos wrote:
>
> >> I am a fairly new linux user, I am looking for soem advice
> >> as far as resources , there's so much, I want some guidance as into
> >> what to read and what is garbage,, I have mastering linux, linux for
> >> dummies,
> > > and special edition using linux, But still There might be something
> > > better,, I got thru the installation part pretty easy,, I only had
> > > trouble with my sound card which is not supported that's all,,,
>
> > Try RedHat Secrets 5.2 (if you're using RedHat). It's a pretty good
> > book and can be found at most big stores.
>
> I just installed RH5.2 on a box, with no apparent probs. Didn't do any
> printer or modem setups at the time, cuz I didn't have any ready to go.
> Did a clean install, workstation style, completely eliminating anything
> previously on the box. I now have Xwindows coming up when booting.
> Four questions here: a) Question 1 is how do I get the CDrom to be
> recognized automatically, and be able to navigate the CD that is in
> there, installing something if I so feel. So far, the only thing I
> can see is that I have to mount it manually, every time I boot. Now,
> admittedly, it's been at least 5 years since I worked on a -Nix
> machine, so maybe that's just the way it is, but 5 years is 15 or
> more generations in pooter life. I would think I could just have the
> CDRom mount automatically, just like I do the Xwindows. Is that
> possible, or am I dreaming? And is there an "Explorer" tool I can
> use to visually inspect what's on the drive?
maybe midnight commander can read the device directly?
allowing any user to mount it will go a long way to making the cdrom
less painful.
you can edit your /etc/fstab and put the word `user' in the options to
allow anyone to mount/umount it. (see man page for fstab. i may have
the name of the option wrong.)
> b) I now wish to install a modem, I have a 56K PCI Modem Blaster. Is
> there anything I need to know to get that recognized as well.
pci == windows only modem
these modems will not work in linux due to (among other things)
manufacturer refusal to release any specs or linux drivers. sorry.
get an external or isa internal. if internal try to get one with
jumpers to set irq and ioport since pnp is a pain in the ass. avoid
winmodems/host controlled modems and inquire about return policy in
case you should despite everything get one.
> c) RH appears to automatically come up in the bash shell, and I am used
> to a cshell. Is there a way to get the flippin' thing to automatically
> boot to a cshell instead?
make sure you have csh and/or tcsh installed first. become root.
edit /etc/passwd. change /bin/bash to /bin/tcsh. or if you're really
a glutton for punishment /bin/csh. the command chsh may work for you
too.
> and... d) Is there a faq for this group? If so, how/where do I get it?
> Does it get posted monthly? I sense that I will have several questions
> over the next few weeks, and hope to not have to inundate you Randy (or
> other RH gurus), or the group, with said questions. Especially if they
> are covered in the faq.
there should be a lot on your harddrive already. go to
/usr/doc/HOWTO. (install the howto package if you haven't already.)
also <URL:http://www.linuxhq.com> is a great source. look at the
links for linux documentation near the bottom of the page and keep
browsing for the howtos and such.
--
johan kullstam
------------------------------
From: [EMAIL PROTECTED] (NF Stevens)
Subject: Re: A bash question
Date: Fri, 14 May 1999 21:32:38 GMT
Do-Hoon Kwon <[EMAIL PROTECTED]> wrote:
>Hello,
> Does anybody know how to check the status of the statement 1 in
>
> statement 1 | statement 2
>
>As I understand, the variable ? holds the exit status of the last
>command. However, I want the exit status of statement 1 while not
>breaking the pipe.
(statement 1; echo $? >result) | statement 2
The bit is brackets gets executed in a sub shell but
that probably wont be a problem.
Norman
------------------------------
From: Mark Hahn <[EMAIL PROTECTED]>
Subject: Re: Food for thought
Date: 14 May 1999 17:46:16 GMT
> why would Linux programs waste THEIR time with garbage, just because
> you bought a cheap modem that really suck (belive me winmodems suck,
> they are just an attempt of some companies to make more money by
> reducing costs), do yourself a favor an get a real modem, even windows
> beneficts with it
this is a stupid, puerile attitude. winmodems are, in principle,
a reasonable cost/benefit tradeoff. the problem is not that they exist,
the problem is *solely* that vendors are not willing to simply tell
their customers what they've bought. this kind of thing is explicitly
ILLEGAL in the auto business, by the way.
again: winmodems are simply not there to Linux until vendors make
programming documentation available. note that this is _zero_ cost
to the vendors, since they already have internal docs, and no support
is required (or wanted).
>>An incredible number of people use these particular modems, why would they
>>waste their time with linux if some of their hardware won't work? I wish I
stop whining.
or rather: complain to your modem vendor, since they're the problem.
------------------------------
From: Frank Sweetser <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.advocacy,comp.os.linux.advocacy,gnu.misc.discuss
Subject: Re: GNU reeks of Communism
Date: 02 May 1999 17:43:23 -0400
Chris Welch <[EMAIL PROTECTED]> writes:
> Not to mention ESR _threatened_ Bruce Perens not to long ago. He was
> serious, and I've heard he's a gun nut.
read it on slashdot. ESR quickly said that he had *no* intenvions for
physical violence whatsoever.
--
Frank Sweetser rasmusin at wpi.edu fsweetser at blee.net | PGP key available
paramount.ind.wpi.edu RedHat 5.2 kernel 2.2.5 i586 | at public servers
Any false value is gonna be fairly boring in Perl, mathematicians
notwithstanding.
-- Larry Wall in <[EMAIL PROTECTED]>
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: SuSE 6.1 for cheap...
Date: Fri, 14 May 1999 21:34:08 GMT
In article <[EMAIL PROTECTED]>,
[EMAIL PROTECTED] wrote:
> I was pleasantly surprised to find that www.chumbo.com sells SuSE 6.1
> for around $23 plus another $5 for shipping.
>
> I really like Suse and KDE.
It is a well done product, I like it too.
--
Louis-ljl-{ Louis J. LaBash, Jr. }
--== Sent via Deja.com http://www.deja.com/ ==--
---Share what you know. Learn what you don't.---
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To: comp.unix.bsd.freebsd.misc
Subject: Re: Proper use of /usr/local (Re: The Best Linux distribution?)
Date: Fri, 14 May 1999 21:32:51 GMT
In article <7hhm6i$38i$[EMAIL PROTECTED]>,
[EMAIL PROTECTED] (Leslie Mikesell) wrote:
> Hindsight is easy, but that is never the way it works. What happens
> in practice is that you install the distribution version, find
> out that it is broken in some way or that some new and needed
> feature has been added in the current version, or find that
> you need some local custom patch right away. Then you grab the
> original source, do anything needed to fix it, and let it
> run from /usr/local/bin (and if that breaks anything, throw
> a symlink in place of the original once you are sure this is an
> improvement). Now when the next release of the distribution
> comes around you have to look at these individually to see if
> the update takes care of the same problem. This really isn't
> much trouble for the dozen or so programs that might have needed
> a quick-fix along the way - unless something else has cluttered
> /usr/local with things that aren't really yours to maintain.
Okay, let me see if I have this straight:
- You install the distribution version into /usr
- You test the distribution version
- You fix whatever problems you find and install the final version into
/usr/local
That means you're using /usr as a staging/testing area for /usr/local.
Now that you've explained your reasons clearly enough, however, I can
finally see the wisdom of what you're doing, if not the method.
Solaris uses the /opt directory for essentially the same thing you're
trying to use /usr for. At least, that's the way it seems to me [I've
never been quite clear on what that damn /opt is all about]. If that's
true, then the whole thing makes more sense to me now.
So, all of this conflict could be resolved if Linux distributors would
adopt the following convention:
/usr - system files
/usr/local - user customized files
/opt - distribution files
That way, your distribution could install optional packages into /opt,
without risking overwrite of your stuff in /usr/local *and* without
intermingling system and non-system stuff. I'd think that would make
everyone happy.
--
-Bill Clark
Systems Architect
ISP Channel
http://locale.ispchannel.com/
--== Sent via Deja.com http://www.deja.com/ ==--
---Share what you know. Learn what you don't.---
------------------------------
From: [EMAIL PROTECTED]
Crossposted-To: linux.debian.user,linux.redhat.misc,comp.os.linux.hardware
Subject: Re: What happened to fdformat
Date: Fri, 14 May 1999 21:30:28 GMT
In article <7hho5b$hn7$[EMAIL PROTECTED]>,
[EMAIL PROTECTED] wrote:
> I know this sounds stupid, but I used to think that you can
low-level
> format a floppy in Linux using "fdformat". Well, on my Debian system
> this is what I get:
> ----------------------------------
> histria ~ # fdformat
> bash: fdformat: command not found
> histria ~ # man fdformat
> No manual entry for fdformat
> -----------------------------------
> (as root). How can I format a floppy ?
Well, I just used it yesterday on a newly installed S.u.S.E. Linux
6.1 system. Isn't it part of the 'mtools' package. Maybe you have
to be 'root' to use it.
Hope these hint are useful.
--
Louis-ljl-{ Louis J. LaBash, Jr. }
--== Sent via Deja.com http://www.deja.com/ ==--
---Share what you know. Learn what you don't.---
------------------------------
From: Richard Caley <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps,comp.unix.bsd.freebsd.misc
Subject: Re: Pro-Unix vs anti-WinTel
Date: 14 May 1999 19:44:27 +0100
In article <[EMAIL PROTECTED]>, Steve Lamb (sl) writes:
sl> What happens when an author releases a new version of his
sl> software?
Same as happens with a packages tree. If you go get it, anything which
uses it is subject to failure without notice. Whether they fail
depends on the effort gone into backwards compatability by the
developer, not on whether you get the result as binary or source.
--
Mail me as rjc not [EMAIL PROTECTED] _O_
|<
------------------------------
From: [EMAIL PROTECTED] (Never spam a hacker)
Subject: Linux Software Archive + more
Date: Fri, 14 May 1999 19:40:44 GMT
Hi
http://members.xoom.com/myvillage/linux
includes a new Linux Help Forum for beginners and advanced users. This
forum is still new, so start posting. But the main feature of this
site is its software archive for Linux. It will soon be having a
search engine but it currently neatly categorized.
Check out this site now.
------------------------------
From: [EMAIL PROTECTED] (NF Stevens)
Subject: Re: page faults
Date: Fri, 14 May 1999 21:32:39 GMT
[EMAIL PROTECTED] (Bob Tennent) wrote:
>On Thu, 13 May 1999 13:08:40 -0400, jason wrote:
> >
> >Bad memory is a very likely possibility. I bought new RAM last year that was
> >bad upon arrival. I used this excellent program to test it and verify it
> >was bad:
> >
> >http://metalab.unc.edu/pub/Linux/system/hardware/memtest86-1.4a.tar.gz
> >
>OK, I've tested my RAM using memtest and it detected no errors.
>So what *else* might be causing apparently random page faults?
I had a couple of those page fault things immediately after
upgrading to kernel 2.2.5. I cleaned all the dust out of the
fan on my processor. It hasn't occurred again yet. (Touch wood).
Norman
------------------------------
From: [EMAIL PROTECTED] (Doug DeJulio)
Crossposted-To: comp.os.linux.development.apps,comp.unix.bsd.freebsd.misc
Subject: Re: Pro-Unix vs anti-WinTel
Date: 14 May 1999 16:51:37 -0400
In article <[EMAIL PROTECTED]>,
Paul D. Smith <[EMAIL PROTECTED]> wrote:
>%% [EMAIL PROTECTED] writes:
>
> z> Ports is probably the only package system that *does* have a
> z> central repository.
>
>What exactly do you mean by that? Debian certainly does. Even RedHat
>does, basically (although they have a lot of unofficial stuff on their
>site, too, and it's less well-organized). I'm not following your
>terminology.
I think I see what he means, but I'm not certain, so I'll ask a pair
of questions (one of which I'll answer).
1) Is it possible to make a Debian package available without doing so
through the central Debian repository? The answer to this one is
"yes".
2) Is it possible to make a "ports package" available without doing so
through the central *BSD "Ports" repository? I really don't know
the answer to this.
If the answer to #2 is "no", then you *know* every single package
installed via "Ports" comes from the central repository. You do not
know this with RPM, and you do not know this with Debian.
--
Doug DeJulio | mailto:[EMAIL PROTECTED]
HKS, Incorporated | http://www.hks.net/~ddj/
------------------------------
From: "G. Sumner Hayes" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps,comp.unix.bsd.freebsd.misc
Subject: Re: Pro-Unix vs anti-WinTel
Date: Fri, 14 May 1999 15:54:45 -0400
Richard Caley wrote:
> What do you want an example of? The internet ois full of binaries of
> who nows what origin.
>
> I was just saying I don't see why I should trust them over verified
> source from the real home of whateveritis.
Trust and source vs. binary are orthogonal issues. FWIW, both Debian
and Red Hat packages carry PGP signatures to protect against tampering,
as well as MD5 checksums to protect against random bitrot. If you're
being attacked by someone who can spoof RSA signatures, you should
probably throw in the towel on the security front.
I'd love to see a cvsup/ports system deployed alongside rpm/dpkg; I
always get the source RPMs and build myself. There's no real reason
to do this in the common case, though; many official binary packages
go through rather extensive regression testing and are more likely
to perform correctly than something you build yourself locally
_assuming_ that you are running the same setup as the packager. So
for people running Debian slink, grabbing a binary from slink is
probably more likely to work than is grabbing the source and rebuilding.
Similarly for Red Hat 6.0.
For me, ports would be a great thing. Unless you audit the source
yourself, though, it's not a security issue. Even then, you normally
get your first compiler from someone; you have to trust them not to
launch a Thompson-style attack anyway, so trusting them for other
binaries is not a huge leap of faith.
Ports makes it a lot easier to upgrade the entire system from source.
It can also reduce bandwidth requirements. But it increases demands on
what you install on the local system and on use of local processor time.
The Red Hat situation is far from ideal if you want to rebuild
everything; it's slightly better than getting .tar.gz files, but not
anywhere near as nice as ports. Ports is far from ideal for installing
binaries.
--Sumner
------------------------------
** 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
******************************
