FreeBSD v.s. Linux
Clyde Wary
This may be something that's been discussed many times in the past, but
I just recently found this newsgroup, so here goes. I have a fair
amount of experience using and programming with Linux, and was going to
install it on my PC at home...while trying to decide which distribution
of Linux to get, I found out about FreeBSD. How do the two OS's
compare? I've heard that FreeBSD is compatible with several commercial
versions of Unix, but Linux may not be. Is this true? How easy is
FreeBSD to install, compared with, say, Red Hat Linux, or Slackware
Linux? I am not a skilled hacker, but I am alot more competent than the
average PC user in these areas. Thanks.
John R. Campbell
FreeBSD uses Berserkeleyish syscall/interrupt behavior.
I don't know what commercial apps you can run on it vs Linux;
I personally prefer Linux but that's only because I have more
experience dealing with SysVish flavoring. If I had more time
dealing w/ BSD flavorings, I would've ended up preferring
FreeBSD. I've already done device drivers for FreeBSD, so
I have experience in both universes.
If you flip a coin, you can't lose. Both systems are VERY
functional and the distributions will carry enough tools
that you will be quite happy.
Of course, if your Unix experiences are primarily w/ Linux
(and SysV) then Linux is the "way to go". If you're more
adventurous and want to learn the Berserkeley way of doing
things, FreeBSD is fine.
Sorry, the two systems are (roughly) equivalent in their
functionality. No matter which one you choose you will
be happy.
As for distributions- there are more Linux distributions.
I've tended towards Slackware in the past, but RedHat's
ability to upgrade the system piecemeal (rather than slick
and reload) is attractive to me- for Linux.
I DO NOT have enough experience w/ FreeBSD to comment. I
don't know enough...
Good luck!
--
John R. Campbell, Speaker to Machines so...@jtan.com
"As a SysAdmin, yes, I CAN read your e-mail, but I DON'T get that bored!"-me
Disclaimer: I'm just a consultant at the bottom of the food chain, so,
if you're thinking I speak for anyone but myself, you must
have more lawyers than sense.
Jordan K. Hubbard
Clyde Wary wrote:
>
> This may be something that's been discussed many times in the past,
You bet.
> of Linux to get, I found out about FreeBSD. How do the two OS's
> compare? I've heard that FreeBSD is compatible with several commercial
> versions of Unix, but Linux may not be. Is this true? How easy is
FreeBSD will run BSD/OS, SCO, SVR4 and Linux binaries. Linux will run
SCO and SVR4 binaries, as far as I know, so you'll miss the BSD/OS
emulation there (you can, of course, run Linux binaries just fine under
Linux :-).
> FreeBSD to install, compared with, say, Red Hat Linux, or Slackware
> Linux? I am not a skilled hacker, but I am alot more competent than
I'd say it's comparable.
--
- Jordan Hubbard
FreeBSD core team / Walnut Creek CDROM.
Ferdinand Goldmann
Clyde Wary <cw...@mailhost.nmt.edu> wrote:
: This may be something that's been discussed many times in the past, but
: I just recently found this newsgroup, so here goes. I have a fair
: amount of experience using and programming with Linux, and was going to
: install it on my PC at home...while trying to decide which distribution
: of Linux to get, I found out about FreeBSD. How do the two OS's
You said it ... one of the chief advantages of FreeBSD over Linux is IMHO the
availability of one central maintained clean nice distribution. Not 10? 15?
as it is in the Linux world. Well... it can be a pain if you have to work with/
manage machines based on different Linux distributions. Each of these products
keeps its config files somewhere else (Although most of them are rather good)
I also think that a FreeBSD machine is easier to manage than a Linux machine.
: versions of Unix, but Linux may not be. Is this true? How easy is
: FreeBSD to install, compared with, say, Red Hat Linux, or Slackware
: Linux? I am not a skilled hacker, but I am alot more competent than the
: average PC user in these areas. Thanks.
Rather easy, much like a standard Linux distribution. Well, it depends on what
you want to do with your computer I guess. If you need support for commercial
software like databases or office software, you are better off using Linux.
If you plan to set up a server, you might go with FreeBSD instead.
I am currently switching from Linux to FreeBSD at home, just because the
distribution feels better and out of interest 8^))
/ferdl
--
<< In Real Life: Goldmann Ferdinand >>
<< IRC : DrAkHaI 8^) >>
<< eMail : fe...@wildsau.idv.uni-linz.ac.at >>
<< ... Let him who hath understanding reckon the number of the GateZ >>
<< for it is a human number, its number is six hundred and sixty six >>
Shawn Ramsey
> You said it ... one of the chief advantages of FreeBSD over Linux is IMHO the
> availability of one central maintained clean nice distribution. Not 10? 15?
> as it is in the Linux world. Well... it can be a pain if you have to work with/
> manage machines based on different Linux distributions. Each of these products
> keeps its config files somewhere else (Although most of them are rather good)
The centralization of FreeBSD is the single biggest advantage that
FreeBSD has over Linux IMO. I recently installed Redhat Linux 4.2 on my
personal workstation, for the hell of it, and to learn it. I miss
FreeBSD. :) The FreeBSD ports/packages collection is so much nicer than
Redhat RPM's, because of the centralized list, and the fact the ports
will automatically fetch required ports that are not already installed.
AFIAK, RHL doesnt do this.
There is no -stable period of Linux. :)
> I also think that a FreeBSD machine is easier to manage than a Linux machine.
Most definetly. This again due to the centralized distribution...
basically all FreeBSD installions are the same, with minor diferences
depending on version.
> : versions of Unix, but Linux may not be. Is this true? How easy is
> : FreeBSD to install, compared with, say, Red Hat Linux, or Slackware
> : Linux? I am not a skilled hacker, but I am alot more competent than the
> : average PC user in these areas. Thanks.
>
> Rather easy, much like a standard Linux distribution. Well, it depends on what
> you want to do with your computer I guess. If you need support for commercial
> software like databases or office software, you are better off using Linux.
> If you plan to set up a server, you might go with FreeBSD instead.
> I am currently switching from Linux to FreeBSD at home, just because the
> distribution feels better and out of interest 8^))
FreeBSD is a breeze to install. RHL wasnt difficult, but I did an FTP
installation. I found it incredibly annoying they didnt bother to
include some default URL's to fetch the distribution from.
stephen farrell
Shawn Ramsey <sh...@cpl.net> writes:
> The centralization of FreeBSD is the single biggest advantage that
> FreeBSD has over Linux IMO. I recently installed Redhat Linux 4.2 on my
I agree with this assessment. The difference is in the details, and
in the details, while freebsd does pretty much the same thing linux
does all-around, it is a lot more thought-out... IMHO.
Also, on a more objective note, another huge benefit I see to freebsd
over the linuxes is the "core" system... This stuff is part of every
freebsd system and is comprehensive. It has what you expect of a unix
system. No playing games trying to find which rpm has the "host"
binary, or which .deb has the "ktrace" (or ptrace or truss or strace)
binary. If it should be there, it is. I like that.
Oh yeah, another substantive difference with freebsd is the stability
of the system libraries and such (libc, ld.so, and so on...) By
"stability" I mean from a programming point of view--they don't
change. With linux, one finds that one needs to, say, upgrade ld.so
to run the latest whatever, but in doing so, one breaks a bunch of
other binaries which depended upon bugs in a previous ld.so.
Similarly, there were tons of problems with libc changing... Netscape
wasa problem for ages (needed to use LD_PRELOAD and a special copy of
the old libc for netscape alone), and the current jdk ships with it's
own copy of libc to avoid compatibility problems. Not only does this
suck in the obvious ways, but it sucks additionally b/c changing these
libraries is one easy way to hose your system. Try to run "mv" on
linux without ld.so =). And now linux is moving AGAIN to another libc
system (glibc), which will perpetuate these problems for a long time
(albeit that glibc is supposed to ultimately fix them... but that
could be years from now). I"ve never encountered any of this BS with
freebsd. I like that too. A lot.
--sf
Dana Booth
Clyde Wary wrote:
>
> This may be something that's been discussed many times in the past, but
> I just recently found this newsgroup, so here goes. I have a fair
> amount of experience using and programming with Linux, and was going to
> install it on my PC at home...while trying to decide which distribution
> FreeBSD to install, compared with, say, Red Hat Linux, or Slackware
> Linux? I am not a skilled hacker, but I am alot more competent than the
> average PC user in these areas. Thanks.
Clyde, both have pros and cons, and a lot of loyalists on both sides
will give you varying opinions. I use both, and can tell you that you'll
be very happy with the first one you install, whichever it is.
Fortunately, there's many people who use both, and the 'holy wars'
aren't as prevalent here as, say, the Windows95 - OS/2 - Mac wars. Hell,
we're all addicted to UNIX-alikes, right? :)
--
------------------------
Dana Booth <da...@oz.net>
------------------------
Albert D. Cahalan
stephen farrell <sfarrel...@farrell.org> writes:
>
> Also, on a more objective note, another huge benefit I see to freebsd
> over the linuxes is the "core" system... This stuff is part of every
> freebsd system and is comprehensive. It has what you expect of a unix
> system. No playing games trying to find which rpm has the "host"
> binary, or which .deb has the "ktrace" (or ptrace or truss or strace)
> binary. If it should be there, it is. I like that.
You'd hate that if your disk was small.
> Oh yeah, another substantive difference with freebsd is the stability
> of the system libraries and such (libc, ld.so, and so on...) By
> "stability" I mean from a programming point of view--they don't
> change. With linux, one finds that one needs to, say, upgrade ld.so
> to run the latest whatever, but in doing so, one breaks a bunch of
> other binaries which depended upon bugs in a previous ld.so.
> Similarly, there were tons of problems with libc changing... Netscape
> wasa problem for ages (needed to use LD_PRELOAD and a special copy of
> the old libc for netscape alone), and the current jdk ships with it's
Netscape is full of bugs. Many Linux systems use a high-performance
memory allocator that won't tolerate code that tries to free data
on the stack, code that frees data twice, and code that goes off the
end of an array. It is not legal C, so it crashes.
This is very good for developers. If the code runs on Linux,
it means that the memory operations (malloc, free...) are being
used correctly. You can thank the Linux libc for making free
software more reliable in general.
Guy Harris
>Not only does this
>suck in the obvious ways, but it sucks additionally b/c changing these
>libraries is one easy way to hose your system. Try to run "mv" on
>linux without ld.so =).
I can do it on SunOS 4.x:
nova% uname -sr
SunOS 4.1.3
nova% ldd /usr/bin/mv
/usr/bin/mv: statically linked
and on SunOS 5.x:
tooting$ uname -sr
SunOS 5.5.1
tooting$ ldd /usr/sbin/static/mv
ldd: /usr/sbin/static/mv: file is not a dynamic executable or shared
object
tooting$ file /usr/sbin/static/mv
/usr/sbin/static/mv: ELF 32-bit MSB executable, SPARC, version 1,
statically linked, stripped
Perhaps the Linux folks should learn from Sun here - there are a few
programs in 4.x that are statically linked so that you *can* recover
from at least some screwups, and "/usr/sbin/static" on 5.x is there for
the same reason (5.x has "cp", "ln", "mv", "rcp", and "tar" - "cp" and
"mv" for the obvious reasons, "ln" to put in any necessary links, "rcp"
to suck a copy from another machine, and "tar" to extract from a "help
help help I screwed up" archive).
Alas, I don't have a FreeBSD machine handy to see what they do in the
way of keeping some emergency-use statically-linked programs around, so
I don't know if they've picked up that idea from Sun or not. (They
should do so, if they haven't done so already.) I'm assuming here,
perhaps incorrectly, that programs are dynamically-linked by default.
--
Reply, or follow up, but don't do both, please.
postmaster@localhost
postmaster@[127.0.0.1]
Giao Nguyen
g...@netapp.com (Guy Harris) writes:
> >Not only does this
> >suck in the obvious ways, but it sucks additionally b/c changing these
> >libraries is one easy way to hose your system. Try to run "mv" on
> >linux without ld.so =).
[snip]
[comments about recoverability and static linking]
> Alas, I don't have a FreeBSD machine handy to see what they do in the
> way of keeping some emergency-use statically-linked programs around, so
> I don't know if they've picked up that idea from Sun or not. (They
> should do so, if they haven't done so already.) I'm assuming here,
> perhaps incorrectly, that programs are dynamically-linked by default.
FreeBSD uses static linking for everything in /bin and
/sbin. Including shells. It's smart like that :) And I agree with you,
having bare basic utilities statically linked is a smart thing.
Timothy J. Lee
g...@netapp.com (Guy Harris) writes:
| tooting$ uname -sr
| SunOS 5.5.1
| tooting$ ldd /usr/sbin/static/mv
| ldd: /usr/sbin/static/mv: file is not a dynamic executable or shared
| object
|
|"mv" for the obvious reasons, "ln" to put in any necessary links, "rcp"
|to suck a copy from another machine, and "tar" to extract from a "help
|help help I screwed up" archive).
Statically linked sh and mount are also useful for crash recovery. On
SunOS 5, root's default shell is /sbin/sh, which is statically linked.
|Alas, I don't have a FreeBSD machine handy to see what they do in the
|way of keeping some emergency-use statically-linked programs around, so
|I don't know if they've picked up that idea from Sun or not.
ldd on FreeBSD 2.2.1 reports that the programs in /bin and /sbin are
"not dynamic executable"s. The programs in /usr/bin are linked with
shared libraries, according to ldd. But /bin and /sbin contain the
essentials for crash recovery, such as sh, mount, fsck, cp, mv, etc.
--
------------------------------------------------------------------------
Timothy J. Lee timlee@
Unsolicited bulk or commercial email is not welcome. netcom.com
No warranty of any kind is provided with this message.
Albert D. Cahalan
g...@netapp.com (Guy Harris) writes:
>> Not only does this suck in the obvious ways, but it sucks
>> additionally b/c changing these libraries is one easy way
>> to hose your system. Try to run "mv" on linux without ld.so =).
Try _not_ removing ld.so.
> Perhaps the Linux folks should learn from Sun here - there are a few
> programs in 4.x that are statically linked so that you *can* recover
> from at least some screwups, and "/usr/sbin/static" on 5.x is there for
The SunOS 5.x way (/usr/sbin/static) makes some sense, though if
you removed /lib then it may be best to hit the power switch before
the disks get synced. I've saved other data on a Linux system
that way. BTW, the disks were in perfect condition after that.
The FreeBSD way (/bin is static) is really dumb. Shared libraries
exist to save memory, and you are NOT USING THEM for commonly
used software. Note that "/bin" comes before "/lib", so it is
likely that your static binaries in /bin will get removed by
an "rm -rf *" in / before ld.so is removed.
Actually, /sbin/static is best of all.
Steinar Haug
[Albert D. Cahalan]
| The FreeBSD way (/bin is static) is really dumb. Shared libraries
| exist to save memory, and you are NOT USING THEM for commonly
| used software.
I think you'll find many people disagree here. Shared libraries do *not*
necessarily save memory, and AFAIK this was not the primary reason for
introducing them in FreeBSD. They usually save disk space, though.
Also, the binaries in /bin are usually rather small programs, and rather
useful to have in emergencies. I'm very happy with /bin the way it is in
FreeBSD.
Steinar Haug, Nethelp consulting, sth...@nethelp.no
Steinar Haug
[Timothy J. Lee]
| The programs in /usr/bin are linked with
| shared libraries, according to ldd. But /bin and /sbin contain the
| essentials for crash recovery, such as sh, mount, fsck, cp, mv, etc.
At least on my system, chflags, gunzip/gzcat/gzip/zcat, ld and tar (all
in /usr/bin) are not dynamically linked.
John S. Dyson
In article <vc73ekl...@mercury.cs.uml.edu>,
acah...@mercury.cs.uml.edu (Albert D. Cahalan) writes:
> g...@netapp.com (Guy Harris) writes:
>
>>> Not only does this suck in the obvious ways, but it sucks
>>> additionally b/c changing these libraries is one easy way
>>> to hose your system. Try to run "mv" on linux without ld.so =).
>
> Try _not_ removing ld.so.
>
> exist to save memory, and you are NOT USING THEM for commonly
> likely that your static binaries in /bin will get removed by
> an "rm -rf *" in / before ld.so is removed.
>
> Actually, /sbin/static is best of all.
Please do some research. Shared libs do not necessarily save memory,
but actually do more to share disk space. (Building a shell such as
"bash" shared can be significantly counterproductive.) Shared libs also
exact a performance penalty, dependent on CPU architecture and MMU design
and how the VM system and compiler deals with those items.
The formula as to what is best to build shared or not isn't trivial. It
would be an interesting paper to read or write. Shared libs have been
problematical for as long as I can remember. Any single solution to the
problem of saving disk, saving memory, maximizing performance, and ease
of creation of shared libs (so far) involves tradeoffs of these goals.
Jordan K. Hubbard
John S. Dyson wrote:
> Please do some research. Shared libs do not necessarily save memory,
jkh: "Whups! Looks like John's speed-reading the headers again. Lesse,
what's faster, typing or talking? Yeah, I think I'll give him a ring.."
Phone: "*breeeeeeet*... *breeeeeeeet*... *click!*"
John: "Hello?"
AT&T: "[line noise]"
jkh: "Heya, John! It's me. Yeah, about that posting you just made to
c.u.b.f.m to the "/bin static is STUPID" guy... Yeah, that one. So, I
was just reading it and wondering here. Did you notice the name on the
header? No? Heh. That was the great Albert D. Cahalan you were talking
to there! And you asked him to do some research! Ha ha! I was fit to
bust laughing. Huh? C'mon, you remember him - he's that occasional
poster who doesn't *do* research or bother with facts or exercise any
discernable form of neural activity whatsoever in his postings. He's
that linux *flamer* guy, remember?"
John: "Ohhhh..."
jkh: "Yeah, that guy. He of ``my TCP stack is better than your TCP stack
if you happen to be measuring it this week and are willing to ignore
last week entirely!'' fame. Hah hah! Right. A real poster child for
the "stop your kid from eating old paint chips!" public awareness
campaign, if you catch my drift."
John: "[mumble mumble, mumble.. mumble?]"
jkh: "Nope. Not even prolonged inbreeding would account for that, I
think. So anyway, don't waste your time on him, OK? If you have to
respond to the guy's trolls at all, do it as a parody or something, but
for god's sake don't take him seriously! Yeesh! Yeah. No problem.
Happy to help."
John: "Why do I have the eeriest feeling that my phone's about to ring?"
Jordan K. Hubbard
Albert D. Cahalan wrote:
>
> You'd hate that if your disk was small.
I'm framing this.
Gary Howland
Giao Nguyen wrote:
> > I don't know if they've picked up that idea from Sun or not. (They
> > should do so, if they haven't done so already.) I'm assuming here,
> > perhaps incorrectly, that programs are dynamically-linked by default.
>
> FreeBSD uses static linking for everything in /bin and
> /sbin. Including shells. It's smart like that :) And I agree with you,
> having bare basic utilities statically linked is a smart thing.
/bin/mv is dynamically linked, but /stand/mv is statically linked. But
what's really nice about the contents of /stand/ is that it is mostly
one program! -:
80 -r-xr-xr-x 58 root bin 802816 Nov 18 1995 init
80 -r-xr-xr-x 58 root bin 802816 Nov 18 1995 kill
80 -r-xr-xr-x 58 root bin 802816 Nov 18 1995 ln
80 -r-xr-xr-x 58 root bin 802816 Nov 18 1995 ls
80 -r-xr-xr-x 58 root bin 802816 Nov 18 1995 mkdir
80 -r-xr-xr-x 58 root bin 802816 Nov 18 1995 mknod
80 -r-xr-xr-x 58 root bin 802816 Nov 18 1995 mount
80 -r-xr-xr-x 58 root bin 802816 Nov 18 1995 mount_cd9660
80 -r-xr-xr-x 58 root bin 802816 Nov 18 1995 mount_msdos
80 -r-xr-xr-x 58 root bin 802816 Nov 18 1995 mount_nfs
80 -r-xr-xr-x 58 root bin 802816 Nov 18 1995 mt
80 -r-xr-xr-x 58 root bin 802816 Nov 18 1995 mv
Notice that they're all inode 80? And there are 58(!) links to that
program. So not only is 'mv' linked to 'cp', it's linked to 'init', to
'kill', to 'mount', etc. So you can recover your system from just one
file (and no libraries)!
Gary
--
pub 1024/C001D00D 1996/01/22 Gary Howland <ga...@hotlava.com>
Key fingerprint = 0C FB 60 61 4D 3B 24 7D 1C 89 1D BE 1F EE 09 06
Niall Smart
Albert D. Cahalan wrote:
> stephen farrell <sfarrel...@farrell.org> writes:
> >
> > Also, on a more objective note, another huge benefit I see to freebsd
> > over the linuxes is the "core" system... This stuff is part of every
> > freebsd system and is comprehensive. It has what you expect of a unix
> > system. No playing games trying to find which rpm has the "host"
> > binary, or which .deb has the "ktrace" (or ptrace or truss or strace)
> > binary. If it should be there, it is. I like that.
>
If you've got a small disk buy a bigger one, bleh.
> Netscape is full of bugs. Many Linux systems use a high-performance
> memory allocator that won't tolerate code that tries to free data
> on the stack, code that frees data twice, and code that goes off the
> end of an array. It is not legal C, so it crashes.
If the Linux libc malloc crashes when the a program does these things
it is by accident not design.
> This is very good for developers. If the code runs on Linux,
> it means that the memory operations (malloc, free...) are being
> used correctly. You can thank the Linux libc for making free
> software more reliable in general.
Relying on the side effects of a memory allocator to judge the
reliability
of your code is a bad idea, if you want to test that your dynamic memory
allocation is correct then use a tool such as PureAtria Purify, if this
not available for your development platform, or you can't afford it,
then
use one of the freely available tools that can help. FreeBSD's libc
malloc has some useful options which can be configured from an
environment
variable, for example.
Niall
stephen farrell
acah...@mercury.cs.uml.edu (Albert D. Cahalan) writes:
> > Oh yeah, another substantive difference with freebsd is the stability
> > of the system libraries and such (libc, ld.so, and so on...) By
> > "stability" I mean from a programming point of view--they don't
> > change. With linux, one finds that one needs to, say, upgrade ld.so
> > to run the latest whatever, but in doing so, one breaks a bunch of
> > other binaries which depended upon bugs in a previous ld.so.
> > Similarly, there were tons of problems with libc changing... Netscape
> > wasa problem for ages (needed to use LD_PRELOAD and a special copy of
> > the old libc for netscape alone), and the current jdk ships with it's
>
> Netscape is full of bugs. Many Linux systems use a high-performance
> memory allocator that won't tolerate code that tries to free data
> on the stack, code that frees data twice, and code that goes off the
> end of an array. It is not legal C, so it crashes.
Unfortunately, I'm well familiar with the reason why Netscape 3 had
problems (d. leay's malloc() replacing gnumalloc() in libc). However,
you miss the point. 1. libc should never have been buggy enough in
the first place to allow netscape (or whomever) to write sloppy code
on it. The fact that it was, and it changed, and that users have to
deal with the consequences is a typical "linuxism". 2. You failed to
address that this problem is *ongoing*--we both well know that the
netscape example is ancient news. But why does the current jdk ship
with it's own copy of libc? Why does it require the latest ld.so that
is not part of any "stable" linux distribution? Why is linux moving a
a whole new libc in the near future?--and what problems is that going
to incur? And by the time the promise of glibc is here and it is
stable, I'm sure there will be an hlibc to take over and cause the
upheaval again. It's just the attitude, and it's not going to change,
I don't think. Would like to be wrong.
> This is very good for developers. If the code runs on Linux,
> it means that the memory operations (malloc, free...) are being
> used correctly. You can thank the Linux libc for making free
> software more reliable in general.
Jordan can frame your other comment; I'll frame this one.
(ps. iirc, only netscape, which is *not* free software, really slammed
into this problem so heinously. most free software is portable and so
such problems are debugged by working on other systems i.e., not
linux).
(pps. I'm a huge linux advocate, really--no sarcasm.)
--sf
stephen farrell
g...@netapp.com (Guy Harris) writes:
> Alas, I don't have a FreeBSD machine handy to see what they do in the
> way of keeping some emergency-use statically-linked programs around, so
[phaedrus]~% ldd `which mv`
ldd: /bin/mv: not a dynamic executable
--sf
stephen farrell
acah...@mercury.cs.uml.edu (Albert D. Cahalan) writes:
> Try _not_ removing ld.so.
Right. This is my point *exactly*. Ironically, I"ve never had to
touch ld.so on freebsd. But on Linux, this is not always an option.
And sure if you're real careful you should never have a problem. But
mistakes happen. And it's b/c this stuff changes so often on Linux
that these issues even come up.
One time I compiled my own libc on linux (gee... why was I doing that?
To build a current libc with the old gnumalloc in it so I could run
netscape too) and I botched... the libc was bogus and guess what
happened when I put it in as the system libc and ran ldconfig?
Right--I was booting off of a floppy.
> > Perhaps the Linux folks should learn from Sun here - there are a few
> > programs in 4.x that are statically linked so that you *can* recover
> > from at least some screwups, and "/usr/sbin/static" on 5.x is there for
>
> The SunOS 5.x way (/usr/sbin/static) makes some sense, though if
> you removed /lib then it may be best to hit the power switch before
> the disks get synced. I've saved other data on a Linux system
> that way. BTW, the disks were in perfect condition after that.
Super.
> The FreeBSD way (/bin is static) is really dumb. Shared libraries
> exist to save memory, and you are NOT USING THEM for commonly
> used software.
Well, it's a trade off. You lose some disk space and some memory, but
you gain speed and reliability. Frankly this is a trade-off I'd
rather make in the direction of freebsd. Plus /bin/mv is 139264,
which doesn't seem grossly huge to me... albeit >4 times the size of
the linux counterpart (29024).
> Note that "/bin" comes before "/lib", so it is
> likely that your static binaries in /bin will get removed by
> an "rm -rf *" in / before ld.so is removed.
Try _not_ doing rm -rf * in /.
> Actually, /sbin/static is best of all.
I guess. But you're wasting even more disk space then, and iirc, you
seem to think that's pretty important... Also I don't see any linux
systems doing this.
--sf
Jeremy Nelson
>[Timothy J. Lee]
>| The programs in /usr/bin are linked with
>| shared libraries, according to ldd. But /bin and /sbin contain the
>| essentials for crash recovery, such as sh, mount, fsck, cp, mv, etc.
>
>At least on my system, chflags, gunzip/gzcat/gzip/zcat, ld and tar (all
>in /usr/bin) are not dynamically linked.
/usr/bin/tar (dynamically linked) wouldnt be really neccesary in crash
recovery since you have /bin/pax (statically linked). You can find a
statically linked gzip in /stand/gzip (you did keep /stand around, didnt
you?), but i keep a statically linked gzip in /bin for my sanity.
I'm not entirely sure i understand why you'd need 'ld' in crash recovery...
What all do most of you put into /bin because at one time your system
crashed and took all of /usr with it, leaving you with tools that you
just cant do without? When i lost my /usr partition long ago, i swore
never to be caught without 'less', 'zsh', 'gzip', and 'vi' again. ;-)
Jeremy
Steven G. Kargl
In article <347AE2...@hotlava.com>,
Gary Howland <ghow...@hotlava.com> writes:
> Giao Nguyen wrote:
>
>> > I don't know if they've picked up that idea from Sun or not. (They
>> > should do so, if they haven't done so already.) I'm assuming here,
>> > perhaps incorrectly, that programs are dynamically-linked by default.
>>
>> FreeBSD uses static linking for everything in /bin and
>> /sbin. Including shells. It's smart like that :) And I agree with you,
>> having bare basic utilities statically linked is a smart thing.
>
> /bin/mv is dynamically linked, but /stand/mv is statically linked. But
> what's really nice about the contents of /stand/ is that it is mostly
> one program! -:
>
/bin/mv is a statically linked binary here.
troutmask:kargl[224] pwd
/bin
troutmask:kargl[225] file mv
mv: FreeBSD/i386 compact demand paged executable
--
Steve
finger ka...@troutmask.apl.washington.edu
http://troutmask.apl.washington.edu/~clesceri/kargl.html
Gary Howland
> acah...@mercury.cs.uml.edu (Albert D. Cahalan) writes:
>
> > The SunOS 5.x way (/usr/sbin/static) makes some sense, though if
> > you removed /lib then it may be best to hit the power switch before
> > the disks get synced. I've saved other data on a Linux system
> > that way. BTW, the disks were in perfect condition after that.
Good point! Perhaps we need a command to delay any syncing to disk for
a few seconds? Jordan, will you add it to the distribution if I write
an "unsync" program?
Seriously, though, the sad fact is that this only worked because linux
machines tend to mount disks asynchronously in order to squeeze out a
little more performance at the expense of reliability.
(I think I'm correct in stating that 'rm' on FreebSD will not return
until the disks have been synced - how else can you maintain a stable
filesystem?)
Ron Echeverri
In article <vc73ekl...@mercury.cs.uml.edu>,
Albert D. Cahalan <acah...@mercury.cs.uml.edu> wrote:
>The FreeBSD way (/bin is static) is really dumb.
"I am a sophomore Computer Science major at the University of
Massachusetts in Lowell."
rone
--
Ron Echeverri <rone...@best.net> | "Who is the boss of you? Me! I am
Systems/Usenet Administration | the boss of you! I am the boss of
Best Internet Communications, Inc. | you! I am the boss of you!"
<URL:http://www.ennui.org/~rone/> | - The Grand Inquisitor
Gary Howland
Steven G. Kargl wrote:
>
> /bin/mv is a statically linked binary here.
> troutmask:kargl[224] pwd
> /bin
> troutmask:kargl[225] file mv
> mv: FreeBSD/i386 compact demand paged executable
>
Yes, you're right. 2.1 FreeBSD doesn't display "compact" :-/
Martin Kammerhofer
Gary Howland (ghow...@hotlava.com) wrote:
: > acah...@mercury.cs.uml.edu (Albert D. Cahalan) writes:
:
: (I think I'm correct in stating that 'rm' on FreebSD will not return
: until the disks have been synced - how else can you maintain a stable
: filesystem?)
You can do the same tradeoff with FreeBSD's native UFS/FFS filesystems.
See the ``async'' option of mount(8).
Generally I recommend against it, except for /tmp and the like.
HTH,
Martin
--
<A HREF="mailto:da...@sbox.tu-graz.ac.at">Martin Kammerhofer</A>.
PGP-key 1024/A96058B1 2B 3D F9 3A F8 B4 18 24 5F 00 2D F6 B9 06 A2 53
for public key finger da...@sbox.tu-graz.ac.at or request it by email
Gary Howland
Martin Kammerhofer wrote:
> You can do the same tradeoff with FreeBSD's native UFS/FFS filesystems.
> See the ``async'' option of mount(8).
> Generally I recommend against it, except for /tmp and the like.
I agree. I also use it for removable media, though, like ZIP drives -
it makes the world of difference, especially if copying lots of small
files. Just don't take the disk out until it's unmounted!
ri...@mordor.net
Martin Kammerhofer <da...@sbox.tu-graz.ac.at> wrote:
> Gary Howland (ghow...@hotlava.com) wrote:
> : > acah...@mercury.cs.uml.edu (Albert D. Cahalan) writes:
> :
> : (I think I'm correct in stating that 'rm' on FreebSD will not return
> : until the disks have been synced - how else can you maintain a stable
> : filesystem?)
> :
> You can do the same tradeoff with FreeBSD's native UFS/FFS filesystems.
> See the ``async'' option of mount(8).
> Generally I recommend against it, except for /tmp and the like.
Really. And it's not much of a performance gain either.
Chris
--
Chris Mauritz | Network Security & Design
Network Engineer | 56k-T3 connectivity to the Net.
ri...@mordor.net <--fun biz--> | http://www.new-york.net/
Eric Lee Green
On Sun, 23 Nov 1997 16:09:58 -0800, Jordan K. Hubbard <j...@FreeBSD.org> wrote:
>Clyde Wary wrote:
>FreeBSD will run BSD/OS, SCO, SVR4 and Linux binaries. Linux will run
>SCO and SVR4 binaries, as far as I know, so you'll miss the BSD/OS
>emulation there (you can, of course, run Linux binaries just fine under
>Linux :-).
According to the README for the iBCS emulator for Linux, it will run:
* Sparc Solaris
* i386 BSD (386BSD, FreeBSD, NetBSD, BSDI/386)
- very alpha, very old.
* SVR4 (Interactive, Unixware, USL, Dell etc.)
* SVR3 generic
* SCO (SVR3 with extensions for symlinks and long filenames)
* SCO OpenServer 5
* Wyse V/386 (SVR3 with extensions for symlinks)
* Xenix V/386 (386 small model binaries only)
* Xenix 286
>> FreeBSD to install, compared with, say, Red Hat Linux, or Slackware
>> Linux? I am not a skilled hacker, but I am alot more competent than
>
>I'd say it's comparable.
Probably.
About the only thing that's stopping me from using FreeBSD right now is lack of
FAT support (I tend to switch between Unix and Win95 a lot, due to there not
being any decent music software for Unix :-( ). Well, plus lack of support
for my tape drive (a Connor/Seagate Tapestor 800), although that may have been
fixed since I last looked.
--
Eric Lee Green ex...@softdisk.com Executive Consultants
Systems Specialist Educational Administration Solutions
You might be a redneck if you put on insect repellant prior to a date.
John Szumowski
Zenin wrote:
>
> Eric Lee Green <e_l_...@hotmail.com> wrote:
> >snip<
> : About the only thing that's stopping me from using FreeBSD right now is lack of
> : FAT support
>
> Actually FreeBSD has had FAT support for years. The only problem
> is that it's vary likely to corrupt your file systems. But then,
> Win95 is also vary likely to currupt your file systems so you see
> this really isn't a problem. :-)
>
> --
> -Zenin
> ze...@best.com
*very* likely? that seems a bit of a stretch...
anyways, mtools does a decent enough job of dealing with FAT.
Zenin
Zenin
John Szumowski <ha...@javanet.com> wrote:
>snip<
: *very* likely? that seems a bit of a stretch...
From man mount_msdos:
"The msdos filesystem is not known to work reliably with filesystems
created by versions of MS-DOS later than version 3.3."
Basically, almost all dos file systems you're likely to find.
"If you see the warning:
mountmsdosfs(): Warning: root directory is not a multiple of the cluster-
size in length
then it is possible that writing to the MS-DOS filesystem would produce
corruption on the disk. This is a shortcoming in the code which needs to
be addressed."
Which is pretty much what you will get if you try and mount almost any
FAT disk that was made with DOS > v3.3, in my experiance. Not only
did it corrupt my disk, it caused the system to crash hard. I've only
played with it a few times, and this has always been the case with >3.3
disks. -Which is why I've got a DOS 3.3 floppy to reformat FAT drives. :)
: anyways, mtools does a decent enough job of dealing with FAT.
It's ok and is much safer then mount_msdos, but it's a klugy system.
--
-Zenin
ze...@best.com
stuart henderson
> Statically linked sh and mount are also useful for crash recovery.
and put a copy in a directory called something like
"dont-move-this-even-if-you-run-out-of-disk-space-in-root" <grin>.
Richard Tobin
In article <EK7uJ...@nonexistent.com> ri...@mordor.net writes:
>> See the ``async'' option of mount(8).
>> Generally I recommend against it, except for /tmp and the like.
>Really. And it's not much of a performance gain either.
Well, that depends on what you're doing. If you're creating lots of
files, it can be a huge win. Restoring to a new filesystem is a good
example: it can be 5 times faster (the FreeBSD install is *nuch*
faster now that it uses it).
-- Richard
--
Because of all the junk e-mail I receive, all e-mail from .com sites is
automatically sent to a file which I only rarely check. If you want to mail
me from a .com site, please ensure my surname appears in the headers.
David E. O'Brien
John S. Dyson <ro...@dyson.iquest.net> wrote:
> (Building a shell such as "bash" shared can be significantly
> counterproductive.)
Sorry to say, that's how we build both bash-1.x and bash-2.x in the ports
system...
--
-- David (obr...@NUXI.com -or- obr...@FreeBSD.org)
James says: "Grad school sucks."
David E. O'Brien
Jeremy Nelson <j...@enteract.com> wrote:
> /usr/bin/tar (dynamically linked) wouldnt be really neccesary in crash
It's also statically linked (and still in /usr/bin) in 2.2.5.
David E. O'Brien
> memory allocator that won't tolerate code that tries to free data
> on the stack, code that frees data twice, and code that goes off the
I'm courious about their design. Do you know it well enough to explain
it to me?
--
-- David (obr...@NUXI.com)
Simon Maurice
I'm a recent convert from Red Hat Linux to FreeBSD and having tried the
'best' Linux distribution and now FreeBSD I'm staying right where I am.
I'm no great OS hacker so I can't really talk about what's going on
under the hood but I can say this:
I'm running the Linux version of Netscape using the Linux
compatability
libraries and Netscape RUNS BETTER UNDER FREEBSD THAN IT DOES UNDER
LINUX!!
Netscape is more stable and loads faster under FreeBSD which I suppose
is like Sun's WABI running Windows apps faster than a native Windows
machine :)
ciao
Simon Maurice
Dana Booth
In article <348395F8...@planet.net.au>, sy...@planet.net.au
says...
> I'm running the Linux version of Netscape using the Linux
>compatability
> libraries and Netscape RUNS BETTER UNDER FREEBSD THAN IT DOES UNDER
>LINUX!!
But it looks like you're running the bsd version??
> X-Mailer: Mozilla 4.03 [en] (X11; I; FreeBSD 2.2.5-RELEASE i386)
^^^
--
----------------------------
Dana Booth <da...@mmi.oz.net>
----------------------------
Jerry Hicks
Dana Booth wrote:
>
> In article <348395F8...@planet.net.au>, sy...@planet.net.au
> says...
>
> > I'm running the Linux version of Netscape using the Linux
> >compatability
> > libraries and Netscape RUNS BETTER UNDER FREEBSD THAN IT DOES UNDER
> >LINUX!!
>
> But it looks like you're running the bsd version??
>
> > X-Mailer: Mozilla 4.03 [en] (X11; I; FreeBSD 2.2.5-RELEASE i386)
I believe this is derived dynamically under emulation. That is, it is
not an indication of the Netscape version, but rather the host OS.
BTW: This isn't the first report of Linux programs executing faster
under FreeBSD emulation.
Cheers,
Jerry Hicks
jerry...@bigfoot.com
Dana Booth
In article <34845214...@ix.netcom.com>,
Jerry Hicks <wghh...@ix.netcom.com> writes:
>> But it looks like you're running the bsd version??
> I believe this is derived dynamically under emulation. That is, it is
> not an indication of the Netscape version, but rather the host OS.
I see... I just wanted to be sure he was being accurate.
--
-----------------------------
Dana Booth <da...@dana.oz.net>
-----------------------------
cu...@journyx.com
In article <348395F8...@planet.net.au>,
sy...@planet.net.au wrote:
>
> I'm a recent convert from Red Hat Linux to FreeBSD and having tried the
> 'best' Linux distribution and now FreeBSD I'm staying right where I am.
> I'm no great OS hacker so I can't really talk about what's going on
> under the hood but I can say this:
>
> compatability
> libraries and Netscape RUNS BETTER UNDER FREEBSD THAN IT DOES UNDER
> LINUX!!
>
> is like Sun's WABI running Windows apps faster than a native Windows
> machine :)
i like both
i've found freebsd has less free stuff already ported and
some commands are slower (such as rm -rf /bigdir)
our product (below) runs on both pretty much identically
__________________________________________________________________
Web-Based Time Tracking journyx WebTime
is FREE for 60 Days at (512)345-8282
http://journyx.com/wts.html cu...@journyx.com
------------------------------------------------------------------
-------------------==== Posted via Deja News ====-----------------------
http://www.dejanews.com/ Search, Read, Post to Usenet
Jordan K. Hubbard
cu...@journyx.com wrote:
> i've found freebsd has less free stuff already ported and
> some commands are slower (such as rm -rf /bigdir)
Hmm. The ports collection is now at 1200 ports, which isn't too shabby
and browsing it certainly exceeds the amount of time I have for looking
at free software. :)
As to the rm -rf /bigdir, I suspect that you simply didn't have the
FreeBSD filesystem mounted with the async (metadata update) flag set.
This would make it operate more or less identically to Linux when
tossing large numbers of files away, albeit at greater risk should
lightning strike your machine at exactly the wrong moment. :)
--
- Jordan Hubbard
FreeBSD core team / Walnut Creek CDROM.
Simon Maurice
>
> > X-Mailer: Mozilla 4.03 [en] (X11; I; FreeBSD 2.2.5-RELEASE i386)
> ^^^
Hmmmm, interesting but when I installed Netscape, the distribution CD
had no BSD distribution but had Linux versions for V1.3 and 2,0 kernels.
I installed the 2.0 Linux kernel version (as I had done on Linux) so as
far as I can tell it's the linux version.
That aside. Let's assume that I'm wrong. Let's assume that the
installation script did something on the sly and didn't install the
Linux version but a native one. The fact is still outragously clear that
under FreeBSD netscape loads faster, is more stable (under Xfree86) and
responds to user input slightly quicker than it did under Red Hat Linux
4.2.
Goodbye Linux
Simon Maurice
bi...@bmorgan.com
Is there much/any differences in terms of POSIX compatability
between FreeBSD and Linux?
TIA
la...@ws6502.gud.siemens.co.at
Simon Maurice <sy...@planet.net.au> writes:
>
> That aside. Let's assume that I'm wrong. Let's assume that the
> installation script did something on the sly and didn't install the
> Linux version but a native one. The fact is still outragously clear that
> under FreeBSD netscape loads faster, is more stable (under Xfree86) and
> responds to user input slightly quicker than it did under Red Hat Linux
> 4.2.
>
I attribute the observed acceleration to the FreeBSD aggressive swapping
strategies.
What FreeBSD VM does is as follows: the machine always pages out, even
some pages that are not even dirty (swap space is cheap), even when there
are no requests for free pages. The result of this is that when a program
requires a lot of free pages, practically all of the RAM will have been
written into swap and the pages can be immediately allocated to the
program that required the memory.
Linux seems to delay pageout until the moment it is really needed which
reduces the swap usage but once the new pages are needed the program has
to wait until the system has paged out enough pages. Windows and NT seem
to do the same.
This difference is advantageous for FreeBSD in a case of somewhat overloaded
machine (or a heavily overloaded machine as long as it's a single user
one--I actually manage to work with Xemacs, gcc, Netscape and latex/xdvi
on a 16 MB machine: task switching is relatively fast as only a pagein
is needed, not a pageout of the previously used task).
Paging out clean pages rather than simply discarding them is advantageous
as well because pagein from swap is faster than pagein from a file.
Naturally, clean page pageout should be done only if the machine is idle
enough (otherwise it is time consuming).
/Marino
--
As far as the differences between BSD and Systems V, that's
simple. System V sucks and BSD doesn't. :) -- Curt Welch
UNIX _is_ user friendly. It's just selective about who its
friends are. -- Marco Molteni
stephen farrell
la...@ws6502.gud.siemens.co.at (marino....@siemens.at) writes:
> Simon Maurice <sy...@planet.net.au> writes:
>
> ..
It is funny, the diversity of experience here. A while back I was
considering switching back to linux seriously. The reason? FreeBSD
(2.2.2) was a memory hog. I could not run Netscape4, AccelX, and
XEmacs (speaking of memory hogs =) comfortably in 64MB of ram, while I
could in Linux (2.0.x). I decided my happiness with FreeBSD otherwise
justified spending $150 (or whatever) for *another* 64MB or RAM. I've
effectively eliminated swapping just to get a usable system. But I
feel gross needing 128MB for an acceptable system.
In my experience, swapping under freebsd... I dunno, you might as well
shoot yourself in the head. It just kills the system (P6-150). The
mouse stops moving for as long as 30 seconds just switching back and
forth between netscape and xemacs on an otherwise unloaded 64M system.
It is horribly unnacceptable. (and yes, it knew about all of the
RAM... though believe me I checked many times!)
Perhaps I have hardware issues (it's a first generation P6
motherboard, EIDE harddrive), but even so, this problem didn't occur
under linux. linux was ok w/32M and similar usage, but 64 was roomy
(as 128 is w/freebsd).
Additionally, there is a very odd problem with Netscape under FreeBSD.
If I have a large HTML page (say 50 or 100k) and do a text search, the
cursor locks up, it starts swapping painfully (the way it used to
before I upgraded), and the mouse stops moving. 10-20 seconds at
times. Viewing with top, I can see that the "Free" amount of RAM
changes from say 20MB to 128K and vacillates like such while the mouse
is locking up. Some serious issues with the memory allocation, I
guess. This happens w/NS3, NS4 and linux NS. It does *not* happen
with linux.
Otoh, as long as it does not need to swap, I've been very happy with
the performance of my FreeBSD box.
--sf
John S. Dyson
In article <874t4qu...@phaedrus.uchicago.edu>,
stephen farrell <sfarrel...@farrell.org> writes:
>
> It is funny, the diversity of experience here. A while back I was
> considering switching back to linux seriously. The reason? FreeBSD
> (2.2.2) was a memory hog. I could not run Netscape4, AccelX, and
> XEmacs (speaking of memory hogs =) comfortably in 64MB of ram, while I
> could in Linux (2.0.x). I decided my happiness with FreeBSD otherwise
> justified spending $150 (or whatever) for *another* 64MB or RAM. I've
> effectively eliminated swapping just to get a usable system. But I
> feel gross needing 128MB for an acceptable system.
>
been working on improvements to help apps that use the old malloc, and
PHK has written a new one. When you run *exactly* the same application,
with *exactly* the same memory footprint, the app will run very nicely
on FreeBSD compared to Linux (I am not saying better or worse, but it
will run well.)
Our paging code depends on locality of reference, and if the needed memory
isn't predictable, then our code will degrade to a thrash (just like any
VM code.) We definitely have been stunted by our userland malloc, and
hopefully that problem is gone. If you run a Linux app, or an app that
was linked with our new malloc, the situation is much better. (I think that
PHK, with his new malloc, has done as much or more good than a year of kernel VM
optimizations.)
We do not prepage at all, but we do tend to push out old, unused pages
pretty quickly. This is due a heuristic that buffer cache pages that are used
a little bit count for some usage. Old pages that might be used only once
in an app eventually get pushed. We try not to push pages that have been
used recently (but we definitely do not use a simple LRU.) We do, as an
attempt to take advantage of locality of reference, sometimes push pages that
have been used recently, but are adjacent to pages that have fallen into
disuse.
YMMV.
--
John
dy...@freebsd.org
jdy...@nc.com
Eric Lee Green
On 03 Dec 1997 19:39:06 +0100, marino....@siemens.at
>I attribute the observed acceleration to the FreeBSD aggressive swapping
>strategies.
>
>What FreeBSD VM does is as follows: the machine always pages out, even
>some pages that are not even dirty (swap space is cheap), even when there
>are no requests for free pages. The result of this is that when a program
This is configurable in Linux. See
/usr/src/linux/Documentation/memory-tuning.txt which explains how to
tune your paging. I've noticed that fiddling with the values in
/proc/sys/vm/freepages makes a great deal of difference in my
percieved performance. (freepages controls when background paging will
commence etc.).
I don't know why the default Linux values for 'freepages' are so stupid :-(.
(Well, yes I do -- they weren't so stupid when you had only 4mb of RAM -- but
they should have been adjusted for today's memory costs).
John S. Dyson
In article <836EFAA2EBE65188.7D0CDB08...@library-proxy.airnews.net>,
e_l_...@hotmail.com (Eric Lee Green) writes:
> On 03 Dec 1997 19:39:06 +0100, marino....@siemens.at
> <la...@ws6502.gud.siemens.co.at> wrote:
>>I attribute the observed acceleration to the FreeBSD aggressive swapping
>>strategies.
>>
>>What FreeBSD VM does is as follows: the machine always pages out, even
>>some pages that are not even dirty (swap space is cheap), even when there
>>are no requests for free pages. The result of this is that when a program
>
> (Well, yes I do -- they weren't so stupid when you had only 4mb of RAM -- but
> they should have been adjusted for today's memory costs).
>
tuning is mostly in a pseudo-mib form.
S Telford
Dana Booth (da...@mmi.oz.netDELETE.CAPS) wrote:
: In article <348395F8...@planet.net.au>, sy...@planet.net.au
: says...
: > I'm running the Linux version of Netscape using the Linux
: >compatability
: > libraries and Netscape RUNS BETTER UNDER FREEBSD THAN IT DOES UNDER
: >LINUX!!
: But it looks like you're running the bsd version??
: > X-Mailer: Mozilla 4.03 [en] (X11; I; FreeBSD 2.2.5-RELEASE i386)
: ^^^
Linux Netscape doesn't have "Linux" hard-coded into it; it gets the OS
identity from a system call. On NetBSD/i386 1.2 it'll say "NetBSD 1.2"
too.
--
Scott Telford "If 386BSD had been available when
Edinburgh Parallel Computing Centre I started on Linux, Linux would
University of Edinburgh, EH9 3JZ, UK. probably never had happened."
s.te...@ed.ac.uk +44 131 650 5978 - Linus Torvalds
la...@ws6502.gud.siemens.co.at
ro...@dyson.iquest.net (John S. Dyson) writes:
> We do not prepage at all, but we do tend to push out old, unused pages
> pretty quickly. This is due a heuristic that buffer cache pages that are used
> a little bit count for some usage. Old pages that might be used only once
> in an app eventually get pushed. We try not to push pages that have been
> used recently (but we definitely do not use a simple LRU.) We do, as an
> attempt to take advantage of locality of reference, sometimes push pages that
> have been used recently, but are adjacent to pages that have fallen into
> disuse.
Funny that, but you should know--you wrote the code (thanks for the
corection, BTW).
I was under impression that prepaging was actually used and was very happy
about it :) I even wrote some tests which seemed to confirm that and
watched the swap usage growth even when the system was idling, but this
could have been an artefact of 16 MB machine which is by definition low
on memory. I'm still running 2.1.5 and old malloc, BTW (waiting for
the 2.2.5 to appear in Austria). OTOH, my disks are SCSI which helps a
lot on a DX-33 486.
However, locality of reference paging seemed to help a lot when I was
rendering largish models in Radiance (20 MB VSIZE, 11 MB RSS; and the
load still does not drop under 80 percent--way to go!) Of course, it
would seem that Radiance renderer behaves very well--it does not access
memory all over the picture.
Chris Taylor
> It is funny, the diversity of experience here. A while back I was
> considering switching back to linux seriously. The reason? FreeBSD
> (2.2.2) was a memory hog. I could not run Netscape4, AccelX, and
> XEmacs (speaking of memory hogs =) comfortably in 64MB of ram, while I
> could in Linux (2.0.x). I decided my happiness with FreeBSD otherwise
> justified spending $150 (or whatever) for *another* 64MB or RAM. I've
> effectively eliminated swapping just to get a usable system. But I
> feel gross needing 128MB for an acceptable system.
>
Just a few words.....
YOU ARE FUCKED. That's the most ridiculous thing I've heard on this newsgroup.
--Chris
stephen farrell
Chris Taylor <ch...@christaylor.com> writes:
> ..
To avoid any misunderstanding, let me start by saying that I am
reading this with a sense of humor, and assume it was written with
one... However I *am* curious as to exactly what you mean... are you
surprised that I would need so much RAM w/FreeBSD? Are you surprised
that I would pay $150 to stick with FreeBSD? Or something else?
--sf
Zenin
stephen farrell <sfarrel...@farrell.org> wrote:
: It is funny, the diversity of experience here. A while back I was
: considering switching back to linux seriously. The reason? FreeBSD
: (2.2.2) was a memory hog.
Hmm, interesting. My current configuration is:
Intel P5/200 mhz
64 MB RAM / 128 MB swap
3 NFS mounted drives
3 IDE drives (2 are striped together with ccd)
Running, 24/7:
MiniSQL 2.0
Apache w/mod_perl and ApacheDBI
NcFTPd
Stock Quake server w/CTF mods (16 megs dedicated)
QW Quake server w/Arena mods (another 16 megs dedicated)
XFree86 at 16 bbp
AfterStep 1.0
Netscape 4.04
WebStats (internal company stats engine that parses our
20 something high traffic websites) -CPU hog mostly
NNTPCache
A dozen or so xterms and copys of joe (my editor)
And a number of misc, smaller servers for internal
company use.
Sure, swap will get a little big but it still doesn't really break
a sweat. Always smooth as glass and rock solid. I attribute
FreeBSD's advanced pager and schedular. Disk space is cheap, and
getting clean pages from swap is much less expensive then from
the file system.
If you've got problems running your rather minimal set of programs
with 64 Megs on a P5, it sounds to me like something is vary out
of wack. How much swap do you have? Have you reconfigured and
recompiled your kernel? What video card/driver are you using? -This
can affect GUI stuff like Netscape and Xemacs a _lot_.
--
-Zenin
ze...@best.com
J Wunsch
ro...@dyson.iquest.net (John S. Dyson) wrote:
> I think that can be partially attributed to the old malloc.
...with the other part (the general sluggishness while paging)
attributing to the IDE disk. ISTR that there's DMA support for IDE
now, but there wasn't in 2.2.2, so it's quite clear that heavy paging
costs a lot of CPU.
I have never observed that general sluggishness with SCSI disks (on a
decent controller, of course). The X11 cursor was moving around until
the machine got loaded enough to really go thrashing (which, of
course, is possible).
--
cheers, J"org
joerg_...@uriah.heep.sax.de -- http://www.sax.de/~joerg/ -- NIC: JW11-RIPE
Never trust an operating system you don't have sources for. ;-)
AP
In article <34872458...@christaylor.com>, ch...@christaylor.com
says...
>
> Just a few words.....
>
> YOU ARE @*!. That's the most ridiculous thing I've heard on this newsgroup.
>
> --Chris
>
>
Do you have to use that language?
Brian Somers
[Article cross-posted to sender]
In article <34872458...@christaylor.com>,
Chris Taylor <ch...@christaylor.com> writes:
: Just a few words.....
:
: YOU ARE FUCKED. That's the most ridiculous thing I've heard on this newsgroup.
Wow ! I'd just love to get criticism *this* constructive to
some of my postings. Chris, didn't your parents ever teach you
*any* manners ? Even if the original post was flame bait (I
don't personally think it was), this is *not* a response I'd
expect from any associate of mine.
Now go back to your parents, and after they've wiped the dribble
from your chin, you can ask them for a lesson in etiquet and
human interaction.
'till then, please refrain from turning this newsgroup into
something that people find offensive.
: --Chris
:
--
Brian <br...@Awfulhak.org> <br...@FreeBSD.org> <br...@OpenBSD.org>
<http://www.Awfulhak.org>
Don't _EVER_ lose your sense of humour !
Ron Echeverri
In article <MPG.ef17ae44...@news.idt.net>,
AP <anth...@mail.idt.net> wrote:
>In article <34872458...@christaylor.com>, ch...@christaylor.com
>says...
>> YOU ARE @*!. That's the most ridiculous thing I've heard on this newsgroup.
>Do you have to use that language?
What language do you mean? "FUCKED", perhaps?
rone
--
Ron Echeverri <rone...@best.net> | "Who is the boss of you? Me! I am
Systems/Usenet Administration | the boss of you! I am the boss of
Best Internet Communications, Inc. | you! I am the boss of you!"
<URL:http://www.ennui.org/~rone/> | - The Grand Inquisitor
Shawn Ramsey
> > > It is funny, the diversity of experience here. A while back I was
> > > considering switching back to linux seriously. The reason? FreeBSD
> > > XEmacs (speaking of memory hogs =) comfortably in 64MB of ram, while I
> > > could in Linux (2.0.x). I decided my happiness with FreeBSD otherwise
> > > justified spending $150 (or whatever) for *another* 64MB or RAM. I've
> > > effectively eliminated swapping just to get a usable system. But I
> > > feel gross needing 128MB for an acceptable system.
> > >
> >
> >
> > YOU ARE FUCKED. That's the most ridiculous thing I've heard on this
> > newsgroup.
> >
> > --Chris
>
> To avoid any misunderstanding, let me start by saying that I am
> reading this with a sense of humor, and assume it was written with
> one... However I *am* curious as to exactly what you mean... are you
> surprised that I would need so much RAM w/FreeBSD? Are you surprised
> that I would pay $150 to stick with FreeBSD? Or something else?
Regardless, FreeBSD is hardly to blame. Netscape is such a HUGE memory
hog... I don't run Xemacs, but I can run Netscape XFree86, and a couple
other things just fine.
Matt Dillon
:In article <iy7g1o8...@ws6502.gud.siemens.co.at>,
:marino....@siemens.at <la...@ws6502.gud.siemens.co.at> wrote:
:>ro...@dyson.iquest.net (John S. Dyson) writes:
:..
:>
:>Funny that, but you should know--you wrote the code (thanks for the
:>corection, BTW).
:>
:>I was under impression that prepaging was actually used and was very happy
:>about it :) I even wrote some tests which seemed to confirm that and
:>watched the swap usage growth even when the system was idling, but this
FreeBSD's paging system is both simple and complex. The basic premis,
though, is that a page must go through several states before it is
actually unrecoverably freed.
In FreeBSD a page can be:
wired most active state
active nominally active state
inactive dirty & inactive (can be reclaimed to active/wired)
cache clean & inactive (can be reclaimed to active/wired)
free truely free
FreeBSD attempts to move wired/active pages to either inactive or cache
depending on whether they are dirty or not. When pages are needed,
FreeBSD will do a combination of paging of inactive pages (which cleans
them and moves them to cache), and movement of pages from cache to free.
Actual page allocation always comes out of free. Sysctl parameters
handle low and high water marks allowing one to control the burstiness
and aggressiveness of the page movement.
In a nominal paging situation, an active process may see a page get
moved to inactive/cache, but if the process references the page relatively
quickly the page will be 'reclaimed' (see vmstat, systat -vm) and moved
back to active/wired without requiring a buffer copy or disk I/O.
In a nominal paging situation, a dirty page in inactive is written to swap
and turned into a clean page in cache, but still not immediately freed.
It is still possible to reclaim the page from cache and move it back to
active even if it has been paged out.
A very heavy paging situation occurs when FreeBSD is unable to maintain
the minimum cache+free paramater. When this occurs, FreeBSD begins to
*swap* whole processes. This is an area where work has been needed
for a while because certain programs such as 'ps' tend to force all
swapped (IW) processes back to idle (I) and the swap-trigger is just
too sensitive... it's trivial to fix and -current already has the
better swapping code, with -stable soon to follow. The new code is
going to have the capability to swap idle processes out based on how long
they've been sleeping (SUNish style swapping). In a heavily loaded
time-share system such as BEST's shell machines, this results in a
constant, slow movement of processes both into and out of the swapped
state and better balances new allocations against frees. In fact, even
swapping out a process does not immediately free its pages... it's still
possible for the process to be swapped in without incuring disk I/O though,
usually, the pages are lost because the swapping only occurs in a heavily
loaded situation anyway.
It should be noted that the disk cache in FreeBSD is comprised of ALL the
paging pools, not just the 'cache' paging pool, but FreeBSD has various
sysctl VM parameters that allow you to separate a true 'disk cache' from
active pages used by processes. A true 'disk cache' infers the caching
of blocks not currently in use. Thus sysctl allows you to specify a
minimum size for the 'cache' queue which in turn allows you to balance
paging against cache maintenance. The larger the minimum 'cache' size,
the more aggressive FreeBSD pages in order to maintain the larger
'cache'. A smaller minimum 'cache' size results in less aggressive
paging, but a higher chance of paging an active page out in a heavily
loaded system.
-Matt
:>/Marino
--
Matthew Dillon Engineering, BEST Internet Communications, Inc.
<dil...@best.net>, include original article w/ any response.
do not under any circumstances send email to joe...@bigspender.idiom.com
and, for gods sake, don't email buck...@popserver.idiom.com
Amancio Hasty
stephen farrell wrote:
Hi,
I run XFree86 16 bit , Enlightenment , Xemacs and couple of apps.
My system does flicker a little when I switch to a virtual desktop .
My setup :
OS: FreeBSD 3.0-current
PPro 200MHz 48Mb
Adaptec 2940
SEAGATE ST34501W 0017 (Cheetah 10000 rpm , 4.3 Gig)
Matrox Millenium 4M Vram
Brootree 848 PCI video capture .
Netscape 4.04 for FreeBSD
As I am typing this I am also compiling the kernel just for kicks .
The response is very good and I can't notice that on one of my virtual
desktops I am compiling the kernel and for sheer pleasure
I am watching TV with fxtv so I am pumping an extra
32MB/sec thru the PCI bus.
I happen to also have 3.2 GIG Fujisu IDE drive which with DMA is very nice
in
fact I sometimes use the disk with fxtv to do software mpeg encoding.
Enjoy,
Amancio
TonyP
In article <669m37$n9t$1...@bofh.noc.best.net>,
rone+...@bofh.noc.best.net says...
> In article <MPG.ef17ae44...@news.idt.net>,
> AP <anth...@mail.idt.net> wrote:
> >In article <34872458...@christaylor.com>, ch...@christaylor.com
> >says...
> >> YOU ARE @*!. That's the most ridiculous thing I've heard on this newsgroup.
> >Do you have to use that language?
>
> What language do you mean? "FUCKED", perhaps?
>
> rone
> --
> Ron Echeverri <rone...@best.net> | "Who is the boss of you? Me! I am
> Systems/Usenet Administration | the boss of you! I am the boss of
> Best Internet Communications, Inc. | you! I am the boss of you!"
> <URL:http://www.ennui.org/~rone/> | - The Grand Inquisitor
>
Ron Echeverri
In article <MPG.efbfab32...@news.idt.net>,
TonyP <anth...@mail.idt.net> wrote:
>ok ronnie.
That's "Ronny", buddy.
rone
--
Ron Echeverri <rone...@best.net> | "Happiness is free, and love is
Systems/Usenet Administration | relatively cheap, but HATE costs
Best Internet Communications, Inc. | more than you have in your wallet."
<URL:http://www.ennui.org/~rone/> | - Tjames Madison <tja...@pigdog.org>
TonyP
In article <6749hl$j3d$1...@bofh.noc.best.net>,
rone+...@bofh.noc.best.net says...
> In article <MPG.efbfab32...@news.idt.net>,
> TonyP <anth...@mail.idt.net> wrote:
> >ok ronnie.
>
> That's "Ronny", buddy.
OK, pal.
Joshua Pincus
Chris Taylor <ch...@christaylor.com> wrote:
> > It is funny, the diversity of experience here. A while back I was
> > considering switching back to linux seriously. The reason? FreeBSD
> > (2.2.2) was a memory hog. I could not run Netscape4, AccelX, and
> > XEmacs (speaking of memory hogs =) comfortably in 64MB of ram, while I
> > could in Linux (2.0.x). I decided my happiness with FreeBSD otherwise
> > justified spending $150 (or whatever) for *another* 64MB or RAM. I've
> > effectively eliminated swapping just to get a usable system. But I
> > feel gross needing 128MB for an acceptable system.
> >
> Just a few words.....
> YOU ARE FUCKED. That's the most ridiculous thing I've heard on this newsgroup.
I wouldn't go that far. I think Mr. Taylor's system hasn't been configured
correctly. I used to administer a system that had 128 MEGs of memory, but it
served 800 users and provided everything from xdm for an entire network to
email. We had/have about 20-25 users on simultaneously at any given point in
time. Perhaps Mr. Taylor isn't using virtual memory? One of the things that I
love about FreeBSD is it's effective use of VM.
The system I administered was blazenly fast. Even when I ran xemacs, netscape,
xfig, fvwm2, several xterms, xdaliclock, xru, xbiff, xdvi, and xanim at the
same time as user number 20. ;)
JP
> --Chris
--
Joshua Pincus
UNIX Programmer
University of Rochester Computing Center
Rochester, New York 14627
Dana Booth
Chris Taylor <ch...@christaylor.com> wrote:
> Just a few words.....
>
> YOU ARE FUCKED.
That seems to be the general attitude from the FreeBSD crowd toward anyone
who has a problem with their software. Isn't it sorta funny? About half of
all the articles written here are 'us v. Linux'. Mention Linux, and the
religious faithful crawl out, insults in hand. If assholes like yourself
spent less time insulting people who have installation or operating woes,
and spend less time defending FreeBSD against that terrible and evil Linux,
and spend a little more time either helping out or playing with your kids,
this NG wouldn't seem like such a bad joke.
--
-----------------------------
Dana Booth <da...@dana.oz.net>
-----------------------------
John S. Dyson
In article <67ra4n$6a8$1...@hourglass.oz.net>,
da...@dana.oz.net (Dana Booth) writes:
> Chris Taylor <ch...@christaylor.com> wrote:
>
>> Just a few words.....
>>
>
> That seems to be the general attitude from the FreeBSD crowd toward anyone
> who has a problem with their software.
>
reasons.
>
> Isn't it sorta funny?
>
No, it is sad. The only good thing that is coming of it is that it appears
that we are starting to attract a younger crowd. Maybe the new group of
kids will participate in, as opposed to trying to reinvent the
economy. (It is a typical child's game to try to change the rules, when
they cannot play.) GPL reinvents the word "free", and tries to reinvent
the software economy.
>
> About half of
> all the articles written here are 'us v. Linux'. Mention Linux, and the
> religious faithful crawl out, insults in hand.
>
Nope, remember, it is partially an issue of GPL vs. freedom and free software.
The zealots are Linux-ites, trolling. Why did you troll? What value is there
in it? Only a fool would think that the statement made above was from a
serious FreeBSD person. Why did you appear to take it seriously?
>
> If assholes like yourself
> spent less time insulting people who have installation or operating woes,
> and spend less time defending FreeBSD against that terrible and evil Linux,
> and spend a little more time either helping out or playing with your kids,
> this NG wouldn't seem like such a bad joke.
>
I don't agree with the above statement either. Linux isn't terrible and
evil, but it is really sad that people have to try to reinvent things, and
fragment free software, with silly, encumbering licenses and political agendas.
Why are you calling the (likely) kid an a**hole? Frankly, you are
perpetuating the childish behavior. He/she might only be as old as a kid
that I could have had.
Let's all make a New Year's resolution of not trolling or starting flamage.
This is not an advocacy newsgroup. If you had something intelligent to say,
I would not have responded the way that I did. Note that I didn't make
unsupported accusations. They were fully supported.
The above notwithstanding,
Merry Christmas.
--
John | Never try to teach a pig to sing,
dy...@freebsd.org | it just makes you look stupid,
jdy...@nc.com | and it irritates the pig.
Ron Echeverri
In article <67ra4n$6a8$1...@hourglass.oz.net>,
Dana Booth <da...@dana.oz.net> wrote:
>Chris Taylor <ch...@christaylor.com> wrote:
>> Just a few words.....
>That seems to be the general attitude from the FreeBSD crowd toward anyone
>who has a problem with their software.
HI DANA!!! R U QTE?? PLEEZ EMAIL ME 4 A GOOD TIME
BIFF!!!1
Matt Dillon
:In article <67rrdv$gtp$1...@bofh.noc.best.net>,
:Ron Echeverri <ro...@bofh.noc.best.net> wrote:
:>In article <67ra4n$6a8$1...@hourglass.oz.net>,
:>Dana Booth <da...@dana.oz.net> wrote:
:>>Chris Taylor <ch...@christaylor.com> wrote:
:>>> Just a few words.....
:>>> YOU ARE FUCKED.
:>>That seems to be the general attitude from the FreeBSD crowd toward anyone
:>>who has a problem with their software.
:>
:>HI DANA!!! R U QTE?? PLEEZ EMAIL ME 4 A GOOD TIME
:>
:>BIFF!!!1
:>--
:>Ron Echeverri <rone...@best.net> | "Happiness is free, and love is
Oh shit! we put too much ext... ahhh caffine in the free drink dispenser
again.
-Matt
W. Scholten
John S. Dyson wrote:
[sad things snipped]
> da...@dana.oz.net (Dana Booth) writes:
> > Isn't it sorta funny?
> >
> No, it is sad. The only good thing that is coming of it is that it appears
> that we are starting to attract a younger crowd. Maybe the new group of
> kids will participate in, as opposed to trying to reinvent the
> economy. (It is a typical child's game to try to change the rules, when
> they cannot play.) GPL reinvents the word "free", and tries to reinvent
> the software economy.
I have no problem with the redefinition of free: this is a process which
is common to most specialized fields, where a term gets a meaning quite
different from the normal, everyday meaning, see e.g. mathematics.
However, when someone of whom you're not sure is working in this field
says something, you cannot interpret this by the rules of the field.
This is what the GNU zealots do al the time. When I say, 'you can't use
GPL programs commercially' EVERYONE knows what I mean. Only these
nincompoops insist on erroneously correcting you which only amounts to
showing their ignorance of life, both down to earth and abstract.
...
> I don't agree with the above statement either. Linux isn't terrible and
> evil, but it is really sad that people have to try to reinvent things, and
> fragment free software, with silly, encumbering licenses and political agendas.
Well, nothing is freeer than a PD license. In the GPL, the 'we preserve
freedom' is actually BS as the original PD version of any PD based
program is still available to whoever wants it and making a new version
proprietary does not reduce the freedom of anyone. Ok, so you don't have
the source/rights to do with the new program what you want, but
otherwise the programmer would have begun writing the program from
scratch and you end up in the same situation which means that all that
was achieved by the GPL is a duplication of effort. GNU zealots don't
understand this and the ones I've seen (gnu.misc.discuss) can't even
reason.
However, I think the priciple 'improvements/bugfixes must be released
freely' is a good one; you get something for free and in return you have
the small obligation of giving back bugfixes/improvements. However, I
use it in a reduced (simplified) and more late 90's agreeable format in
a license I dubbed CGPL (Class GPL, see the URL below for the license)
which I use for my eiffel classes and which allows compilation into any
program.
> The above notwithstanding,
> Merry Christmas.
Merry Christmas to BSD users (from a linux user).
Wouter
--
eiffel's future is so bright, we ought to wear sunglasses
http://www.cistron.nl/~wouters/eiffel/eiffel.html
J Wunsch
da...@dana.oz.net (Dana Booth) wrote:
>> Just a few words.....
>>
>> YOU ARE FUCKED.
> That seems to be the general attitude from the FreeBSD crowd toward
> anyone who has a problem with their software.
Hardly. I have read far more words of the above flavor here thrown by
you into various directions (including mine) than have been sent by
actual FreeBSD developers within the same period of time.
If some troll (like the one you've been picking for your example) is
doing wrong, this gives you no right to generalize his actions, and
imply that anyone else in the FreeBSD camp is agreeing with him. Your
followup has been sent out after that troll has already been
publically put down for his words.
Dana, if you don't like FreeBSD, don't use it. Nobody has been
forcing you to use it. If you wanna use it, and have questions,
you've been told before to pose your questions in a way that helps
solving them (i.e., concentrate on technical details, as opposed to
emotional reports that barely contain any technical description of
your actual problem at all). If you do this, you'll quickly realize
that people here are really helpful. Read the entire group, and
you'll find numerous examples for people who have been helped
successfully.
This is my last attempt to tell you that your attitude (as shwon in
the past) is considered inappropriate here. If you continue to ignore
our pleas, you only risk being ignored yourself (mainly by the means
of kill files). That's usually the only defense one has in Usenet.
Bill Gunshannon
In article <67rbve$3...@snews2.zippo.com>, ro...@dyson.iquest.net (John S. Dyson) writes:
|>
|> Let's all make a New Year's resolution of not trolling or starting flamage.
|>
John,
I'll bet you still believe in Santa Claus, too.
All the best.
bill
--
Bill Gunshannon | de-moc-ra-cy (di mok' ra see) n. Three wolves
bi...@cs.uofs.edu | and a sheep voting on what's for dinner.
University of Scranton |
Scranton, Pennsylvania | #include <std.disclaimer.h>
Destination Unkonwn
what is better linux or freebsd?
Ron Echeverri
In article <69lqe7$q...@mtinsc02.worldnet.att.net>,
CP/M.
rone
--
Ron Echeverri <rone...@best.net> | "Happiness is free, and love is
TonyP
In article <69lqe7$q...@mtinsc02.worldnet.att.net>, fro...@writeme.com
says...
> what is better linux or freebsd?
If you're a "joiner" get Linux. If you are independent get FreeBSD. ;)
Tony
(Have I been winking too much? My eye hurts!)
Jorge Goncalves
Ron uses lots of FreeBSD machines at Best Internet Communtications, Inc.
And of course FreeBSD is much, much better than Linux.
IMO.
Jorge Goncalves
Ron Echeverri <ro...@bofh.noc.best.net> wrote:
> In article <69lqe7$q...@mtinsc02.worldnet.att.net>,
> Destination Unkonwn <fro...@writeme.com> wrote:
> >what is better linux or freebsd?
> CP/M.
te...@portsoft.com
In <MPG.f284ccee...@news.idt.net>, anthonp...@mail.idt.net (TonyP) writes:
>In article <69lqe7$q...@mtinsc02.worldnet.att.net>, fro...@writeme.com
>says...
>
>
>Tony
>(Have I been winking too much? My eye hurts!)
And if your a "lamer" you don't have to do anything other than sit here and
read about it, and make the occassional inflammatory postings.
(Sorry all, I couldn't resist)
Marcus Meissner
Real Life ?
Marcus
--
<URL:http://wwwcip.informatik.uni-erlangen.de/~msmeissn/>
GNU is the only way to build pyramids WITHOUT slaves.
-- Klaus Schilling in a BSD vs GPL thread
Lee Cremeans
In article <69m25i$9np$1...@freefall.telepac.pt>,
Jorge Goncalves <j...@freefall.telepac.pt> writes:
> Ron uses lots of FreeBSD machines at Best Internet Communtications, Inc.
>
> And of course FreeBSD is much, much better than Linux.
>
> IMO.
Um, you're not helping. I personally am sick of seeing all these pointless
.advocacy-style threads in c.u.b.f.m, and would appreciate it if people
would just STOP. Otherwise, I plonk the thread, nuff said.
Followups directed to the bit-bucket, for everyone's good.
--
Lee C. -- Manassas, VA, USA (WakkyMouse on DALnet #watertower)
A! JW223 YWD+++^ri P&B++ SL+++^i GDF B&M KK--i MD+++i P++ I++++ Did
$++ E5/10/70/3c/73ac/95/96 H2 PonPippi Ay77 M | hcremean (at) vt.edu
FreeBSD/Linux/Unix hacker...Win95 and M$ evil! (go see www.freebsd.org)
My home page: http://wakky.dyn.ml.org/~lee | finger me for geek code
TonyP
In article <69n9lo$9l2$1...@tom.pppl.gov>, te...@portsoft.com says...
> In <MPG.f284ccee...@news.idt.net>, anthonp...@mail.idt.net (TonyP) writes:
> >In article <69lqe7$q...@mtinsc02.worldnet.att.net>, fro...@writeme.com
> >says...
> >
> >
> >Tony
> >(Have I been winking too much? My eye hurts!)
>
> And if your a "lamer" you don't have to do anything other than sit here and
> read about it, and make the occassional inflammatory postings.
>
> (Sorry all, I couldn't resist)
>
for your own good. So there. :ţ
Tony