* Kragen Javier Sitaker <kragen at canonical.org> [2009-01-23 01:15]:
> Some bastard pretended to be me in order to get the message
> through to the mailing list.

I’ve seen this mail in several places now. It always spoofs
a real From. So it’s quite possible that it was not targetted
specifically at your mailing list but just happened to have a
matching From and To by freak accident.

Regards,
-- 
Aristotle Pagaltzis // <http://plasmasturm.org/>

* Kragen Javier Sitaker <kragen at canonical.org> [2009-01-23 01:15]:
> As should be apparent from the formatting, content, and trailer
> of this message, I did not send it

This is why routine PGP/GPG signing is a good idea – “if I had
sent it it would be signed.”

(Yes, I know – I’m not doing that myself either. Always thought I
should have, but am yet to get around to it…)

Regards,
-- 
Aristotle Pagaltzis // <http://plasmasturm.org/>

On Thu, Jan 22, 2009 at 09:42:08PM +0100, Kragen Sitaker supposedly wrote:
> If you want to know who the real establishment is in America and
> around the world, the real power behind the so-called
> ''military-industrial complex'', the real maleficent power that has
> led this world to inexorable conflict, war, hatreds, destruction of
> real human values, morality, conscience ? it is the global media
> establishment.
> 
> This same media witnessed the Zionist slaughter ...

As should be apparent from the formatting, content, and trailer of this
message, I did not send it; some bastard pretended to be me in order to
get the message through to the mailing list.  I'm sorry about that.
Mailman doesn't support authentication of where messages come from, as
far as I know, so there isn't a strong way to prevent this from
happening.

Kragen

If you want to know who the real establishment is in America and around the world, the real power behind the so-called ''military-industrial complex'', the real maleficent power that has led this world to inexorable conflict, war, hatreds, destruction of real human values, morality, conscience — it is the global media establishment.

This same media witnessed the Zionist slaughter of 1,300 people in Gaza, and the maiming of 6,000 more, half of them women and children. More than 20,000 homes and buildings were destroyed or damaged and thousands became homeless. This horrendous mass slaughter of Palestinians is compared to the loss of 3 Israeli
civilians. Yet, the media never dare to call this slaughter exactly what it is: terrorism on a horrendous scale! This same American and global media has not informed the American people or the people of the world of the damning truth about Barack Obama.

In truth, Obama is a complete creation of extremist Jewish Zionists and he has already begun to serve Israel by his top appointments.

The Jewish-dominated media in America is promoting the Obama inauguration as akin to the Second Coming of Jesus Christ. (If you question Jewish control of the media read Who Runs the Media? and you will find documentation proving the Jewish control being far greater than you might suspect) The rest of the world’s
powerful media (Which also has a powerful Jewish presence in the UK, France, Spain, Sweden, Russia, and the rest of Europe) takes its cues from the NY/Hollywood media axis, and presents Obama as kind of a superhuman agent of real change.

In fact, it is a Huge Lie!

Obama is completely controlled by the same forces that have controlled George Bush, so much so that The Chicago Tribune quotes leading Jews bragging that Obama is so in the hands of the Zionists that he is ''the first Jewish President.'' Obama received about 80 percent of the Jewish vote in the United States!

The Three critical things that made Obama President:

1)Jewish political influence

Obama’s chief handler is David Axelrod, a radical Jewish Zionist whose previous claim to fame was the fact that he was the Zionist political hit man against Illinois Senator Charles Percy, who was not deemed to be pro-Israel enough. (He was only 99.9 percent Israel’s lackey, not the required 100 percent) Obama went to Israel
and made sure he pronounced himself even more radically pro-Zionist than Bush or McCain.

2) Massive amounts of Jewish money

Obama’s campaign was kicked in high gear by Jewish Hollywood and financial moguls. In just one of the early fundraising parties in Hollywood by Spielberg, Geffen and Katzenburg it raised over a million dollars for Obama. Obama raised more money than any political candidate in American history and his biggest contributors
were overwhelmingly Jewish activists, Jewish international finance and banking firms and Jewish globalists. His largest single contributor was the international Jewish Banking firm of Goldman Sachs. He received more money from the same international bankers that ripped off trillions of dollars in the recent economic scandals
than even John McCain. Is a man totally under the power of the most powerful financial organizations in the entire world, really an agent of ''change?''

3) Overwhelming positive support in the Jewish-dominated media

By a large margin, Jewish-owned media officially endorsed Obama, and that is not counting tens of thousands of positive articles by Jewish; owned publications and pundits.

Obama’s first act as President-elect was to appoint a Jewish extremist, dual citizen, Rahm Emmanuel, as his chief of staff. As the people of Gaza were slaughtered, Obama would not make a single statement to stop this murder and maiming of thousands of innocents!

Israel, very carefully timed its terrorist attack on Gaza to be in the remaining days of the Bush Administration. The day before Obama’s inauguration Israel announced that it would be completely out of Gaza by the time Obama took his oath of office.

Why?

Before Obama took office, Israel could make this terrorist slaughter against Palestinians and Obama would still be perceived as having clean hands. Because Obama is completely under their control, they want him to have an image of fairness, honor and peacefulness, and as representing a new direction of American policy, as
he begins to deal with the Mideast turmoil.

Since he is Israel’s boy in the White House, what better scenario could they have than a President perceived as practically the Second Coming, of high moral conviction, and dedicated to fairness,  but who is actually bound hand-and-foot to the Zionist agenda, just as the last president’s have been. Talk about a perfect shill.
And the game is working, for even many Palestinians are filled with hope that the new President will work to end their long suffering.

With an almost godlike positive image around the world, a Barack Obama who is in reality controlled by Israel, is a much bigger danger than was even George Bush. We know what Bush is, but too many Americans and other peoples around the world are falling for the lies about the new ''savior'' of the world: Barack Obama. He
already says we are to put thousands of new troops in Afghanistan and has rattled the sabers against Iran. Obama with highest positive image (created by media) in the world is in a more dangerous position to lead us into catastrophic wars than George Bush ever was.

Jewish screenwriters in Hollywood couldn’t have written a better script for the Zionist agenda than what is being played out on inauguration day, USA.

Of course, why should they write it, they did it.

God save America, God save the Palestinians, God save peace and justice in the world!

–david duke

Source:

 http://www.davidduke.com/general/wake-up-world-meet-the-new-boss-same-as-the-old-boss_7225.html

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


You or someone using your email adress is currently subscribed to the Lawrence Auster
Newletter. If you wish to unsubscribe from our mailing list, please let us know by calling to 1 212 865 1284

Thanks,

Lawrence Auster,
238 W 101 St Apt. 3B
New York, NY  10025
Contact: lawrence.auster at att.net


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

I hassled around with trying to get it to install via PXE (enabling PXE
boot on the Asus board required that I go into BIOS setup
and enable the Ethernet option BIOS, in BIOS setup → Advanced →
Onboard Devices Configuration → Onboard PCIE GbE LAN → LAN Option ROM)
but then I gave up in favor of the following approach instead.

Installing Ubuntu Intreped Ibex on my ASUS P5KPL-AM with UNetbootIn
===================================================================

(This document is at <http://canonical.org/~kragen/unetbootin.html>.)

I downloaded unetbootin_304-4_i386.deb from SourceForge and installed
it on my Debian Etch laptop with `dpkg -i`, which required that I
`apt-get install libqt4-gui libqt4-core` first.  Then I ran
`unetbootin` to create a boot-install image on my USB key, selecting
“Ubuntu → 8.10_NetInstall_x64”.

I’ve [previously described the hassles of booting from USB on the P5KPL-AM]
(http://lists.canonical.org/pipermail/kragen-journal/2009-January/000563.html)
but one thing I hadn’t realized was that every time you power-cycle
the machine or boot it without the USB key, it forgets that you wanted
to make it “Force FDD” on the USB device.

The USB key thus created doesn’t actually boot successfully.  Each
time the boot countdown finished, it would restart.  I hit Esc to get
to the `boot:` prompt and typed `unetbootindefault` and hit Enter;
this explained that the problem was, “Could not find kernel image:
/ubnkern”.  Other people have successfully booted from USB and done
installs with this version of `unetbootin`; I don’t know why things
are different on my machine.

I changed the `syslinux.cfg` file to be much shorter:

    default ubnkern
    append initrd=ubninit

That got it to boot the kernel and start the install.  Unfortunately
the AMD64 (“x64”) boot image fails to get DHCP (or ping the default
gateway when manually configured), but the `i686` version works fine,
except that it is going to install a 32-bit operating system.
(`dmesg` says it’s using the `r8169` driver, which I’m pretty sure is
what the 64-bit one is using too.  The 64-bit kernel can detect the
link going up and down; it just can’t pass packets.)

Fixing GRUB
-----------

After the install finished, the disk would fail to boot; I just saw a
blinking underscore cursor on a black screen when I tried to boot it.

I booted again from the USB stick, and after it got to the point where
it had figured out that I had a hard disk, I went to the second
virtual console (Alt-F2) and hit Enter to get a prompt.  Then I did
something like the following (from memory):

    mkdir /foo
    mount -t ext3 /dev/sdb1 /foo
    mount -t proc none /foo/proc
    mount -t sysfs none /foo/sys
    chroot /foo /bin/bash
    /etc/init.d/udev start
    grub-install /dev/sdb

At this point I noticed that `grub-install` was saying, `Installing
GRUB to /dev/sdb as (hd1)...`, but that was wrong.  The disk will be
the first one in the system when GRUB tries to boot off of it, so it
should be `(hd0)`, not `(hd1)`.

So I edited `/boot/grub/device.map`, which said

    (hd0)	/dev/sda
    (hd1)	/dev/sdb

so that it said the opposite:

    (hd0)	/dev/sdb
    (hd1)	/dev/sda

and ran `grub-install /dev/sdb` again.  This time it said, `Installing
GRUB to /dev/sdb as (hd0)...`, and thereafter it worked correctly.

<link rel="stylesheet" href="http://canonical.org/~kragen/style.css" />

I decided I'd better get a new machine; my current laptop ("thrifty") is
10 years old.  So I bought an Asus P5KPL-AM motherboard, an Intel
Celeron E1200 1.6GHz dual-core processor (really a slow Core Duo with
512kB of cache), a 2GB DIMM of Kingston DDR2-800 RAM, a cheap-junk power
supply and USB keyboard, a 500GB SATA disk, and a 1280×1024 LCD monitor.
No case yet.  At this point the total cost is roughly US$400.

So far I haven't been able to get anything installed on it due to lack
of suitable install media.  Here is a list the significant mistakes I
know I have made so far:

- ATX 12V power supplies require that you connect *two* power cables to
  the motherboard, the big wide one and the little 12V square one.  Just
  one won't make it boot or even POST.

- I spent far, far too long actually buying the thing, in parts.
  Assembling the parts was easy and fun (aside from the problem above),
  but comparison-shopping for them really sucked.

- I probably should have just bought a CD-ROM drive instead of trying to
  do the install from a USB key.

- I tried to use various automated programs to make a bootable USB key
  instead of doing it myself.  Given a USB key already partitioned with
  a single FAT16 partition, you just need to `syslinux /dev/sda1` and
  copy an Ubuntu `vmlinuz` and `initrd.gz` onto it and put this in
  `syslinux.cfg`:

        default vmlinuz
        append initrd=initrd.gz

  However, I don't currently have a working flash drive large enough to
  load an Ubuntu install CD image from, and for whatever reason, the
  Ubuntu `initrd.gz` doesn't automatically load the Debian Etch netinst
  ISO I put on the flash drive.

Here are some obstacles I've run into:

- The 2GB Kingston/Trend Micro flash key I bought for this purpose for
  AR$49 (US$14) doesn't actually work.  I'm currently using Beatrice's
  256MB flash key.

- The `liveusb` package doesn't work because of a variety of bugs; as
  mentioned before, I should have just done it myself.

- However, it took me a while to find that out, because there are no
  tarballs of `liveusb`.  You have to check it out from source control
  using `bzr`.

- The version of `bzr` in Debian Etch is too old to access the `liveusb`
  repository on Launchpad, so I installed `bzr` from backports.

- But my backports GPG ring was out of date, so I had to update it.

- Thrifty (a Thinkpad T20) won't get past POST with either of my CD-ROM
  drives plugged in.  This is a new feature in the last few months.

- But you can't plug a CD-ROM drive into it once it's already booted;
  Linux won't notice it.

- You can plug a CD-ROM drive into it after it's out of the BIOS but
  before Linux has scanned for IDE devices, and that mostly works.

- Except that both of the CD-ROM drives I have for it can't reliably
  read the Ubuntu 8.10 CD that Beatrice burned on her Mac, probably
  because they're 10 years old this year.  So there was really no point
  in getting one of them to work.

- The AMI BIOS on the "mother" (as they call it here) doesn't really
  have a "boot from USB" option in the setup.  You have to press F8 at
  boot time to get a list of possible boot devices including the USB
  stick.

- But first you have to configure the BIOS to boot from it as if it were
  a floppy drive: press DEL to get into setup, and in the Advanced tab,
  under "‣ USB Configuration", under "‣ USB Mass Storage Device
  Configuration", set the "Emulation Type" to "Forced FDD".  Neither
  "Floppy" nor "Hard Disk" nor "Auto" works; all result in the error
  message "Missing operating system".  "CDROM", the other option, just
  results in hard system hangs during boot.

- The Ubuntu `initrd.gz` doesn't automatically load the Debian Etch
  netinst ISO I put on the flash drive.  I was able to mount it myself
  with commands like the following:

        modprobe vfat
	mkdir /mnt
	mount -t vfat /dev/sdb1 /mnt
	modprobe loop
	losetup -f /mnt/debian-40r6-amd64-netinst.iso
	mkdir /cdrom
	modprobe iso9660
	mount -t iso9660 /dev/loop0 /cdrom
	cp /cdrom/install.amd/initrd.gz /tmp
	cd /tmp
	gzip -d initrd.gz
	cpio -i < initrd
	chroot . bin/ash

  I'm not sure if all of that ceremony was needed, but it did work.
  Unfortunately I don't know enough about how the Debian netinst disk
  works to figure out how to run it from here.

Dave Long <dave.long at bluewin.ch> writes:
> A design that I'd like to try sometime would be to try an "Elephant"- 
> like language, in which all values would, in principle, be  
> recoverable from a logfile.  This could eliminate GC, as frequently  
> used values would tend to be instantiated in a cache, and values that  
> are never referenced again would cost just their portion of storage,  
> presumably amortized by bulk sequential i/o patterns.

This is a fascinating connection!  

(For the readers who don’t know what Dave’s talking about, Elephant
was John McCarthy’s recent proposal for a really different kind of
programming system; as I recall, the basic idea is that you specify
the kinds of promises your program has to keep, and the system figures
out how to keep those promises based on a complete log of the past I/O
behavior of your program.)

I’ve been thinking a bit over the years about logfile-based
programming as well; the “rumor-oriented programming” post was about
it.  (The basic theory is that if you write your application to not be
sensitive to the order of items in the logfile and if there’s no
mutual-exclusion validity criterion for logfiles, you automatically
get an application-agnostic synchronization protocol, and you forever
avoid the problem of database schema migration.)  My theory was that
you could define the objects of interest to your application in terms
of queries on the history, in a FRP-like way.  (Although I was
thinking SQL, not Haskell.)

But the connection between GC and paging sounds really fruitful.
Suppose you chose to eliminate GC by telling the OS that the pages you
allocated for results derived from history didn’t need to be preserved
at all; if the OS wanted to evict them, it could do so merely by
deleting them from your mapping and zeroing them for another process
to use.  Now if you try to access them, you get a segfault, and the
segfault handler can put the requesting thread to sleep and enqueue a
query to recompute the data that it had wanted.  So you rely on the
OS’s paging mechanism to manage the caching of your computational
results.  Is that more or less what you have in mind?

This way, the usual overhead of checking weak-reference validity, or
sending weak-reference notifications when finalizing a
weakly-referenced object, gets shoved off on the MMU hardware.

It seems like it might have poor space efficiency: you might end up
with 100 pages that stay alive because they contain 150
frequently-used objects, each of which is only, say, 32 bytes, while
the rest of those pages is taken up by dead objects.

In the context of parsing, I was thinking that references from the
stack (from function arguments or partly evaluated expressions) would
be strong references that would promote objects to older generations;
does that generalize to the paging scheme?  You could perhaps allocate
space for the values of subexpressions on the stack so they don’t get
deallocated before you use them, then move them to the heap once they
were no longer part of a live subexpression or function argument, but
it seems like somewhere you need to store away the context you need to
recreate those results if you need them again after they get recycled.
In the PEG-parsing case, that information is just a position in the
source text and a reference to a nonterminal.  (The source text and
nonterminals don’t need to be garbage-collectable.)  But in an
Elephant-like system it seems that your query could be some
arbitrarily complex expression.  Can you guarantee not needing GC for
the query too?

Alternatively you could realize your Elephant idea using a
generational GC in the same way I was suggesting for caching of PEG
results, with a more traditional implementation of strong and weak
references.

> Any idea how mutation rates compare for various workloads with main  
> memory and/or disk bandwidths?

I don’t even know where to begin.  Maybe some of Ben Zorn’s
quantitative research on garbage collection and allocator performance
might come close?

Kragen

> One approach to implementing this, which I've also been thinking of  
> for
> maintaining Bicicleta intermediate results, would be to use a value- 
> weak
> hashtable for the memoization table, along with a garbage collector.
> Any value that is needed again while it's still in the nursery will  
> not
> need to be recomputed, but most values will be reaped because at the
> time of the next minor collection, the only reference to them will  
> be in
> the value-weak hashtable.  A few won't be --- the ones whose values  
> are
> currently referenced from some parsing rule currently being tried ---
> and those will be promoted.
> ...
> I think this design may be useful for laziness in general, but I  
> haven't
> tried it yet, despite having come up with it a year or two ago.

A design that I'd like to try sometime would be to try an "Elephant"- 
like language, in which all values would, in principle, be  
recoverable from a logfile.  This could eliminate GC, as frequently  
used values would tend to be instantiated in a cache, and values that  
are never referenced again would cost just their portion of storage,  
presumably amortized by bulk sequential i/o patterns.

Any idea how mutation rates compare for various workloads with main  
memory and/or disk bandwidths?

-Dave