In this particular case, the anouncement had already
been published.
Going off-topic (relative to the powerfull) I might
compare it to the ongoing debate about full disclosure and
proof-of-concept exploits. I will be interested to see
how it turns out...
IED is short for improvided explosive device, such as
a large quantity of commerial explosive in a 3" diameter
iron pipe, with the detonator connected to a cell-phone
ringer circuit.
I loved the chips and the techies, but the legalists
at SGI/Mips were startingly evil.
An unnamed former employer found their contract
contained a clause that said the purchaser indemnified
the vendor in the event of a patent suit brought in Canadian court...
--dave
I used to fence, and fencing shoes run the $50-$80.00 range,
despite being almost a one-off product. Nevertheless, these
relatively cheap shoes eliminated a severe pain.
Conclusion: good mass-market shoes sould cost less than $50.00
My old team at Sun used to port everything from
soup to nuts to Solaris, Linux and later versions of
middleware, so that sounds like an opportunity for
some fixed-price ports (;-))
And the dongle approach has failed three times: once with
Apple ][, again with CP/M and once more with DOS.
Only a few rare examples still exist in the wild:
everyone else, including my former employers, found
it was so expensive and worked so badly it was
more expensive than the projected loss
from merely-copyrighted softeware.
Sorry, I was being ambiguous: I meant that the position
that Theo and RMC always espouse was a reason to
create alternate-firmware and similar projects, not that
their comments about this project motivated it.
Indeed, the tone of their comments sometimes demotivates me (;-))
RMS and Theo are good folks to have, to keep us from
wandering down a blind alley. In the case of
OLPC, their position has caused the launch of a subproject
to create free replacements for the proprietary bits.
At the same time, those replacements don't exist yet, and
OLPC is constrained (by power and mesh-networking issues) to use the proprietary
bits in the meantime, to be able to ship product.
Sounds to me like a good plan: they know they want both
laptops and free software, so they're working on a plan
to have both. Which is a very healthy approach!
They experimentally released "Using Samba" under a free
license, and found to their surprise that despite
bing about the fourth book into that market,
it jumped off the shelf.
People were using it as an on-line reference and
printing snippets to carry around. Printing the whole
book was inconvenient, so people went out and bought copies.
And if they were talking to my old system at Ork,
they'd be getting a (possibly false) positive on every
verification that was syntactically valid.
Not A Good Algorithm (;-))
They are probably doing a VRFY <person@place> on another
connection, which is part of the RFCs, but which is
often turned off by MTAs.
Back in the dusty beginnings of time when I was working on
mail,
the same MTAs that turned off VRFY always said they'd accept
RCPT To:<person@place>, so as to be able to avoid any microsecond-consuming checking (;-))
Sounds like a trade-auditing project I was once on.
If the 10,000 trades are easily broken into small groups,
such as by the initial letter of the ticker symbol, and
if all the data for the analysis is fetched in the first
step, you can in fact spread the processing over 26-odd machines
for a speedup of (fixed part + (per-ticker-symbol part/26)).
I have an article on doing the load-balancing part of this
kind of processing, albeit on a large multiprocessor, at
http://www.sun.com/blueprints/0605/819-2888.pdf[In PDF].
As you've already guessed, sometimes the problem doesn't decompose
nicely into parts that can be distributed to machines
far from the database.
The rule of the thumb is that grid does distributed computation,
where you ship small amounts of data to many CPUs. If you have
large amounts of data, you need to have previously distributed
data stores, and then you ship the processing to reside with it,
instead of the other way around. Alas, some folks call the latter grid,
when it should be called something like "data grid" (;-))
We used to something of the sort with background and batch jobs in Multics:
DRBrown.absentee was my background instance, and I could
grant it more or fewer permissions if I felt like it.
In Unix, one could have a davecb.browser role
with only the right to read web pages, and a davecb.email
that could connect to pop, imap and smtp, and read/write
$HOME/Mail, but nothing else. I suspect the latter could
be built fairly easily in selinux...
My personal and work SCSI disks gave superior service, with
3 year on-site guarantee and 5 years return-to-depot.
We used the on-site replacement just once, out of many many disks, about a week
after purchase.
My personal backup drive was replaced last year,
at (or beyond) the end of the 5-year warranty period, which wasn't
unusual. The replacements were new, not refurb, and have the same warranty
starting with their "purchase" date.
Slashdot Mirror
In this particular case, the anouncement had already been published.
Going off-topic (relative to the powerfull) I might compare it to the ongoing debate about full disclosure and proof-of-concept exploits. I will be interested to see how it turns out...
--daveAnd so a corollary is that any security researcher who exposes a risk or danger is a criminal (;-))
--dave
IED is short for improvided explosive device, such as a large quantity of commerial explosive in a 3" diameter iron pipe, with the detonator connected to a cell-phone ringer circuit.
--dave
The article is intended to sell magazines, and for that purpose it is well crafted.
- To someone who doesn't know Stallman, it's misleading.
- To someone who does know him and dislikes him, it's justification.
- To someone who does know him and likes him, its a troll.
All of the above sell dead trees. All of the above are evil.--dave
I loved the chips and the techies, but the legalists at SGI/Mips were startingly evil.
An unnamed former employer found their contract contained a clause that said the purchaser indemnified the vendor in the event of a patent suit brought in Canadian court...
--dave
I used to fence, and fencing shoes run the $50-$80.00 range, despite being almost a one-off product. Nevertheless, these relatively cheap shoes eliminated a severe pain.
Conclusion: good mass-market shoes sould cost less than $50.00
--dave
--dave
My old team at Sun used to port everything from soup to nuts to Solaris, Linux and later versions of middleware, so that sounds like an opportunity for some fixed-price ports (;-))
--dave
Good point, but build the intelligence at the network layer wherever you can, so that if smtp server 26 goes down, no-one will even notice (;-))
-dave (my sysadmin once said "clusters are less reliable than uniprocessors") c-b
And the dongle approach has failed three times: once with Apple ][, again with CP/M and once more with DOS.
Only a few rare examples still exist in the wild: everyone else, including my former employers, found it was so expensive and worked so badly it was more expensive than the projected loss from merely-copyrighted softeware.
--dave
Many machine rooms and all houses are designed for fewer pieces of hot equipment than are common now, which has repeatedly led me to
The large-room trick also means that if you use a local air conditioner, it doesn't have to be turned down to sub-zero (;-))
--dave
Only if they already know how to avoid accidentaly accepting the jurisdiction of a U.S. court under .S. law.
One hopes their Canadian lawyers are following this and tell their U.S. lawyers to watch out (;-))
--dave
Sorry, I was being ambiguous: I meant that the position that Theo and RMC always espouse was a reason to create alternate-firmware and similar projects, not that their comments about this project motivated it.
Indeed, the tone of their comments sometimes demotivates me (;-))
--dave
RMS and Theo are good folks to have, to keep us from wandering down a blind alley. In the case of OLPC, their position has caused the launch of a subproject to create free replacements for the proprietary bits.
At the same time, those replacements don't exist yet, and OLPC is constrained (by power and mesh-networking issues) to use the proprietary bits in the meantime, to be able to ship product.
Sounds to me like a good plan: they know they want both laptops and free software, so they're working on a plan to have both. Which is a very healthy approach!
--dave
They experimentally released "Using Samba" under a free license, and found to their surprise that despite bing about the fourth book into that market, it jumped off the shelf.
People were using it as an on-line reference and printing snippets to carry around. Printing the whole book was inconvenient, so people went out and bought copies.
Which pleased me immensely!
--dave
At least the Canadian federal and provincial governments love them
--dave
And if they were talking to my old system at Ork, they'd be getting a (possibly false) positive on every verification that was syntactically valid. Not A Good Algorithm (;-))
--dave
Thanks, kind sir (;-))
Yup: if you could do it at run-time, you'd use a bin-packing algorithm to create N equally-sized buckets of trades (;-))
--dave
They are probably doing a VRFY <person@place> on another connection, which is part of the RFCs, but which is often turned off by MTAs.
Back in the dusty beginnings of time when I was working on mail, the same MTAs that turned off VRFY always said they'd accept RCPT To:<person@place>, so as to be able to avoid any microsecond-consuming checking (;-))
--dave
Sounds like a trade-auditing project I was once on.
If the 10,000 trades are easily broken into small groups, such as by the initial letter of the ticker symbol, and if all the data for the analysis is fetched in the first step, you can in fact spread the processing over 26-odd machines for a speedup of (fixed part + (per-ticker-symbol part/26)).
I have an article on doing the load-balancing part of this kind of processing, albeit on a large multiprocessor, at http://www.sun.com/blueprints/0605/819-2888.pdf[In PDF].
As you've already guessed, sometimes the problem doesn't decompose
nicely into parts that can be distributed to machines
far from the database.
The rule of the thumb is that grid does distributed computation, where you ship small amounts of data to many CPUs. If you have large amounts of data, you need to have previously distributed data stores, and then you ship the processing to reside with it, instead of the other way around. Alas, some folks call the latter grid, when it should be called something like "data grid" (;-))
--dave
It's commonly called "disintermdiation", by the way... see http://en.wikipedia.org/wiki/Disintermediation
--dave
We used to something of the sort with background and batch jobs in Multics: DRBrown.absentee was my background instance, and I could grant it more or fewer permissions if I felt like it.
In Unix, one could have a davecb.browser role with only the right to read web pages, and a davecb.email that could connect to pop, imap and smtp, and read/write $HOME/Mail, but nothing else. I suspect the latter could be built fairly easily in selinux...
--dave
If the editors are good, they find the things I and others are interested in, and filter out the dreck. Their magazines are worthwhile.
If they'e not, all you're paying for is a bundle of ads (;-))
--dave
My personal and work SCSI disks gave superior service, with 3 year on-site guarantee and 5 years return-to-depot.
We used the on-site replacement just once, out of many many disks, about a week after purchase. My personal backup drive was replaced last year, at (or beyond) the end of the 5-year warranty period, which wasn't unusual. The replacements were new, not refurb, and have the same warranty starting with their "purchase" date.
--dave