When the only type of tool we have is massively parallel systems, what kind of problems do you think we will apply that tool to, and what kind of problems do you think we will start ignoring? I would rather see us tackle both kinds of problems.
I suspect, however, we will end up with "the contact lens effect", where someone loses their context lens in a dark alley, then looks for it under a streelight "because the light is better over here".
I _really_ fail to understand the rationale for DHCPv6.
IPv6 was designed o that stateless autoconfig resulted in routable addresses.
Combine that with ZEROCONF, and you can discover everything that a DCHP server is going to be able to tell you, and more.
The only technical rationale I've ever heard is for reverse DNS, to prevent someone getting on the local net without authorization and relaying through your SMTP server, but that requires that you configure your DHCP server to only serve to "trusted" MAC addresses. It's also totally useless with DNSUPDAT, since anyone who gets an address can update the reverse in their home domain, and relay out that instead (which is more secure anyway).
So the only rationale I see is controlling access to network dialtone (a business rationale, based on the business model of selling packets rather than selling pipes - a model I happen to disagree with allowing to continue to exist).
It's the little azure ball to the south of the stepped scarlet pyramid of the Eastern Seaboard Fission Authority burning beyond the green cubes of Mitsubishi Bank of America.
I highly recommend using an Ono-Sendai Cyberspace 7 computer deck.
Stay away from Sense/Net if you're a n00b, or you're likely to get iced.
I've had BOINC running for a while without a GUI on my iPhone using the hacker SDK.
Now that they've documented things, the roadblocks are gone from the GUI, and understandable battery and "on external power" notifications will let me know when not to run.
Except it IS documented. He was looking in the header file that documented the API, rather than the one that defined the manifest constant. If he had looked in the one with the manifest constant, he would have seen this comment:
// CoreGraphics deferred updates are disabled if WebKitEnableCoalescedUpdatesPreferenceKey is set // to NO, or has no value. For compatibility with Mac OS X 10.4.6, deferred updates are OFF by // default. #define WebKitEnableDeferredUpdatesPreferenceKey @"WebKitEnableDeferredUpdates"
- Every company thinks they are a great place to work. - Every company thinks they only hire great people. - Every company thinks their problems are interesting and hard. - Every company claims to treat its employees well.
instead, give concrete examples. Are you trying to solve the protein folding problem with something other than the brute force IBM is is trying to use, or are you trying to figure out the best billing algorithm to use on credit cards so you collect the most late payments by skirting banking regulations? Is Paul Vixie on your board of directors? Did you just successfully hire Dennis Ritchie because your ideas are compelling to brilliant people? Would all your employees say that they would enthusiastically have their friends and family come to work for you?
(2) Have smart people working for you already; smart people like to work with smart people
(3) Ask the smart people who already work for you to refer other smart people they know
(4) Poach; if you can't poach smart people away from their current gigs, it's really, really doubtful that you are half as compelling a company as you think you are
BS: Having dealt with Adaptec and Diamond and trying to get them to work with OpenSource drivers, I have to say that most EEs could not write modular software to save their lives.
In the mid 1990s, we were unable to program drivers for the Diamond video cards because some asshat had decided that there was no need to put the PAL dot clock selector input latch tables in a sane location with a recognizable table terminator, because he/she decided that "everyone will be using INT 10 BIOS calls in real mode in order to set video modes on the cards, so why bother?". This was almost as much for hardware protectionism as it was to allow them to swap out PAL contents and ROM contents in tamdem without redesigning hardware - particularly, component layout and printed circuit artwork - to get "whole new video cards". We ended up doing MD5s of the ROMs and setting up our own PAL tables based on what we got as a checksum.
Adaptec did a similar thing on their AHA29xx SCSI controllers, for fear that drivers written for them would would with someone else hardware, just like the BusLogic controllers worked with the AHA152x, 154x, 172x, and 174x card drivers. So they invented an arcane abstraction layer called "HIM", for downloading the SCSI sequencer software to RAM on the card, and relied on the license on the sequencer software to save them from clones. But they screwed up the implementaiton of the HIM, and the OpenSource community ended up having to write their sequencer software to work around the bad abstraction.
So we ended up with a bunch of cards that didn't work in non-Intel hardware (MIPS, DEC Alpha, PPC, etc.), or with non-Microsoft supplied OSs. All because some firmware writers couldn't be bothered to think in terms of interface abstractions, and felt they could write better code than a CS or SWE person (at the time, CE was not an option as a major).
So I call BS on the ability of people who understand the glass understanding how the glass is going to be used by the cusomers that buy it.
For the record:
- I am educated as a physicist, computer scientist [software engineering emphasis], and applied mathematician
- I consider myself a member of the previous generation of embedded software developers, though not my entire career has been embedded.
- I think the next generation of embedded developers are coming places like Taiwan and the poorer countries in Europe, where limited access to modern hardware means that they have to work harder with less, and lack of "a big market" means half of them are off hacking things like iPhones to make them work with the local infrastructure.
The Everest base camps have WiFi; 802.11b, to be specific, so anything that can do 802.11b should be OK.
At that altitude, or anywhere else about 3000m (10,000 ft), you will want something that has solid state storage.
But that's probably not worth worrying about, since cheap liquid crystal typically freezes at about -10C (14F), and the best at about-40C (-40F). It also stops phase-changing at about 80C (176F), so you probably don't have to worry about the deserts.
Can't they respect their own traditions without imposing technologically enforced access controls? What do they do when someone uses hard-copy information, or, to take an example from the article, a man viewing woman's rituals?
What is the point of building an access control system like this?
The other parts are interesting, too, but the part that grabbed me is that this permits large radio and television conglomerates to prop up the ailing print newspaper media, which in the US anyway, is in dire need of propping.
I think it's actually a good thing that they are now allowing the purchases of these companies, which would otherwise go out of business.
As to the Comcast issue - it's not this particular part of the public being ignored - if anything, I'd like to see the cable infrastructure nationalized and leased out to the highest bidder; it would save me on property taxes, too.
The developer community in India is not that large; computer ownership is 14 per 1000 people, which is barely over 1%, compared to them having ~5.2% of the population with cable television [Source: http://blogs.officialexportguide.com/country/].
Add to this the fact that most of the ownership of these machines is centralized, either in large corporations as business equipment, software shops where the employee only has access to work on what the software shop wants worked on, or cyber cafes, where they are not used for development at all, and you get a picture of a country where the resources to support a self-growing Open Source community really aren't there.
For India itself, given that they have ~ 1.6 million computers, and 42,000,000 Internet users [Source: http://www.internetworldstats.com/asia/in.htm], most of these people are using public access facilities, such as cyber cafes.
ASIDE: cyber cafes do not grow up in areas where private computer ownership is prevalent; California, for example, has a total of 11 [Source: http://www.globalcomputing.com/Cafes.html].
So it's unlikely that India will become an Open Source power house any time soon, unless ownership of the means to participate in the community moves into the hands of the individual contributor, without being gated by either hardware availability or employer requirements.
It's far, far more likely to be Western, then Eastern Europe, followed by Asia, which displaces the U.S., if it's to happen any time soon. So much for those who complain about outsourcing...
An intelligently botted computer will shut down when it's told to, go to sleep when it's told to, disconnect from the network when it's told to, and might even make QOS guarantees for other applications on the computer, when they start using lots of network bandwidth.
The last thing a correctly written bot wants to do is show that it's there by doing anything that will draw undue attention to itself.
It may even rate limit its sending of SPAM (if it's a SPAM bot sending SPAM through your ISPs mail server).
If it can use your address book to get at your contacts computers, it will also try that, before turning your machine into a leaf node.
I notice by the relative ratio of the length of the index finger to that of the ring finger that the hand is modelled on a male hand (the index digit is shorter than the ring digit). Have you done any studies on dexterity as to which configuration is better for manipulating objects?
So far MeetingMaker is still top dog; I give it a 2 out of 5 (5 being the best).
There really is no really good group calendaring/scheduling software, period.
All of the iCal based systems suffer from not having notifications, other than active data, and the server based systems (like MeetingMaker) tend to have connectivity problems (this is fixable, but the various vendors never fix it).
Connectivity is more or less required for real time notifications of schedule changes from the central calendar store to the calendar viewing application. Without a real time notification mechanism, high latency mechanisms tend to be used instead (this is the iCal method). But unfortunately, Lucent locked up the very obvious LTAP modifications to LDAP servers very early on with a patent, and since then, everyone has routed around the Lucent damage by adopting "push model" mechanisms with high latency.
The most common iCal push model implementation is via email with active message content. Like OutLook, the email you receive of this type causes code to run on our computer, and when that happens as a result of receiving data, and then the code *acts* on the *content* of the data received, you are just opening yourself up to exactly the same kind of exploits that have plagued OutLook forever.
So the best-of-breed out there right now is a client-server model, aand the least heinous of these is unfortunately still MeetingMaker.
Which is really, really sad, since I first used MeetingMaker some 15 years ago, and it really hasn't gotten much better since then.
This is more about scientists looking at their maps and sending instructions before the 20-40 minute round trip lag between the time the rover says "bing! I am done with the last task!" and the time the scientists hear the "bing!" and start plotting the next target.
You plot the next target and send it 10-20 minutes before the rover can possibly get done with its last target, and there is no lag: the rover just immdiately goes after the new target.
"scientists like to study the surveys taken first before/if they pick a spot to investigate close-up. It takes at least a few hours to do this"
Read Isaac Asimov; in one of his stories, a very remote spacecraft gets into trouble and they are wondering what to tell the astronauts to try, since the communications lag means that they won't get a chance to get an answer for more than two attempts. The mission control director's mother suggests that they gossip: just keep talking at both ends about things to try and results from trying them, until the problem is resolved.
Any scientist who let the rover sit idle after sending orders for investigating something, rather than taking that time to pick the next target and send it to the rover so that when it's done with its previous target, it immediately goes on to the next one, is an idiot, didn't read science fiction as a kid, doesn't understand sliding window protocols, or all three.
No, the problem with Parallels was CPU rendezvous.
They were incorrectly assuming that the CPU broadcast IPI would be handled synchronously, wwn they were hoisting up the OS and installing their hypervisor, so some CPUs would be running under it, and others would not. You can prove this to yourself by setting ncups=1 in the boot arguments (see the MacOS X kernel debugging tutorial on http://developer.apple.com/
According to their web site (if you read it), this was fixed in the last beta release.
Slashdot Mirror
I do NOT look forward to our parallel overlords.
When the only type of tool we have is massively parallel systems, what kind of problems do you think we will apply that tool to, and what kind of problems do you think we will start ignoring? I would rather see us tackle both kinds of problems.
I suspect, however, we will end up with "the contact lens effect", where someone loses their context lens in a dark alley, then looks for it under a streelight "because the light is better over here".
-- Terry
I found it eerily inaccurate.
I threw a bunch of random samplings of English text from Project Gutenberg at it. It claims almost everyone is male.
-- Terry
I _really_ fail to understand the rationale for DHCPv6.
IPv6 was designed o that stateless autoconfig resulted in routable addresses.
Combine that with ZEROCONF, and you can discover everything that a DCHP server is going to be able to tell you, and more.
The only technical rationale I've ever heard is for reverse DNS, to prevent someone getting on the local net without authorization and relaying through your SMTP server, but that requires that you configure your DHCP server to only serve to "trusted" MAC addresses. It's also totally useless with DNSUPDAT, since anyone who gets an address can update the reverse in their home domain, and relay out that instead (which is more secure anyway).
So the only rationale I see is controlling access to network dialtone (a business rationale, based on the business model of selling packets rather than selling pipes - a model I happen to disagree with allowing to continue to exist).
So whose idea was it to turn on DHCPv6?
-- Terry
Finding things in IPv6 Cyberspace...
"So what's the Gregorian music website?"
It's the little azure ball to the south of the stepped scarlet pyramid of the Eastern Seaboard Fission Authority burning beyond the green cubes of Mitsubishi Bank of America.
I highly recommend using an Ono-Sendai Cyberspace 7 computer deck.
Stay away from Sense/Net if you're a n00b, or you're likely to get iced.
-- Terry
I've had BOINC running for a while without a GUI on my iPhone using the hacker SDK.
Now that they've documented things, the roadblocks are gone from the GUI, and understandable battery and "on external power" notifications will let me know when not to run.
Woo hoo!
-- Terry
-- Terry
Can we maybe start referring to them as just "The Department"?
-- Terry
Be a superstar company to work for...
(1) Drop the buzzwords:
- Every company thinks they are a great place to work.
- Every company thinks they only hire great people.
- Every company thinks their problems are interesting and hard.
- Every company claims to treat its employees well.
instead, give concrete examples. Are you trying to solve the protein folding problem with something other than the brute force IBM is is trying to use, or are you trying to figure out the best billing algorithm to use on credit cards so you collect the most late payments by skirting banking regulations? Is Paul Vixie on your board of directors? Did you just successfully hire Dennis Ritchie because your ideas are compelling to brilliant people? Would all your employees say that they would enthusiastically have their friends and family come to work for you?
(2) Have smart people working for you already; smart people like to work with smart people
(3) Ask the smart people who already work for you to refer other smart people they know
(4) Poach; if you can't poach smart people away from their current gigs, it's really, really doubtful that you are half as compelling a company as you think you are
-- Terry
Did they change their name?
I would suggest "Not So Immaculate Mary"... shades of Apple renaming "Sagan" to "BHA"?
-- Terry
BS: Having dealt with Adaptec and Diamond and trying to get them to work with OpenSource drivers, I have to say that most EEs could not write modular software to save their lives.
In the mid 1990s, we were unable to program drivers for the Diamond video cards because some asshat had decided that there was no need to put the PAL dot clock selector input latch tables in a sane location with a recognizable table terminator, because he/she decided that "everyone will be using INT 10 BIOS calls in real mode in order to set video modes on the cards, so why bother?". This was almost as much for hardware protectionism as it was to allow them to swap out PAL contents and ROM contents in tamdem without redesigning hardware - particularly, component layout and printed circuit artwork - to get "whole new video cards". We ended up doing MD5s of the ROMs and setting up our own PAL tables based on what we got as a checksum.
Adaptec did a similar thing on their AHA29xx SCSI controllers, for fear that drivers written for them would would with someone else hardware, just like the BusLogic controllers worked with the AHA152x, 154x, 172x, and 174x card drivers. So they invented an arcane abstraction layer called "HIM", for downloading the SCSI sequencer software to RAM on the card, and relied on the license on the sequencer software to save them from clones. But they screwed up the implementaiton of the HIM, and the OpenSource community ended up having to write their sequencer software to work around the bad abstraction.
So we ended up with a bunch of cards that didn't work in non-Intel hardware (MIPS, DEC Alpha, PPC, etc.), or with non-Microsoft supplied OSs. All because some firmware writers couldn't be bothered to think in terms of interface abstractions, and felt they could write better code than a CS or SWE person (at the time, CE was not an option as a major).
So I call BS on the ability of people who understand the glass understanding how the glass is going to be used by the cusomers that buy it.
For the record:
- I am educated as a physicist, computer scientist [software engineering emphasis], and applied mathematician
- I consider myself a member of the previous generation of embedded software developers, though not my entire career has been embedded.
- I think the next generation of embedded developers are coming places like Taiwan and the poorer countries in Europe, where limited access to modern hardware means that they have to work harder with less, and lack of "a big market" means half of them are off hacking things like iPhones to make them work with the local infrastructure.
-- Terry
I would be happy to do this... if:
- I could opt out on a per machine basis
- I was allowed to charge them rent on the disk space; I have decided that rent is $5000/month.
If I have to pay for hosting services for my software, why shouldn't they for theirs?
-- Terry
WiFi, Solid state, liquid crystal
The Everest base camps have WiFi; 802.11b, to be specific, so anything that can do 802.11b should be OK.
At that altitude, or anywhere else about 3000m (10,000 ft), you will want something that has solid state storage.
But that's probably not worth worrying about, since cheap liquid crystal typically freezes at about -10C (14F), and the best at about-40C (-40F). It also stops phase-changing at about 80C (176F), so you probably don't have to worry about the deserts.
-- Terry
Given your comment, I'm wondering...
Can't they respect their own traditions without imposing technologically enforced access controls? What do they do when someone uses hard-copy information, or, to take an example from the article, a man viewing woman's rituals?
What is the point of building an access control system like this?
-- Terry
"I've heard the opposite- that slot-load drives are bad for schools because kids like to stick things in them."
And I've heard that what they stick in the slots is pieces of the trays they snap off from other machines that have (had?) tray loading drives.
-- Terry
This could save print journalism.
The other parts are interesting, too, but the part that grabbed me is that this permits large radio and television conglomerates to prop up the ailing print newspaper media, which in the US anyway, is in dire need of propping.
I think it's actually a good thing that they are now allowing the purchases of these companies, which would otherwise go out of business.
As to the Comcast issue - it's not this particular part of the public being ignored - if anything, I'd like to see the cable infrastructure nationalized and leased out to the highest bidder; it would save me on property taxes, too.
-- Terry
I tried looking for that reference...
But the googles, they do nothing!
-- Terry
This is unlikely, near term.
The developer community in India is not that large; computer ownership is 14 per 1000 people, which is barely over 1%, compared to them having ~5.2% of the population with cable television [Source: http://blogs.officialexportguide.com/country/].
Add to this the fact that most of the ownership of these machines is centralized, either in large corporations as business equipment, software shops where the employee only has access to work on what the software shop wants worked on, or cyber cafes, where they are not used for development at all, and you get a picture of a country where the resources to support a self-growing Open Source community really aren't there.
For India itself, given that they have ~ 1.6 million computers, and 42,000,000 Internet users [Source: http://www.internetworldstats.com/asia/in.htm], most of these people are using public access facilities, such as cyber cafes.
ASIDE: cyber cafes do not grow up in areas where private computer ownership is prevalent; California, for example, has a total of 11 [Source: http://www.globalcomputing.com/Cafes.html].
So it's unlikely that India will become an Open Source power house any time soon, unless ownership of the means to participate in the community moves into the hands of the individual contributor, without being gated by either hardware availability or employer requirements.
It's far, far more likely to be Western, then Eastern Europe, followed by Asia, which displaces the U.S., if it's to happen any time soon. So much for those who complain about outsourcing...
-- Terry
You are talking about the badly written ones.
An intelligently botted computer will shut down when it's told to, go to sleep when it's told to, disconnect from the network when it's told to, and might even make QOS guarantees for other applications on the computer, when they start using lots of network bandwidth.
The last thing a correctly written bot wants to do is show that it's there by doing anything that will draw undue attention to itself.
It may even rate limit its sending of SPAM (if it's a SPAM bot sending SPAM through your ISPs mail server).
If it can use your address book to get at your contacts computers, it will also try that, before turning your machine into a leaf node.
These are not the early days of botnets.
-- Terry
Interesting about the shadowrobot.com hand...
I notice by the relative ratio of the length of the index finger to that of the ring finger that the hand is modelled on a male hand (the index digit is shorter than the ring digit). Have you done any studies on dexterity as to which configuration is better for manipulating objects?
Thanks,
-- Terry
So far MeetingMaker is still top dog; I give it a 2 out of 5 (5 being the best).
There really is no really good group calendaring/scheduling software, period.
All of the iCal based systems suffer from not having notifications, other than active data, and the server based systems (like MeetingMaker) tend to have connectivity problems (this is fixable, but the various vendors never fix it).
Connectivity is more or less required for real time notifications of schedule changes from the central calendar store to the calendar viewing application. Without a real time notification mechanism, high latency mechanisms tend to be used instead (this is the iCal method). But unfortunately, Lucent locked up the very obvious LTAP modifications to LDAP servers very early on with a patent, and since then, everyone has routed around the Lucent damage by adopting "push model" mechanisms with high latency.
The most common iCal push model implementation is via email with active message content. Like OutLook, the email you receive of this type causes code to run on our computer, and when that happens as a result of receiving data, and then the code *acts* on the *content* of the data received, you are just opening yourself up to exactly the same kind of exploits that have plagued OutLook forever.
So the best-of-breed out there right now is a client-server model, aand the least heinous of these is unfortunately still MeetingMaker.
Which is really, really sad, since I first used MeetingMaker some 15 years ago, and it really hasn't gotten much better since then.
-- Terry
This is more about scientists looking at their maps and sending instructions before the 20-40 minute round trip lag between the time the rover says "bing! I am done with the last task!" and the time the scientists hear the "bing!" and start plotting the next target.
You plot the next target and send it 10-20 minutes before the rover can possibly get done with its last target, and there is no lag: the rover just immdiately goes after the new target.
-- Terry
Communications lag is irrelevant.
"scientists like to study the surveys taken first before/if they pick a spot to investigate close-up. It takes at least a few hours to do this"
Read Isaac Asimov; in one of his stories, a very remote spacecraft gets into trouble and they are wondering what to tell the astronauts to try, since the communications lag means that they won't get a chance to get an answer for more than two attempts. The mission control director's mother suggests that they gossip: just keep talking at both ends about things to try and results from trying them, until the problem is resolved.
Any scientist who let the rover sit idle after sending orders for investigating something, rather than taking that time to pick the next target and send it to the rover so that when it's done with its previous target, it immediately goes on to the next one, is an idiot, didn't read science fiction as a kid, doesn't understand sliding window protocols, or all three.
-- Terry
"No business case [for municipal WiFi]..."
It seems to me that there is no business case for public parks, either.
Not everything has to be about turning a profit for someone.
-- Terry
The next time... ...the car bombs which don't blow up will be in larger SUVs, and will have scary faces painted on the front of them.
-- Terry
No, the problem with Parallels was CPU rendezvous.
They were incorrectly assuming that the CPU broadcast IPI would be handled synchronously, wwn they were hoisting up the OS and installing their hypervisor, so some CPUs would be running under it, and others would not. You can prove this to yourself by setting ncups=1 in the boot arguments (see the MacOS X kernel debugging tutorial on http://developer.apple.com/
According to their web site (if you read it), this was fixed in the last beta release.
-- Terry