Anyways, writing zeros, or writing something else sequentially should essentially be the same. More dumb Flash based SSDs actually respond to writing zeros, so I just remembered "here you can do the same zeroing trick".
You just need a background task that zeros blocks used by deleted files. EasyCo's MFT implements such a log filesystem as a generic block device for x86 Linux and Windows.
Andy Tanenbaum has a nice story about that. On a conference they wanted to use a modulated laser to beam an internet connection to another building. Except they calibrated the target at night, it tested fine. But during the day the air would warm and break the connection. He found it rather odd they didn't just use a cable the next day, but the management insisted on using the broken technology. Didn't work they entire conference.
Troll? What? It's just a technical question. Just like I would want to see a working Xerox Alto, I might want see an "Archimedes" when it is anything special.
Hmm, I dropped an 's there. Should have been "[..] than the servers they [..]". The rate limiting is good point yes, but their unicornscan tool is for scanning large amounts of server (I think so..).
They -at Outpost24- are basically doing TCP/IP without the per connection timeout timers. By keeping the state in a SYN-cookie-like hashtable and drinking directly from the firehose (network). This enables their scan program to work faster and with less resources than the server they are connecting to. As far as I get it, this part is already publicly available in their Unicornscan program.
Furthermore they discovered that *after* you have done the three way TCP/IP handshake, the OSes are less paranoia than is good for them. There seem to be positions in the statespace in which the TCP/IP stack will keep resending the answer packet. That keeps a timer slot busy, a resource which can be exhausted. Apparently bugs with the same symptom appear in all major implementations.
No, half a second can be right. Try loading a lot of tabs, browse in them until the browser dies. Now start Safari again. It will be slow, probably cleaning up caches.
I'm seeing it all the time, as this is my usual way of needing to start the browser (again).
Windows Server 2003 also was the better version of XP. It runs very smooth on a lowly EPIA 800 (P2-350MHz equivalent). XP is kind of a slouch on that system. I guess they think they won't sell enough Windows Server licences when it's not at least 'feeling' faster than their desktop offerings.
Anyway, what could you do to give debuggers special privileges that you could prevent other people from using?
Only make programs 'unprotected' if they are started by a debugger. For example, run them in VM in the debugger address space. This way you can't hijack already running programs.
But yeah, I am aware that there always is an 'outside' to a thread, program, kernel or computer. From the outside you could read values directly. Ex: a computer that is not running could have it's kernel changed so it gives memory dumps of certain programs.
The 6310i has 14 days standby time out of the box. Mine is 3y old and can maintain 12 days. Plus, the battery used in the 6310i is still sold for about 30,-. There are even some other devices (like GPSes) that have standardised on this battery formfactor, so it safe to asume they will be available for some time.
What you could do, is use Bluefrog from Blue Security. Except that service is not available anymore. But it did work. There are people working on more reliable replacement services.
It has been policy for some years already, or at least it has been in the news at voting days in The Netherlands since several years. The obvervations are both used as a protective measure, and as a way for people from foreign countries to experience how voting systems may work. These 'observers' aren't just random people, but most of them have interest in fair voting procedures and how to implement them. They visit lots of countries to sample what works and what does not.
The desktop probably won't see much from having a realtime kernel.
Well, the changes to the kernel to support some sort of usable realtime system have improved the responsiveness of the kernel greatly. Probably none of the programs on my desktop will ever need hard realtime. But boy, do I like it if my programs keep responsive under heavy I/O from other programs.
I do not believe you need kHz-MHz resolution for any desktop application for timing.
Well, to each his own. Like I said probably none of the programs I run will use realtime 'contracts' with the kernel. But I expect this change to be as import as using a "server grade" operating system on the desktop. Just like everybody likes stable computer programs (and OSes), they also like responsive programs. And OSes will need to support that too, and for this support they'll need to change, for the better.
Slashdot Mirror
I think slightly misinterpreted this article:
http://www.pcper.com/article.php?aid=669&type=expert&pid=5
Anyways, writing zeros, or writing something else sequentially should essentially be the same. More dumb Flash based SSDs actually respond to writing zeros, so I just remembered "here you can do the same zeroing trick".
It is not the equivalent, because more data needs to transfered. But the effect is the same, according to Intel. Also works for most SSDs.
You just need a background task that zeros blocks used by deleted files. EasyCo's MFT implements such a log filesystem as a generic block device for x86 Linux and Windows.
Andy Tanenbaum has a nice story about that. On a conference they wanted to use a modulated laser to beam an internet connection to another building. Except they calibrated the target at night, it tested fine. But during the day the air would warm and break the connection. He found it rather odd they didn't just use a cable the next day, but the management insisted on using the broken technology. Didn't work they entire conference.
Troll? What? It's just a technical question. Just like I would want to see a working Xerox Alto, I might want see an "Archimedes" when it is anything special.
You mean the Acorn Archimedes ?
'Wrinkle' as in 'annoying person'? You are an educated troll?
Hmm, I dropped an 's there. Should have been "[..] than the servers they [..]". The rate limiting is good point yes, but their unicornscan tool is for scanning large amounts of server (I think so..).
They -at Outpost24- are basically doing TCP/IP without the per connection timeout timers. By keeping the state in a SYN-cookie-like hashtable and drinking directly from the firehose (network). This enables their scan program to work faster and with less resources than the server they are connecting to. As far as I get it, this part is already publicly available in their Unicornscan program.
Furthermore they discovered that *after* you have done the three way TCP/IP handshake, the OSes are less paranoia than is good for them. There seem to be positions in the statespace in which the TCP/IP stack will keep resending the answer packet. That keeps a timer slot busy, a resource which can be exhausted. Apparently bugs with the same symptom appear in all major implementations.
Is that you, Sam Harris?
No, half a second can be right. Try loading a lot of tabs, browse in them until the browser dies. Now start Safari again. It will be slow, probably cleaning up caches.
I'm seeing it all the time, as this is my usual way of needing to start the browser (again).
Like this patent ? :: http://www.wipo.int/pctdb/en/wo.jsp?IA=WO2002044935&wo=2002044935&DISPLAY=CLAIMS
Why can't you just write out the names of the networking companies? Cisco I know. The others escape me.
http://www.penny-arcade.com/comic/2004/3/19/
Windows Server 2003 also was the better version of XP. It runs very smooth on a lowly EPIA 800 (P2-350MHz equivalent). XP is kind of a slouch on that system. I guess they think they won't sell enough Windows Server licences when it's not at least 'feeling' faster than their desktop offerings.
They run ClamAV on the mailserver of Leopard Server, yes. So that's not an overall resident scanner.
Webkit svn 29385 reaches 70/100. But doesn't show the colored squares either.
Anyway, what could you do to give debuggers special privileges that you could prevent other people from using?
Only make programs 'unprotected' if they are started by a debugger. For example, run them in VM in the debugger address space. This way you can't hijack already running programs.
But yeah, I am aware that there always is an 'outside' to a thread, program, kernel or computer. From the outside you could read values directly. Ex: a computer that is not running could have it's kernel changed so it gives memory dumps of certain programs.
It's not like debuggers couldn't have special privileges, instead of all processes having access rights to other programs memory space.
The 6310i has 14 days standby time out of the box. Mine is 3y old and can maintain 12 days. Plus, the battery used in the 6310i is still sold for about 30,-. There are even some other devices (like GPSes) that have standardised on this battery formfactor, so it safe to asume they will be available for some time.
What you could do, is use Bluefrog from Blue Security. Except that service is not available anymore. But it did work. There are people working on more reliable replacement services.
See: http://en.wikipedia.org/wiki/Blue_Frog
Btw, there are (foreign) election observers in the USA:
+ OR+audit)+(election+OR+voting)+(usa+OR+america)
http://www.google.com/search?q=foreign+(observers
CoreDuoTemp reads from the temperature diode inside the CPU. The fans are wired to a sensor that measures (more or less) the chassis temperature.
Yet, the Dutch will have FOREIGN observers?
It has been policy for some years already, or at least it has been in the news at voting days in The Netherlands since several years. The obvervations are both used as a protective measure, and as a way for people from foreign countries to experience how voting systems may work. These 'observers' aren't just random people, but most of them have interest in fair voting procedures and how to implement them. They visit lots of countries to sample what works and what does not.
The desktop probably won't see much from having a realtime kernel.
Well, the changes to the kernel to support some sort of usable realtime system have improved the responsiveness of the kernel greatly. Probably none of the programs on my desktop will ever need hard realtime. But boy, do I like it if my programs keep responsive under heavy I/O from other programs.
I do not believe you need kHz-MHz resolution for any desktop application for timing.
Well, to each his own. Like I said probably none of the programs I run will use realtime 'contracts' with the kernel. But I expect this change to be as import as using a "server grade" operating system on the desktop. Just like everybody likes stable computer programs (and OSes), they also like responsive programs. And OSes will need to support that too, and for this support they'll need to change, for the better.