What's you favorite quote from a language advocate?
My favorite was from Philip Greenspun. He's a professor of CS at MIT and runs the www.photo.net site. He was at my school giving a lecture on database backed web sites. Having heard of Zope and Python I asked him "what do you think of Python?" He gave me a blank look and said "better languages have been designed 30 years ago" (obviously lisp). This from a guy who writes his web stuff in tcl. ROTFL.
First of all, photons are gamma and electrons are beta-.
But it's the heavier particles (like iron nuclei) that do the most damage. They tend to rip up the target more as they zip through. Neutrons are interesting because they can get 'captured' by the target atom, changing it into a different (possibly radioactive) isotope. Slow alpha aren't particularly dangerous because have very little penetration. Fast alphas on the other hand are nasty. Gammas aren't particularly bad because they tend not to hit anything as they pass by.
On windows much of the keyboard/mouse logic is handled by the OS (common dialogs/common controls). Doesn't linux (gtk/E/gnome/swamill/whatever) do something similar? Or does everyone have to reinvent the wheel?
It IS obvious. One of my projects is a web search engine and I have given dns-based session control quite a bit of thought. As a spider designer, it gives me the heebie-jeebies.
No one uses dns encoding because it poisons dns caches. Remember, dns lookups that aren't cached on a nearby server require sending a request/response from at least two other machines. Here's what a session might look like.
////////////////////
First query the local server. Very fast since the connection is probably ethernet. me -> dns.ryans.dhs.org
Now my local dns goes off and searches for the ip address.
Local dns queries the root servers. dns.ryans.dhs.org -> b.root-servers.net dns.ryans.dhs.org ns1-auth.sprintlink.net dns.ryans.dhs.org - ns1-auth.sprintlink.net
Local dns sends me the answer me - dns.ryans.dhs.org
Start tcp session
////////////
As you can see, the name lookup needed one short-haul and two long-haul roundtrips. If it was cached only one short-haul conversation would have been needed.
So what if it's not perfect? Who cares about perfection? Bad ip routes block legitimate sites too, but it is sufficiently rare that nobody cares. Mozilla probably blocks more sites than Norton ever did.
You ever-present parent solution will have a higher failure rate than software filters because no parent will be ever-present. Browsing time will have to be reduced (which may be a good thing).
Your solution is just like saying "prevent drunk drivers by giving the cops all car keys. You have to ask for permission to start your engine." It's too inconvienent to be effective.
BTW, the net is a lot more important than a car for many people. Like all us college students who brought our computers to school and left our cars at home.
First off, performance and real usability issues should always take priority over eye candy. I don't have resources to waste on pretty bs.
Why does mozilla break all the user interface rules (like middle button scrolling)? This pisses me off because they must have spent a bundle of time reimplementing the entire keyboard/mouse logic (incorrectly). Don't fix [break] it if it isn't broken.
For an OS that started on text terminals, linux sure jacked up it's keyboard handling. Back in my windows days I didn't use the mouse (ever, 'cept browsing). With linux I have to use it all the time. I suppose it's really the windows manager / x server / apps fault but it makes the whole system suck.
If you disagree you can post you reasons. If you have no reasons moderate me down instead.
Yes, your phone does have a very fast processor. The processor is designed with a very specialized instruction set that is an extreme form of risc called a Parallel Logic Ladder (PLL). These processors are typified by the PLLatinum line from National Semiconductor.
PLLs have an instruction set that is sparse even compared to conventional risc chips. Most PLLs only have one instruction, integer divide. This makes programming a bit cumbersome at times but modern compilers make the job easier.
Over the last decade PLLs have worked their way into almost every wireless communications device. Their rapid growth has been attributed to the massive computational demands of the NSA's voice recognition software.
Exactly right. When applied to processors (or other computational components) MHz refers to inverse time. This phenomenon is an obvious consequence of Gates' Law and Moore's Law.
Moore: Speed of hardware *2 every 18 months. Gates: Speed of software 1/2s every 18 months.
As you can see, there is an inverse relationship between hardware and software speed, that is, software = INV(hardware). This is where the term 'inverse time' comes from.
Yeah, but my Nokia 21 inch smacks your tv. Also, the ps2 does not have a 3600 Hz refresh rate. It has a 60 Hz interlaced refresh rate (duh). This is equivalent to a fuzzy 30 Hz monitor.
With only 40 columns a tv's text capabilities suck, to say nothing of real graphics.
> On the other hand, increasing processor power > means that the overhead from running the > 'desperate' software virtualizer fortunately > becomes more irrelevant each day.
That would be true if the overhead of the virtualizer was less than O(n) (ie sqrt or log or something). I would guess that scanning is O(n) wrt instructions executed. If scanning is only performed once (O(n) wrt executable size) then you are correct.
Officially, we're only supposed to use the network for academic stuff (I think). In practive, network services lets you run whatever you want. There is no proxy, no firewall, just a pipe to the backbone.
In recent months napster traffic has become a significant portion of the residence hall internet traffic and is getting too expensive for Network Services. A few weeks ago Network Services instituted a 15 megabit/s cap on dorm internet traffic. The rest of the campus is unaffected. No services are blocked. Everything is just slower.
What is the workers don't make products? What then? You're forgetting the teachers, lawyers, delivery men, accountants, and everyone else in the service sector (which is more than half of the US economy, btw).
A memory heirarchy can ALWAYS be faster and larger than a memory system with only one type of ram. Memory heirarchies exploit the temporal and spatial locality of programs. Caches put the speed where it does the most good, thereby achieving better performance for a given amount of money.
It's sort of like the arguments about mp3 vs. raw audio. Mp3 audio compression puts the bits where they'll do the most good according to a psychoacoustic model of the human ear and brain. You'll see fools saying that a cd sounds much better that a 128k mp3. Of course it does, the cd has 10 times as many bits to work with. But compare a 128k mp3 to a 128k raw file and you'll see that the mp3 is much, much better.
Sure, a $10,000 sram bank will be marginally faster than a $500 dram bank, but who cares? I'd rather have 10 times the ram and take the 5% speed hit (assuming sram is 10x faster than dram).
You fail to grasp that fast memory will always be more expensive than slow memory. That should be obvious, if for no other reason than fast memory needs six transistors where dram needs one.
Here's an example. Say fast memory is 10x the price of slow memory and 10x the speed. Also, assume that a cache one tenth the size of the main ram has a miss rate of 0.5%. Given these specs, you can have either:
N megs of fast memory with an average access time of Q.
5N megs of dram with N/2 megs of cache with an average access time of 1.045Q.
That's five times the capacity for a 4.5% speed hit. In the real world the situation is even more favorable for a memory heirarchy.
BTW, nobody just uses ram anymore, everyone has swap space too. Why don't you argue for disks that run at core speed?
If yo want to run 1600 just go buy a 21 inch monitor. It's pretty neat having a display that big. My desk is a bit small so I put my monitor about 8 inches off center. My eyes have to refocus when I look at the far right edge. Just enough to notice, not the least bit annoying.
Slashdot Mirror
Yes, lithography with a hundred billion micron feature size. Suck that, intel!
Ryan
Now that the language wars are going to start...
What's you favorite quote from a language advocate?
My favorite was from Philip Greenspun. He's a professor of CS at MIT and runs the www.photo.net site. He was at my school giving a lecture on database backed web sites. Having heard of Zope and Python I asked him "what do you think of Python?" He gave me a blank look and said "better languages have been designed 30 years ago" (obviously lisp). This from a guy who writes his web stuff in tcl. ROTFL.
Ryan Salsbury
First of all, photons are gamma and electrons are beta-.
But it's the heavier particles (like iron nuclei) that do the most damage. They tend to rip up the target more as they zip through. Neutrons are interesting because they can get 'captured' by the target atom, changing it into a different (possibly radioactive) isotope. Slow alpha aren't particularly dangerous because have very little penetration. Fast alphas on the other hand are nasty. Gammas aren't particularly bad because they tend not to hit anything as they pass by.
Ryan
On windows much of the keyboard/mouse logic is handled by the OS (common dialogs/common controls). Doesn't linux (gtk/E/gnome/swamill/whatever) do something similar? Or does everyone have to reinvent the wheel?
Ryan
How do you know it's not obvious, Andrew Cady?
RYan
It's sort of like patenting the use of teeth to open beers. Everyone's thought of it but nobody actually does it for obvious reasons.
It IS obvious. One of my projects is a web search engine and I have given dns-based session control quite a bit of thought. As a spider designer, it gives me the heebie-jeebies.
No one uses dns encoding because it poisons dns caches. Remember, dns lookups that aren't cached on a nearby server require sending a request/response from at least two other machines. Here's what a session might look like.
////////////////////
First query the local server. Very fast since the connection is probably ethernet.
me -> dns.ryans.dhs.org
Now my local dns goes off and searches for the ip address.
Local dns queries the root servers.
dns.ryans.dhs.org -> b.root-servers.net
dns.ryans.dhs.org ns1-auth.sprintlink.net
dns.ryans.dhs.org - ns1-auth.sprintlink.net
Local dns sends me the answer
me - dns.ryans.dhs.org
Start tcp session
////////////
As you can see, the name lookup needed one short-haul and two long-haul roundtrips. If it was cached only one short-haul conversation would have been needed.
Ryan
I'll mirror it. (response emailed too).
Ryan Salsbury
So what if it's not perfect? Who cares about perfection? Bad ip routes block legitimate sites too, but it is sufficiently rare that nobody cares. Mozilla probably blocks more sites than Norton ever did.
You ever-present parent solution will have a higher failure rate than software filters because no parent will be ever-present. Browsing time will have to be reduced (which may be a good thing).
Your solution is just like saying "prevent drunk drivers by giving the cops all car keys. You have to ask for permission to start your engine." It's too inconvienent to be effective.
BTW, the net is a lot more important than a car for many people. Like all us college students who brought our computers to school and left our cars at home.
Ryan
There are between 500M and 2G pages on the web. That's a lot of stuff to read (though keyword matching can mark most of it 'safe').
Ryan
Not to mention the slashdot effect. Mmmm.. 404 free porn.
First off, performance and real usability issues should always take priority over eye candy. I don't have resources to waste on pretty bs.
Why does mozilla break all the user interface rules (like middle button scrolling)? This pisses me off because they must have spent a bundle of time reimplementing the entire keyboard/mouse logic (incorrectly). Don't fix [break] it if it isn't broken.
For an OS that started on text terminals, linux sure jacked up it's keyboard handling. Back in my windows days I didn't use the mouse (ever, 'cept browsing). With linux I have to use it all the time. I suppose it's really the windows manager / x server / apps fault but it makes the whole system suck.
If you disagree you can post you reasons. If you have no reasons moderate me down instead.
Ryan
Yes, your phone does have a very fast processor. The processor is designed with a very specialized instruction set that is an extreme form of risc called a Parallel Logic Ladder (PLL). These processors are typified by the PLLatinum line from National Semiconductor.
PLLs have an instruction set that is sparse even compared to conventional risc chips. Most PLLs only have one instruction, integer divide. This makes programming a bit cumbersome at times but modern compilers make the job easier.
Over the last decade PLLs have worked their way into almost every wireless communications device. Their rapid growth has been attributed to the massive computational demands of the NSA's voice recognition software.
Ryan
> The processor speed is not a frequency.
Exactly right. When applied to processors (or other computational components) MHz refers to inverse time. This phenomenon is an obvious consequence of Gates' Law and Moore's Law.
Moore: Speed of hardware *2 every 18 months.
Gates: Speed of software 1/2s every 18 months.
As you can see, there is an inverse relationship between hardware and software speed, that is, software = INV(hardware). This is where the term 'inverse time' comes from.
Ryan KE6FFQ
Yeah, but my Nokia 21 inch smacks your tv. Also, the ps2 does not have a 3600 Hz refresh rate. It has a 60 Hz interlaced refresh rate (duh). This is equivalent to a fuzzy 30 Hz monitor.
With only 40 columns a tv's text capabilities suck, to say nothing of real graphics.
Ryan
> On the other hand, increasing processor power
> means that the overhead from running the
> 'desperate' software virtualizer fortunately
> becomes more irrelevant each day.
That would be true if the overhead of the virtualizer was less than O(n) (ie sqrt or log or something). I would guess that scanning is O(n) wrt instructions executed. If scanning is only performed once (O(n) wrt executable size) then you are correct.
Ryan
You think edus don't have porn? Edus probably host the largest concentration of teenage males with web sites on the net. Beware *.dorm.college.edu
Ryan
Officially, we're only supposed to use the network for academic stuff (I think). In practive, network services lets you run whatever you want. There is no proxy, no firewall, just a pipe to the backbone.
In recent months napster traffic has become a significant portion of the residence hall internet traffic and is getting too expensive for Network Services. A few weeks ago Network Services instituted a 15 megabit/s cap on dorm internet traffic. The rest of the campus is unaffected. No services are blocked. Everything is just slower.
Ryan
What is the workers don't make products? What then? You're forgetting the teachers, lawyers, delivery men, accountants, and everyone else in the service sector (which is more than half of the US economy, btw).
Ryan
BTW, I think most pcmcia (hate that acro) are isa. PCMCIA is little more than isa with a smaller connector.
Ryan
A memory heirarchy can ALWAYS be faster and larger than a memory system with only one type of ram. Memory heirarchies exploit the temporal and spatial locality of programs. Caches put the speed where it does the most good, thereby achieving better performance for a given amount of money.
It's sort of like the arguments about mp3 vs. raw audio. Mp3 audio compression puts the bits where they'll do the most good according to a psychoacoustic model of the human ear and brain. You'll see fools saying that a cd sounds much better that a 128k mp3. Of course it does, the cd has 10 times as many bits to work with. But compare a 128k mp3 to a 128k raw file and you'll see that the mp3 is much, much better.
Sure, a $10,000 sram bank will be marginally faster than a $500 dram bank, but who cares? I'd rather have 10 times the ram and take the 5% speed hit (assuming sram is 10x faster than dram).
Ryan
You fail to grasp that fast memory will always be more expensive than slow memory. That should be obvious, if for no other reason than fast memory needs six transistors where dram needs one.
Here's an example. Say fast memory is 10x the price of slow memory and 10x the speed. Also, assume that a cache one tenth the size of the main ram has a miss rate of 0.5%. Given these specs, you can have either:
N megs of fast memory with an average access time of Q.
5N megs of dram with N/2 megs of cache with an average access time of 1.045Q.
That's five times the capacity for a 4.5% speed hit. In the real world the situation is even more favorable for a memory heirarchy.
BTW, nobody just uses ram anymore, everyone has swap space too. Why don't you argue for disks that run at core speed?
Ryan Salsbury
I'm on an ethernet lan and slashdot is a dog. And yes, the ethernet is the slowest link.
Ryan
Did you know that you can get 100% of all dialy vitamins from 43 pints of Guinness and a glass of milk? Mmmm, healty!
If yo want to run 1600 just go buy a 21 inch monitor. It's pretty neat having a display that big. My desk is a bit small so I put my monitor about 8 inches off center. My eyes have to refocus when I look at the far right edge. Just enough to notice, not the least bit annoying.