> especially when Swing has the far more modern, Windows XP look and feel.
Well if your app has the Luna L&F and the desktop is configured to look like W2000 (my configuration I don't like the "fisher price" look), it looks weird too, is-it the case with Swing?
> You can't hate any language as much as some people hate Java until it's really reached critical mass.
Untrue, there is no need to wait to hate Java, I tried Java in 98-99 and hated it, its libraries were full of bugs (Swing, the GC was leaky on 1.1.8) and you spent your time working around the bugs.
> I don't know, that's for the advanced people at Intel,etc. to research.
Intel doesn't make DRAM anymore, but anyway what makes you think DRAM producers haven't already researched this topic?
Maybe the fact they are not successfull is because the law of physic (aka speed of light) is not helping?
> Given that the speed of the ram bus is typically a fraction of the CPU (heard of 133MHz SDRAM ? 333MHz and some above are common now though)
On one part you talk about RAM access time, on the other part you talk about the frequency of the access bus, sorry but that's two different thing: when Intel supported Rambus their RDRAM has much increased bus bandwith, but the RAM access time was longer! Because CPU RAM are external link, it is much harder to increase frequency, but a way to increase the frequency is to go serial (like RAMBUS did) but there are usually serialisation delay so this means *increased* latency!
The thing which *could* help for bandwith is "edge connected" chip (Sun is researching this, there was an article on/. on this quite some time ago) with connexion of CPU and RAM through this edge of the chip not through traces on the motherboard, but this is much more likely to be usefull for CPU, L3 interface not for DRAM (at first).
For access time, the only real possibility would be a new type of RAM such as MRAM but while they claim SRAM-like access time (not sure if it isn't just marketing speak), density is way lower than DRAM currently and it won't come cheap at first: this is a major development!
And how do you think RAM access time can be improved? Where is the time lost that could be regained?
"Make it faster" is just wishful thinking, apparently it is not so easy: RAM with low latency cost much more than other regular RAM..
The only "easy" way to improve significantly RAM access I can think of would be to bundle CPU and RAM in the same package, but this limits upgrade of course..
Unless you slow down a lot the CPU, this is unlikely to happen: in the same technology, propagation delays implies that the CPU will always run much faster than the RAM.
Using cache and several links from CPU to RAM to add bandwith, helps but the delay is still there and will be felt in case of cache miss due to branch mispredict for example.
> the greatest offenders in me experience are... err... frigging IBM Java apps
That's because you didn't try Sun management console Java app, it "integrates" several old management tools which had non-consistent but useful GUI inside one with a consistent but worthless and slow as hell (sigh Java..) GUI tool.
It is so painful to use that I learned the command-line tools to avoid the @#$%^ GUI! Now maybe this is Sun's goal?
Yes, but of course what is needed is not a package that will last until any radioactivity has disappeared, but a package that will last until the remaining radioactivity is negligible compared to the normal background radioactivity.
But Itanium's competition is both with Power (at the high end) and with x86-64 (at the low end).
I agree with you that x86 (and to some extent x86-64) are dead ugly (I refuse to program in x86 assembly, urgh) but you said "it will be junked sooner or later", this I doubt very, very much!
Think a little bit about the cost of porting software, they would be absolutely huge! So it means that the new thing would have to have perfect x86 compatibility so what would be the incentive to port the new software to the new instruction set?
Performance gain? Tss, like we're missing CPU power nowadays..
I disagree when you wear a watch, you tend to look at it a bit too often, when you don't wear one, you tend to relax more in normal situation..
Of course in cases where your late, it is the reverse, it feels annoying to have to get your phone to get the time, instead of just looking at your watch.
Well you forget something in the equation, the price! Because if money is not in the equation, I would rather like to have a lunar based telescope, but it may be a 'little bit expensive'.
Mmm, you know probably better than me, but having a secure server (one without security holes either in the OS or in the app) is the easy part (yes, I know it is hard).
Being able to resist to a DDOS seems to me the hard part! That's why they paid 100k in the article..
> Personally, I don't see a whole lot that says the modern Catholic Church is a Bad Thing(tm) as an institution
Let me try: - discreminating against women (no woman as priest), gay (which is ironic considering the number of priest who are gay). - way too many pedophile problems, probably helped by preventing priest from marrying and being not very helpful for the victims. - being generally against sex education, condoms or pills and abortion on a *very* crowded planet.
It depends on the person: usually you skydive because free-falling feels very nice (nice adrenalin shoot), not at all like falling: you don't see the ground getting closer (if you are, you wayyy too low), but more like flying.
There is a saying on the web: "If riding in an airplane is flying, then riding in a boat is swimming. If you want to experience the element, get out of the vehicle." And it is very true!
Well I agree that the system is robust for intelligent or non-lazy users, but the providers of the system may still get a lot of call from dumb or lazy users..
OK at the beginning it is small, but once you've reached terminal velocity, you're skydiving quite fast and this 'slight air resistance' is for you a very strong wind!!
Well if you're getting out of either an helicopter or a small plane going not very fast, the wind is quite small during the first seconds of the freefall: so I suppose that 0g is very much similar to these first seconds.
It don't last long, but it still last much more than during a jump on the ground..
For skydivers, this is a quite rare situation as quite often we're getting out of plane going quite fast and when it happens, it is always a bit weird having little 'air grip': moving so slow in the air that you can not really use the air to move your body as you want.
You're playing on words and even in this case you're wrong: installing things on standardised locations helps the configuration of software for the user and for the compilation scripts..
>> 1. Cut and Paste don't work > Always worked fine for me, since about 1998.
Here you're exagerating: Windows users are used to cut/paste images, complex document without too many surprise and in 1998 this wasn't working well in Linux: too many different toolkit gave problems..
Now the toolkit interoperability has improved but it was still troublesome in 1998 unless: you only copy/pasted simple text or you select applications to use only one toolkit, so while it was working for you, it doesn't mean that copy/paste was good on Linux on 98..
*As for the games, frankly the number or quality of games available for Linux is very, very inferior to Windows, why are you in denial? (I like IL2: a beautiful combat flight simulator on Windows)
*For the other points I agree with you.. I remember in 99 how I kept fiddling with my computer settings to try to improve fonts rendering on Linux which was truly awful, now the situation is much, much better! It isn't perfect: font kerning could be better, pdf/ps rendering often use awful fonts but it is good.
RTFA as Cringely said: data should be encrypted before they go into the system, which suppose that the system encrypt the data and which also give the interesting problem: where are the key stored?
If the key are stored by the client, there *will* be problems of lost key, so the key must be also stored on the backup server.
Slashdot Mirror
> especially when Swing has the far more modern, Windows XP look and feel.
Well if your app has the Luna L&F and the desktop is configured to look like W2000 (my configuration I don't like the "fisher price" look), it looks weird too, is-it the case with Swing?
> You can't hate any language as much as some people hate Java until it's really reached critical mass.
Untrue, there is no need to wait to hate Java, I tried Java in 98-99 and hated it, its libraries were full of bugs (Swing, the GC was leaky on 1.1.8) and you spent your time working around the bugs.
> I don't know, that's for the advanced people at Intel,etc. to research.
/. on this quite some time ago) with connexion of CPU and RAM through this edge of the chip not through traces on the motherboard, but this is much more likely to be usefull for CPU, L3 interface not for DRAM (at first).
Intel doesn't make DRAM anymore, but anyway what makes you think DRAM producers haven't already researched this topic?
Maybe the fact they are not successfull is because the law of physic (aka speed of light) is not helping?
> Given that the speed of the ram bus is typically a fraction of the CPU (heard of 133MHz SDRAM ? 333MHz and some above are common now though)
On one part you talk about RAM access time, on the other part you talk about the frequency of the access bus, sorry but that's two different thing: when Intel supported Rambus their RDRAM has much increased bus bandwith, but the RAM access time was longer!
Because CPU RAM are external link, it is much harder to increase frequency, but a way to increase the frequency is to go serial (like RAMBUS did) but there are usually serialisation delay so this means *increased* latency!
The thing which *could* help for bandwith is "edge connected" chip (Sun is researching this, there was an article on
For access time, the only real possibility would be a new type of RAM such as MRAM but while they claim SRAM-like access time (not sure if it isn't just marketing speak), density is way lower than DRAM currently and it won't come cheap at first: this is a major development!
Yes, of course we all know that praying has nothing to do with religion/superstition..
So I guess that the dictionnary is wrong then?
Prayer: 1 a (1) : an address (as a petition) to God or a god in word or thought
And how do you think RAM access time can be improved?
Where is the time lost that could be regained?
"Make it faster" is just wishful thinking, apparently it is not so easy: RAM with low latency cost much more than other regular RAM..
The only "easy" way to improve significantly RAM access I can think of would be to bundle CPU and RAM in the same package, but this limits upgrade of course..
Unless you slow down a lot the CPU, this is unlikely to happen: in the same technology, propagation delays implies that the CPU will always run much faster than the RAM.
Using cache and several links from CPU to RAM to add bandwith, helps but the delay is still there and will be felt in case of cache miss due to branch mispredict for example.
> the greatest offenders in me experience are... err... frigging IBM Java apps
That's because you didn't try Sun management console Java app, it "integrates" several old management tools which had non-consistent but useful GUI inside one with a consistent but worthless and slow as hell (sigh Java..) GUI tool.
It is so painful to use that I learned the command-line tools to avoid the @#$%^ GUI!
Now maybe this is Sun's goal?
Yes, but of course what is needed is not a package that will last until any radioactivity has disappeared, but a package that will last until the remaining radioactivity is negligible compared to the normal background radioactivity.
>If all we'll be stuck with will be x86, it'll be quite sad.
I don't think this will happen because of embedded CPUs especially since Intel is also selling ARMs.
But for servers, I wouldn't bet against it if Intel decided to change its mind..
>If its competition was Power or MIPS, then OK,
But Itanium's competition is both with Power (at the high end) and with x86-64 (at the low end).
I agree with you that x86 (and to some extent x86-64) are dead ugly (I refuse to program in x86 assembly, urgh) but you said "it will be junked sooner or later", this I doubt very, very much!
Think a little bit about the cost of porting software, they would be absolutely huge!
So it means that the new thing would have to have perfect x86 compatibility so what would be the incentive to port the new software to the new instruction set?
Performance gain? Tss, like we're missing CPU power nowadays..
I think that your sig show your bias:
>"It is impossible to rightly govern the world without God and the Bible"--George Washington, 9/17/1796
Many European countries have a very strict separation of church and state and do live very well, so apparently George Washington was just mistaken..
>All religion is being removed from the public eye
Well, you know what?
Religion is a private matter, not public!
I disagree when you wear a watch, you tend to look at it a bit too often, when you don't wear one, you tend to relax more in normal situation..
Of course in cases where your late, it is the reverse, it feels annoying to have to get your phone to get the time, instead of just looking at your watch.
>so which would you rather have?
Well you forget something in the equation, the price!
Because if money is not in the equation, I would rather like to have a lunar based telescope, but it may be a 'little bit expensive'.
Mmm, you know probably better than me, but having a secure server (one without security holes either in the OS or in the app) is the easy part (yes, I know it is hard).
Being able to resist to a DDOS seems to me the hard part! That's why they paid 100k in the article..
> Personally, I don't see a whole lot that says the modern Catholic Church is a Bad Thing(tm) as an institution
Let me try:
- discreminating against women (no woman as priest), gay (which is ironic considering the number of priest who are gay).
- way too many pedophile problems, probably helped by preventing priest from marrying and being not very helpful for the victims.
- being generally against sex education, condoms or pills and abortion on a *very* crowded planet.
It depends on the person: usually you skydive because free-falling feels very nice (nice adrenalin shoot), not at all like falling: you don't see the ground getting closer (if you are, you wayyy too low), but more like flying.
There is a saying on the web: "If riding in an airplane is flying, then riding in a boat is swimming. If you want to experience the element, get out of the vehicle." And it is very true!
Well I agree that the system is robust for intelligent or non-lazy users, but the providers of the system may still get a lot of call from dumb or lazy users..
> slight air resistance
OK at the beginning it is small, but once you've reached terminal velocity, you're skydiving quite fast and this 'slight air resistance' is for you a very strong wind!!
>What about the wind?
Well if you're getting out of either an helicopter or a small plane going not very fast, the wind is quite small during the first seconds of the freefall: so I suppose that 0g is very much similar to these first seconds.
It don't last long, but it still last much more than during a jump on the ground..
For skydivers, this is a quite rare situation as quite often we're getting out of plane going quite fast and when it happens, it is always a bit weird having little 'air grip': moving so slow in the air that you can not really use the air to move your body as you want.
>For people who are afraid of heights
No.. You don't suffer from the fear of height when you're exiting the plane: you're too high (and no I'm not joking).
I know: I have the fear of height and have 1000+ skydive jumps, actually quite a few skydivers suffer from the fear of height..
Of course, getting out of the plane is quite frightening at the beginning, but this has nothing to do with the 'fear of feight'.
You're playing on words and even in this case you're wrong: installing things on standardised locations helps the configuration of software for the user and for the compilation scripts..
Cool, thanks a lot for your work!
>> 1. Cut and Paste don't work
> Always worked fine for me, since about 1998.
Here you're exagerating: Windows users are used to cut/paste images, complex document without too many surprise and in 1998 this wasn't working well in Linux: too many different toolkit gave problems..
Now the toolkit interoperability has improved but it was still troublesome in 1998 unless: you only copy/pasted simple text or you select applications to use only one toolkit, so while it was working for you, it doesn't mean that copy/paste was good on Linux on 98..
*As for the games, frankly the number or quality of games available for Linux is very, very inferior to Windows, why are you in denial?
(I like IL2: a beautiful combat flight simulator on Windows)
*For the other points I agree with you..
I remember in 99 how I kept fiddling with my computer settings to try to improve fonts rendering on Linux which was truly awful, now the situation is much, much better!
It isn't perfect: font kerning could be better, pdf/ps rendering often use awful fonts but it is good.
RTFA as Cringely said: data should be encrypted before they go into the system, which suppose that the system encrypt the data and which also give the interesting problem: where are the key stored?
If the key are stored by the client, there *will* be problems of lost key, so the key must be also stored on the backup server.