I've been using SL for a little more than a year so far. Went there with the explicit idea of that I'd probably script something, as the idea sounded interesting. So I got a SL account, and now it got to the point I pretty much have a monitor dedicated to it.
I use SL mostly as a glorified chatroom, and don't move around much, primarily hanging around in Luskwood. If you want a concentration of furries then check it out, but have in mind that right now it'll be quite empty, as most of the population is American. There will be a lot more people in a few hours. I'd say the problem with SL is the same as with any IRC server, until you find a place for yourself it's hard to figure out what to do there.
I also like SL as a base for certain coding ideas. I run a reputation system as an alternative to the one provided by SL, and also do some work on the SL source code. In that sense, SL is appealing because it's already there, so I only have to add my ideas to it, and it offers a large potential user base. I think that SL is appealing for a programmer, builder or artist in that it's a very convenient medium for saying "hey look what I made" and getting a reaction (and perhaps even cash). You see people working on all sorts of interesting stuff.
So what do I do in SL? I mostly hang around and talk to people. It's a bit nicer than IRC in that you can have a more RL-like conversation. People can easily gather in a group and talk about whatever they want without having to form a separate channel. Sometimes I wander around and check out cool stuff people made. I try building a bit. Sometimes I try playing chess with rather bad results. I script and change the SL client. IMO, SL is a bit overhyped right now, but it's still pretty fun to be in.
That's just the Linux kernel having a stupid behavior by default.
By default, it lets processes overcommit memory. That means you can malloc more than there actually is. This is done with the expectation that programs allocate extra memory they don't actually use. Problem is that an excessive allocation succeeds, but then the system can't satisfy it, so it has to kill some random process.
Do this:
# echo 2 >/proc/sys/vm/overcommit_memory
This will turn off overcommit completely. When some program tries to request too much, malloc simply fails. No random processes get killed. The program that tired to allocate the memory is given a chance to handle the failure, unlike what happens with the default setting.
Indeed. I've been badly burned with memory problems before, so now I only buy ECC.
I had a particularly nasty incident. My firewall had been running for months without problems, until one day it crashed. I thought oh well, maybe it hit an obscure kernel bug. Rebooted it. Several days later it crashed again. Rebooted it again. The next time it crashed but didn't boot again as it had corrupted its disk, and I had a really fun day reinstalling it with no internet connection.
My current box takes DDR2 800. I was going to get ECC, but they only had DDR2 666 available. I decided to go with the normal stuff. It passed memtest86, and I started installing gentoo. After a few hours bizarre compilation errors started to happen, and gcc started segfaulting. Turns out the RAM was very slightly bad. So slightly that I had to test it for 10 hours straight to see the problem.
Went back to the shop, exchanged it for the slower ECC. Can't see any noticeable performance difference, and it's rock solid now. It really irks me how for some bizarre reason there's error correction everywhere, on every kind storage media, except RAM, where for some reason it's an "enterprise feature".
I have my box freeze once in a while due to RAM exhaustion. It goes into swap, and swaps so heavily I can't even move the mouse around anymore. The problem with swap these days is that it's horribly slow compared to RAM. While memory keeps getting faster and larger, the hard disk's speed doesn't grow anywhere that fast. So once you're swapping and using a significat amount of it (just 256MB say) everything grinds to a halt.
Current OSes seem to try too hard to keep stuff running. That might have made sense in the past, but I don't want that anymore. If something goes insane and tries to get all the RAM for itself I want it to die, quickly, without freezing the machine solid for several minutes.
Second life is a game, education is not. Get the education through known (quantifiable) channels before playing with games.
It took me a lot of grinding, but I finally levelled up in my scripting skill!;-)
Seriously, SL is just as much of a game as IRC or Slashdot is. You can play a game through it (by posting chess moves for instance), but that doesn't really make it a game in itself.
If I met someone who acquired their education via Second Life, I would laugh hysterically at them. Then I would toss my spare change into their tin can and while I continue on my way to work.
So what exactly would make you feel so superior? I would hope that the prestige of one's education is based on its quality, rather than on where you got it.
Here in Spain we have the UNED, a distance university. I think it's the spanish university with the highest number of students. You can pretty much study everything at home, although newsgroups and forums are available. By your logic, acquiring an education in USENET and web boards must be really funny as well.
If they decided to open a place in SL, it wouldn't stop being what it currently is, would it? The exam certainly wouldn't get any easier. While I have my doubts regarding whether SL would be an improvement over forums, I think such a thing could be tried quite successfully.
But it is chat. Only chat. Chat that you can't archive, that is done with word bubbles, and without a moderation system. What on earth would make you think that this would be a good platform for instruction?
Chat can be archived, you can access a log in SL (ctrl+H), and it can be moderated. If you own the land you're speaking on you can kick people out, ban them, or make the land accessible only to people on the allow list.
- It's distracting as all hell - your students will spend all their time customizing/scoping out each others' avatars
You can make your talk be in a place that's not distracting, and people don't mess with their avatars all that much. I've looked the same for more than half a year so far.
The best argument for threads v processes is Apache. Personally, I agree with the Apache group that Apache 2 with its thread-based model is better. They should know.
Are you joking? That switch was one of the main reasons why apache2 wasn't adopted for a very long time, as the myriad of PHP extensions wouldn't work because they weren't thread safe.
In fact, I didn't upgrade to apache2 until I found out that threads can be completely removed from it, so that it works the same way as apache1 did:
What you say would be true if money had the same value to everybody. Let's say that we have $1000 to split. You make the division: You take $950 for yourself, and give me $50. Now, I'm not that poor that $50 is a big matter to me, I might as well derive more satisfaction from ensuring you don't get $950 for being a bastard than from getting $50.
A prerequisite for person B to say "no" would be that the amount of money they'd get would be below their threshold. Maybe they're filthy rich and even $10000 would be a drop in the bucket.
Reasons why person B may say "no":
They have an axe to grind: You might control the division, but they have the power to decide whether you get away with it, and may consider punishing you for a bad action more worthy
They might be in a bad mood and feel like screwing over a random stranger
They might think that it's just an experiment and that they won't really get money in any case.
They might think it's a theoretical experiment, and that after it both people will get the same amount for their collaboration
They're perfectly aware that this is an experiment and want to show decent, moral people
Note that all of those, even the last one are perfectly selfish.
Re:What I'd like to see (and plan to implement soo
on
TrueCrypt 4.3 Released
·
· Score: 1
Only problem there is that I'm not a large customer, just a normal guy with a server in the closet
Re:What I'd like to see (and plan to implement soo
on
TrueCrypt 4.3 Released
·
· Score: 1
That has some problems.
It'd require an extra disk, which I don't really need right now as I have enough space. Overall it'd be less reliable (3 drives are more likely to fail than 2). It's also a suboptimal configuration for a database server. And it's a lot more error prone, as any component of a RAID-1 can be used independently if needed, while that can't be done with a RAID-5.
My concern is that data can't be recovered from a broken disk sent for replacement, but while it's here I'd like to have as much convenience as possible.
Re:What I'd like to see (and plan to implement soo
on
TrueCrypt 4.3 Released
·
· Score: 1
That would need to be the heck of a magnet. From what I hear, a magnet strong enough would bend the platters. If I had the cash and right place (I'm definitely not going to use something like that anywhere near anything valuable) for that sort of thing I wouldn't really worry about getting warranty replacements.
What I'd like to see (and plan to implement soon)
on
TrueCrypt 4.3 Released
·
· Score: 1
I'd like to have a system where all disks are encrypted, but a password isn't needed to mount them.
Problem: I run a server with a RAID-1 setup. After a few months a disk fails. I remove it and want to get it replaced under warranty. The problem is that the disk isn't in a good enough condition to be able to fully overwrite it, and something sensitive could remain in some obscure area, like reallocated sectors. Server stores a quite large amount of rather private data I feel really uncomfortable letting go. What if it gets fixed and somebody else ends up recovering my stuff?
Solution idea: Make the box boot from an alternate media, such as a CD or a flash drive that would contain kernel, initrd and keys. Whole hard disk would go through dmcrypt, then RAID, then LVM. If disk fails, it can be removed and exchanged without leaving anything readable on it. If the boot media is removed any stored data is quickly made inaccessible.
I'd really like to see something like this being offered as an advanced install option in a distribution.
As fast as possible without making it annoyingly loud. At the very least, dual core. It would be possible to go up to 2 x quad core CPUs, but that's probably too wasteful and loud.
At least 4GB ECC RAM. Better 8GB. ECC because I REALLY hate memory problems.
Silent. Must be possible to sleep right next to it
Quiet, fast and reliable storage. Perhaps a Flash drive.
Assuming Flash can't be large enough, separated mass storage, with hard disks in RAID-1 (with hot spares) located somewhere I can't hear the noise.
Tape library for backups.
Automatic backups of small amounts of vital data to an offsite server (tape is for full backups, this is for small things like source code)
Decent video card. Doesn't have to be anything specially overkill
4 20" LCD monitors, minimum 1600x1200 resolution.
Optimus keyboard, or something similar.
Fully Linux compatible, with no closed source drivers required.
Easily maintainable. Must have a good case that makes it convenient to replace components if needed.
Located somewhere reachable but out of sight.
A second box, with exactly the same hardware, kept in case the primary one fails for some reason.
System configured in such a way that everything is automatically backed up, and a complete hardware failure can be recovered from in a couple of hours
This should be quite doable right now with current technology, except for the stuff that hasn't been released yet.
Besides incredibly trivial things like replacing our light bulbs with CFDs and turning off our power when we go outside, which is a *trivial* amount of savings, there's not much humans can do to save power.
You're kidding, right? There are many places where power can be saved. Example:
My laptop uses about 25W, while my desktop uses probably more than 200. Benchmarks say that the Pentium M 1600 was roughtly equivalent to one CPU in my dual Athlon MP 2000+, while the whole laptop uses less power than one CPU in the desktop.
CFD bulbs use less than half the power of incandescents
LCD monitors use much less power than CRTs. OLEDs seem to be an even further improvement
Improving insulation in the home can easily cut the heating bill in half
No alternative energy source is capable of cheaply producing mass amount of energy, and even most of these run into problems with environmentalists as well (damns stop fish from migrating, windmills chop up endangered birds, etc.)
Sure it is, after all the whole planet is solar powered.
Also, I don't get what's the deal with "windmills chop birds". Yes, they do sometimes. But birds fly into skyscrapers quite often too, and for some reason that doesn't seem to stop anybody from building them. There are also types of windmills that rotate slower, and are less dangerous to birds.
Now, I have absolutely nothing against nuclear power, if well done, but what you say simply isn't true. Other viable alternatives exist as well, and nuclear isn't the only option.
Why WoW? There are more things out there. I bet quite a few people would like to use this in Second Life. Make an avatar based on yourself first, then modify it to perfection.
Yeah, as if people are going to use trigonometry to build shapes. It's a sort of neat idea, I suppose, but doesn't seem to be very suitable for the general SL population.
IIRC, the shape constraints in SL are mostly imposed by Havok, the physics engine they use.
SL users also don't really have much power to spare: right now I'm getting 20 FPS, on a dual Athlon 64 X2 5200+, with a GeForce 7900GS. For most people, SL is seriously slow and they wouldn't like extra load to be added to the client.
In any case, the main problem in SL right now is scalability. Nobody really cares much how shapes generated in comparison to that, and people manage to make really neat things with them anyway.
Err, why exactly? The whole point of SL is that it doesn't have the restrictions of RL, otherwise what fun it would be?
Incidentally, from what I hear it did work the way you say in the past, and you had to fly to get anywhere. Doesn't work that way anymore, and since the previous system was abandoned, they're unlikely to return to it.
In SL, land is far from infinite. Land is mapped directly to server resources. If you buy your own island (65536 m^2), you're effectively paying for a CPU on a server to be dedicated to you. Considering what servers cost, they're not overcharging much.
While land could be infinite, in the sense that you'd pay for resources consumed instead of chunks of space, it currently doesn't work that way in SL, although I've heard rumors of that they'd like it to work eventually that way. Then you could have all the land you'd like, but you'd have to pay for your CPU usage. In that case I imagine you'd pay depending on the amount of server time your scripts use, the number of objects you have, etc.
Assumption is mostly correct, except with one thing, you're not really paying 70 dollars, you're paying almost nothing. Shortly, most of what you pay actually pays for your stipend, as if you were regularly buying it at the Linden Exchange. So you can recover most of that.
As a fairly old resident, I keep my L$500/week stipend for my premium account, which earns me about $16 a year. It can't get much better than having the company paying you instead of the other way. In fact, I wrote that blog entry because I was tired of all the complaints due to stipend reduction. They were needed, or it wouldn't make sense to have premium accounts in the first place.
My understanding is that primitives are a requirement if you want physics. Calculating the physics of objects with an arbitrary shape would be seriously complicated, while it's trivial to calculate say, the volume of a sphere.
Regarding advanced graphical features, LL are adding reflections, but really, I don't think most people at the time want LL to work on scalability. Everybody I see is saying "scalability first, everything else second". So massive graphical improvements are probably going to come from other people. Somebody on the mailing list mentioned making a patch for a stereo version of SL, which sounds very cool, but I haven't seen it yet.
Very well, there are too many. What are you going to do about it?
You can complain until you're blue in the face, but the thing is, the authors of those distros don't give a damn you think there are too many. They work on them because they like it. I had an idea for a distro myself once, and while it didn't materialize due to the lack of time, I can tell you I'd do such a thing only for my own enjoyment. It's absolutely not a matter of what "we" (whoever that is) need, it's a matter of what I want to do. Your opinion about the amount of distros being excessive doesn't enter into it.
Like I said, it's a pointless discussion. You can sit here and whine all you want, but really, nobody responsible cares, so nothing is going to come out of it.
Slashdot Mirror
Yep, I do.
I've been using SL for a little more than a year so far. Went there with the explicit idea of that I'd probably script something, as the idea sounded interesting. So I got a SL account, and now it got to the point I pretty much have a monitor dedicated to it.
I use SL mostly as a glorified chatroom, and don't move around much, primarily hanging around in Luskwood. If you want a concentration of furries then check it out, but have in mind that right now it'll be quite empty, as most of the population is American. There will be a lot more people in a few hours. I'd say the problem with SL is the same as with any IRC server, until you find a place for yourself it's hard to figure out what to do there.
I also like SL as a base for certain coding ideas. I run a reputation system as an alternative to the one provided by SL, and also do some work on the SL source code. In that sense, SL is appealing because it's already there, so I only have to add my ideas to it, and it offers a large potential user base. I think that SL is appealing for a programmer, builder or artist in that it's a very convenient medium for saying "hey look what I made" and getting a reaction (and perhaps even cash). You see people working on all sorts of interesting stuff.
So what do I do in SL? I mostly hang around and talk to people. It's a bit nicer than IRC in that you can have a more RL-like conversation. People can easily gather in a group and talk about whatever they want without having to form a separate channel. Sometimes I wander around and check out cool stuff people made. I try building a bit. Sometimes I try playing chess with rather bad results. I script and change the SL client. IMO, SL is a bit overhyped right now, but it's still pretty fun to be in.
By default, it lets processes overcommit memory. That means you can malloc more than there actually is. This is done with the expectation that programs allocate extra memory they don't actually use. Problem is that an excessive allocation succeeds, but then the system can't satisfy it, so it has to kill some random process.
Do this: This will turn off overcommit completely. When some program tries to request too much, malloc simply fails. No random processes get killed. The program that tired to allocate the memory is given a chance to handle the failure, unlike what happens with the default setting.
Indeed. I've been badly burned with memory problems before, so now I only buy ECC.
I had a particularly nasty incident. My firewall had been running for months without problems, until one day it crashed. I thought oh well, maybe it hit an obscure kernel bug. Rebooted it. Several days later it crashed again. Rebooted it again. The next time it crashed but didn't boot again as it had corrupted its disk, and I had a really fun day reinstalling it with no internet connection.
My current box takes DDR2 800. I was going to get ECC, but they only had DDR2 666 available. I decided to go with the normal stuff. It passed memtest86, and I started installing gentoo. After a few hours bizarre compilation errors started to happen, and gcc started segfaulting. Turns out the RAM was very slightly bad. So slightly that I had to test it for 10 hours straight to see the problem.
Went back to the shop, exchanged it for the slower ECC. Can't see any noticeable performance difference, and it's rock solid now. It really irks me how for some bizarre reason there's error correction everywhere, on every kind storage media, except RAM, where for some reason it's an "enterprise feature".
That it definitely doesn't do.
I have my box freeze once in a while due to RAM exhaustion. It goes into swap, and swaps so heavily I can't even move the mouse around anymore. The problem with swap these days is that it's horribly slow compared to RAM. While memory keeps getting faster and larger, the hard disk's speed doesn't grow anywhere that fast. So once you're swapping and using a significat amount of it (just 256MB say) everything grinds to a halt.
Current OSes seem to try too hard to keep stuff running. That might have made sense in the past, but I don't want that anymore. If something goes insane and tries to get all the RAM for itself I want it to die, quickly, without freezing the machine solid for several minutes.
If you're planning to acquire a whole sim, then you can get a nice discount on it.
I'm not sure if a whole sim would be needed though, but they have advantages over normal land as you get a lot more control over them.
It took me a lot of grinding, but I finally levelled up in my scripting skill!
Seriously, SL is just as much of a game as IRC or Slashdot is. You can play a game through it (by posting chess moves for instance), but that doesn't really make it a game in itself.
So what exactly would make you feel so superior? I would hope that the prestige of one's education is based on its quality, rather than on where you got it.
Here in Spain we have the UNED, a distance university. I think it's the spanish university with the highest number of students. You can pretty much study everything at home, although newsgroups and forums are available. By your logic, acquiring an education in USENET and web boards must be really funny as well.
If they decided to open a place in SL, it wouldn't stop being what it currently is, would it? The exam certainly wouldn't get any easier. While I have my doubts regarding whether SL would be an improvement over forums, I think such a thing could be tried quite successfully.
Chat can be archived, you can access a log in SL (ctrl+H), and it can be moderated. If you own the land you're speaking on you can kick people out, ban them, or make the land accessible only to people on the allow list.
You can make your talk be in a place that's not distracting, and people don't mess with their avatars all that much. I've looked the same for more than half a year so far.
Are you joking? That switch was one of the main reasons why apache2 wasn't adopted for a very long time, as the myriad of PHP extensions wouldn't work because they weren't thread safe.
In fact, I didn't upgrade to apache2 until I found out that threads can be completely removed from it, so that it works the same way as apache1 did: Might be a bit slower, but I'll very gladly sacrifice a tiny bit of performance for reliability.
What you say would be true if money had the same value to everybody. Let's say that we have $1000 to split. You make the division: You take $950 for yourself, and give me $50. Now, I'm not that poor that $50 is a big matter to me, I might as well derive more satisfaction from ensuring you don't get $950 for being a bastard than from getting $50.
A prerequisite for person B to say "no" would be that the amount of money they'd get would be below their threshold. Maybe they're filthy rich and even $10000 would be a drop in the bucket.
Reasons why person B may say "no":
Note that all of those, even the last one are perfectly selfish.
Only problem there is that I'm not a large customer, just a normal guy with a server in the closet
That has some problems.
It'd require an extra disk, which I don't really need right now as I have enough space. Overall it'd be less reliable (3 drives are more likely to fail than 2). It's also a suboptimal configuration for a database server. And it's a lot more error prone, as any component of a RAID-1 can be used independently if needed, while that can't be done with a RAID-5.
My concern is that data can't be recovered from a broken disk sent for replacement, but while it's here I'd like to have as much convenience as possible.
That would need to be the heck of a magnet. From what I hear, a magnet strong enough would bend the platters. If I had the cash and right place (I'm definitely not going to use something like that anywhere near anything valuable) for that sort of thing I wouldn't really worry about getting warranty replacements.
I'd like to have a system where all disks are encrypted, but a password isn't needed to mount them.
Problem: I run a server with a RAID-1 setup. After a few months a disk fails. I remove it and want to get it replaced under warranty. The problem is that the disk isn't in a good enough condition to be able to fully overwrite it, and something sensitive could remain in some obscure area, like reallocated sectors. Server stores a quite large amount of rather private data I feel really uncomfortable letting go. What if it gets fixed and somebody else ends up recovering my stuff?
Solution idea: Make the box boot from an alternate media, such as a CD or a flash drive that would contain kernel, initrd and keys. Whole hard disk would go through dmcrypt, then RAID, then LVM. If disk fails, it can be removed and exchanged without leaving anything readable on it. If the boot media is removed any stored data is quickly made inaccessible.
I'd really like to see something like this being offered as an advanced install option in a distribution.
This should be quite doable right now with current technology, except for the stuff that hasn't been released yet.
You're kidding, right? There are many places where power can be saved. Example:
Sure it is, after all the whole planet is solar powered.
Also, I don't get what's the deal with "windmills chop birds". Yes, they do sometimes. But birds fly into skyscrapers quite often too, and for some reason that doesn't seem to stop anybody from building them. There are also types of windmills that rotate slower, and are less dangerous to birds.
Now, I have absolutely nothing against nuclear power, if well done, but what you say simply isn't true. Other viable alternatives exist as well, and nuclear isn't the only option.
Why WoW? There are more things out there. I bet quite a few people would like to use this in Second Life. Make an avatar based on yourself first, then modify it to perfection.
Yeah, as if people are going to use trigonometry to build shapes. It's a sort of neat idea, I suppose, but doesn't seem to be very suitable for the general SL population.
IIRC, the shape constraints in SL are mostly imposed by Havok, the physics engine they use.
SL users also don't really have much power to spare: right now I'm getting 20 FPS, on a dual Athlon 64 X2 5200+, with a GeForce 7900GS. For most people, SL is seriously slow and they wouldn't like extra load to be added to the client.
In any case, the main problem in SL right now is scalability. Nobody really cares much how shapes generated in comparison to that, and people manage to make really neat things with them anyway.
Err, why exactly? The whole point of SL is that it doesn't have the restrictions of RL, otherwise what fun it would be?
Incidentally, from what I hear it did work the way you say in the past, and you had to fly to get anywhere. Doesn't work that way anymore, and since the previous system was abandoned, they're unlikely to return to it.
In SL, land is far from infinite. Land is mapped directly to server resources. If you buy your own island (65536 m^2), you're effectively paying for a CPU on a server to be dedicated to you. Considering what servers cost, they're not overcharging much.
While land could be infinite, in the sense that you'd pay for resources consumed instead of chunks of space, it currently doesn't work that way in SL, although I've heard rumors of that they'd like it to work eventually that way. Then you could have all the land you'd like, but you'd have to pay for your CPU usage. In that case I imagine you'd pay depending on the amount of server time your scripts use, the number of objects you have, etc.
Assumption is mostly correct, except with one thing, you're not really paying 70 dollars, you're paying almost nothing. Shortly, most of what you pay actually pays for your stipend, as if you were regularly buying it at the Linden Exchange. So you can recover most of that.
As a fairly old resident, I keep my L$500/week stipend for my premium account, which earns me about $16 a year. It can't get much better than having the company paying you instead of the other way. In fact, I wrote that blog entry because I was tired of all the complaints due to stipend reduction. They were needed, or it wouldn't make sense to have premium accounts in the first place.
My understanding is that primitives are a requirement if you want physics. Calculating the physics of objects with an arbitrary shape would be seriously complicated, while it's trivial to calculate say, the volume of a sphere.
Regarding advanced graphical features, LL are adding reflections, but really, I don't think most people at the time want LL to work on scalability. Everybody I see is saying "scalability first, everything else second". So massive graphical improvements are probably going to come from other people. Somebody on the mailing list mentioned making a patch for a stereo version of SL, which sounds very cool, but I haven't seen it yet.
Unfortunately, IRL I can't be blue and furry, so for now I'll have to conform with SL ;-)
For me it's mostly like 3D IRC. In the places I visit, advertising is pretty much inexistent.
Very well, there are too many. What are you going to do about it?
You can complain until you're blue in the face, but the thing is, the authors of those distros don't give a damn you think there are too many. They work on them because they like it. I had an idea for a distro myself once, and while it didn't materialize due to the lack of time, I can tell you I'd do such a thing only for my own enjoyment. It's absolutely not a matter of what "we" (whoever that is) need, it's a matter of what I want to do. Your opinion about the amount of distros being excessive doesn't enter into it.
Like I said, it's a pointless discussion. You can sit here and whine all you want, but really, nobody responsible cares, so nothing is going to come out of it.