I'd really hate to see how this "I'm better than you attitude" plays out in real life where it means jack how epic your loot is
He must really not get out much, if he hasn't noticed the abundance of rice racers / SUVs with 24" rims / Harleys with gold-plated trim and covers / etc. rolling around. Rolexes, Izod shirts, Ping clubs[1], Porsches — all these things are emblems of attitude. The concept of "epic loot" is alive and well, and has been since long before WoW coined the phrase.
My first thought when I read this was that this might be good for some kind of secure data transfer, but they can't seriously be considering this for consumers. Then I realized I was an idiot, and lots of games either are or will eventually be shipping on DVD, so maybe that's the target market. It'd probably still have to be PC-only games, though, as I can't see someone slotting a disk on their console and then standing there plugging and unplugging a dongle. Maybe if there was a USB port on the controller, but that seems like a classic chicken-and-egg situation.
If the encryption's sufficiently hard, and the key length is sufficiently long (and with even a 256MB USB key, that shouldn't be too difficult), then maybe the military or intelligence communities would be interested, but I'm guessing they already have something similar.
Re:Red Hat has no worries with this
on
Oracle Linux?
·
· Score: 1
1) Automatic memory managment 2) Automatic storage managment 3) Automatic segment managment 4) Automatic tablespace maagment 5) automtic DDM
and the list goes on and on and on
Heh, probably "automatic" as in "we removed the tools and hid the config parameters". So now consultants/DBAs who can still figure out how to do this can command 3x their current rates, because ordinary DBAs don't have a prayer of fixing this when it goes wrong.
I'm reminded of something similar with SQL Server, something like you don't have to rebuild index files, it's automatic. However, if one get corrupted, you don't have the option/tools to rebuild it. Such a deal!
Re:Definitely has uses but..
on
Oracle Linux?
·
· Score: 1
I'd imagine not. Companies that use Oracle usually wind up buying enough hardware that they establish good relationships with hardware vendors, and need enough expertise that they hire their own DBAs. They're not likely to dump their vendor relationships and lay off highly critical employees without some serious guaranteed payback.
The only companies who would be interested in such things would be companies transtioning from small to mid-sized status. I doubt there's enough of them to generate enough revenue to make those offerings profitable for Oracle.
Each major revision of Oracle...obsoleted our systems EVERY TIME
That's interesting, considering the installation manual for 8i on SPARC Solaris only requires 128M of memory, twice the amount of system memory available as swap space, and a CD-ROM drive. I would imagine that the requirements for AIX or HP-UX were similarly onerous. Was your systems group just trying to get by with SuSE with an old laptop or something? The Oracle 9i Migration Guide explicitly says that if you don't change your configuration, you don't need to upgrade your hardware when you move from 8i to 9i.
Sounds like somebody just wanted some new toys, and used Oracle upgrades as an excuse to get them...
Re:Definitely has uses but..
on
Oracle Linux?
·
· Score: 2, Informative
I would guess that they would offer a complete package that has Oracle running with Linux pre-configured to run Oracle. The idea would be that nothing else would be run on that box, except perhaps for a few utilities the customer run to monitor, backup, etc.
In that case, why go with Linux? I'd think they'd want some kind of a minimal RTOS: a scheduler, demand-paged VM, TCP/IP stack and a simple filesystem. Basically all it's going to do is switch among Oracle threads and a network daemon, and hammer the disk. Why have a complete multiuser environment for running (essentially) an embedded program?
If they provided (or closely specced) the hardware they could even get old-school on it and hand-write assembler code for the critical parts of the disk driver or network stack. Anyone know if Oracle's raw device I/O is faster than a good RAID set-up?
I got to a client site early Saturday afternoon to move my app into their QA environment. I installed the app, but when I tried to configure the database connection, I couldn't get through. I noticed I also couldn't get through from the test app, which had been working fine the day before. So I try to connect via the command line — nada. Ping? Nope. Called Operations, told them I couldn't find BBCPL0BT (whatever...) and could he check on it? Guy says he'll call me back. Fifteen minutes later, he does. Op: Found your box. Me: Great, when will it be back up? I need to get this app set up for $GROUP so it'll be ready Monday. Op: It should be up by, ummmm...Tuesday, it says here. Me: !!! TUESDAY !!! Why so long? Op: It's on a truck.
Turns out they were migrating their data center and the group I was working for hadn't been paying attention to which machines were moving when. At least the explanation was good enough for the client, who was very serious about holding me to a fixed delivery date and not so serious about agreeing to a fixed feature set...
Re:OpenSolaris?
on
Oracle Linux?
·
· Score: 3, Informative
If they want their own OS, they're probably going to want something that'll support clustering and a fast file system. Currently GPFS is the top dog in that area, and it's only available (currently) for AIX and Linux. It'd probably make more sense to put effort into improving this than porting it to Solaris.
Agreed that Solaris would provide more enterprise-grade (<—marketing term) features than Linux, although zones are becoming less compelling given the rise of virtualization, and I hear that ZFS doesn't provide the performance boost on SANs that it does on JBODs.
Thompson's criticism also presumes that the judge merely sat passively and watched. Who's to say he wasn't sitting there telling the player "Hey, that kid with the glasses! Punch him! Again! Kick him! Take his lunch money! YEAAAH!"
seriously whats up? I was looking to get a nice top of the line Dell system with Red Hat for our work server for months, but they offered nothing with the (then) top performing Athlon 64.
From what I can tell, Dell is positioning AMD as the head of its "value" line. In fact, I think most of their AMD offerings are still single-core. No idea when or if the four-processor server they announced back in May will ever ship. It was originally scheduled for "later in 2006", but even if it was going to ship in time for the holidays, I'd expect to see a little pre-release publicity by now.
AMD is staking a good part of their future on the high end server market, where Intel has never been a huge player.
Yes, despite providing a CPU designed by a high-end server maker (the Itanium was originally going to be the next generation of PA-RISC) and pumping hundreds of millions of dollars (and convincing [coercing?] others to throw in millions too) into marketing and development. It's not that Intel hasn't tried to score in the high-end server market, it's that it tried and failed miserably.
I was wondering if it was a subtle reference to "Zima", another allegedly genre-busting product that made an ineffective splash and has all but faded into the woodwork.
Actually, while I was writing this comment, I was thinking this would be a fun toy to write for an informal network. Just collect a group of friends, and pass all "suspicious" traffic (say any HTTP request that doesn't refer to the original page's server, or any graphics file requests) out to one of them randomly. They make the request, and pass the result back to you. You might even get civic-minded citizens setting up public servers to handle this sort of traffic.
Of course there's all kinds of separate issues this brings up, but if you really suspect you need protection from this sort of thing, it wouldn't be too hard to set up.
It's not because you want to.</clue> I pulled a couple of these at the worst job I ever had. Problem was, it was during the post-bubble meltdown, and the unemployment rate for software developers in my region was somewhere around 30%. So when you get laid off and your first baby's on the way, you take whatever you can get. The company worked us that hard because they could. They knew that as soon as the economy recovered, we'd be gone in a heartbeat (especially at the wages they were paying), so they got as much work out of us as they could. I finally left when they threatened to fire me if I took the vacation I had told them about two months before.
I guess I got off a little more lightly than some of my cow-orkers, though. They kept getting more and more stock options "in recognition of their hard work and dedication". Then the company was sold to (Earthlink?) and all their options became worthless. Kind of sums it up.
If you must follow a questionable URL of dubious provenance, consider actually using an OLDER browser version.
When I'm sufficiently paranoid, I actually use a minibrowser I wrote based on (frankly, a simple wrapper around) the Apache Jakarta HttpClient code. No javascript, no image loading, just "give me the html for http://slashdot.org/". It's really just a debugging tool, but sometimes you just want to know what people are really sending to your browser.
The next step would be an html feature to have images directly in the html
Actually, the next step would be to move to a proper document format, like PDF. People don't really care how their messages are encoded, all they want is to be able to put salutations in 48-pt Monotype Corsiva in putrid pink on a bright green background. It's just a shame you can't embed a looping audio track of a small child farting and laughing, which I fear will hinder the mass acceptance of PDF as a "family-friendly" mail format.
I'm not so sure that this is a trivial matter. One of the great advantages of Gmail is that I can access it from anywhere (anywhere I can get net access with a browser, anyway). Your assumption that a suitably-clued Thunderbird or other MUA is available isn't valid. And in this day and age of people e-mailing video clips and multi-megabyte PDFs about, the storage implications would be considerable, even given Google's capacity.
I think a better solution would be for each URL in every message to be rewritten to prepend "http://gmail.google.com/proxyFetch?uri=..." to the front. This would at least cause the request in the log file to point back to Google, instead of the user's actual machine. It doesn't address the issue of URLs with unique identifiers, but does at least cloak the user's location.
Is it just me, or is Gaiman just OK since he started writing novels? His stuff is interesting, but (IMHO) nowhere near as good as his work on the Sandman series. I've read Stardust and more recently American Gods (which I was really looking forward to), but neither of them were something I'd rush to recommend to anyone. He has interesting characters and good plot devices, but the writing itself just kind of lays there.
Maybe it's just me, American Gods did take home the Hugo...
Kinda, but not as bad as IBM's PC division. In the mid-90s I was using a SPARCstation ELC -- one of the early SPARC machines, and it could run rings around a '286. Problem was, the '486s were just coming out, and could wipe the floor with an ELC (especially the dic^Hskless ones we had). Sun still had the edge in graphics, as long as you could get by with four-bit greyscale. The ELC had a 17" monitor with 1152x900 resolution, back when cutting-edge PCs were lucky to have EGA (640x350, albeit with 16 colors).
More importantly, Sun had been around for awhile, and there was a lot of special software available for it. Linux was still really new, and Windows 95 hadn't come out yet (yet alone NT), so if you needed to do high-end circuit simulation or run the top-end software engineering toolkits, you bought a Sun. We used ours largely because they ran FrameMaker, a seriously wicked document creation tool. This software availability kept people buying Suns (and Auspex servers...) long after their performance was eclipsed by mainstream PCs.
That was my first thought: this OLPC is going to be bigger than Windows or cell phones? Pretty major hubris for an outfit that hasn't even started shipping in quantity yet!
486? 486?!?!? Back in my day (about the time dirt went beta), we used the trusty old TI 765. With real acoustic couplers, none of your fancy-pants alligator clips. And everything was PRINTED on PAPER, so you had a permanent record of your telnet session to ucbvax to prove you'd been there.
You kids with your microprocessors and your CRTs and computers you can lift — GET OFF MY ROCKS!
I think my WRT-54G allows me to force a disconnect for a particular connection. Don't most WAPs? ISTR I had to use this when I was trying to get a new laptop configured, and I wound up eating all my DHCP leases (I have it set to five to discourage mass leeching).
Offtopic question: do any consumer-grade WAPs support both WPA and WEP simultaneously? It's a hassle when my dad drops by, as his old laptop only handles WEP, and I've got all our systems at home configured to use WPA. Is there some technical (protocol-related?) reason why this can't be done?
Nope, won't work./dev/dsk entries are just symlinks to the actual block or character special files, and are definitely not readable or writable by user-mode programs.
oracle@fm-orc01 09:33:30> ls -l/dev/dsk/c0t0d0s2 lrwxrwxrwx 1 root root 49 May 4 13:40/dev/dsk/c0t0d0s2 ->../../devices/pci@0,0/pci-ide@1f,1/ide@0/sd@0,0:c oracle@fm-orc01 09:33:51> ls -l/devices/pci@0,0/pci-ide@1f,1/ide@0/sd@0,0:c brw-r----- 1 root sys 55, 2 Oct 12 09:34/devices/pci@0,0/pci-ide@1f,1/ide@0/sd@0,0:c oracle@fm-orc01 09:34:17> od -x/dev/dsk/c0t0d0s2 od: cannot open/dev/dsk/c0t0d0s2: Permission denied
Hmmm, I think I see your point, though — I should have said non-privileged user mode programs can't get access to the raw disk. Obviously, users running as root can open the raw drive and do whatever they want. Is that the catch under Windows, too — can only programs run as an Administrator open the raw disk? Obviously this is much more common than users running as root on Unix, but it'd still be nice if they restricted access at least that much.
Jesus, is this for real? User mode programs get raw access to the disk? WTF? I'm not buying the "Well, it would break disk editors" argument, if you're using a disk editor and have raw disk access, you can modify any on-disk data structures, which makes it merely an interesting exercise to bypass file security. Seriously, I studied OS design almost twenty years ago, and this was a known issue even back then. You don't allow users raw disk access, you don't allow user mode writes to system files, and you don't allow users to access devices they don't have rights to. I really can't believe they didn't fix this in NT, at the very least.
I didn't even bother to read the rest of the article. If a regular non-privileged user's got raw disk access, then they can patch and plug whatever the hell they want. Game over, just make a copy of NTOSKRNL.SYS, patch the fsck out of it (or into it, if that's yer karma), then just do a raw disk transfer: read a block, write a block, repeat.
Slashdot Mirror
He must really not get out much, if he hasn't noticed the abundance of rice racers / SUVs with 24" rims / Harleys with gold-plated trim and covers / etc. rolling around. Rolexes, Izod shirts, Ping clubs[1], Porsches — all these things are emblems of attitude. The concept of "epic loot" is alive and well, and has been since long before WoW coined the phrase.
[1] No, not cyber cafes...
My first thought when I read this was that this might be good for some kind of secure data transfer, but they can't seriously be considering this for consumers. Then I realized I was an idiot, and lots of games either are or will eventually be shipping on DVD, so maybe that's the target market. It'd probably still have to be PC-only games, though, as I can't see someone slotting a disk on their console and then standing there plugging and unplugging a dongle. Maybe if there was a USB port on the controller, but that seems like a classic chicken-and-egg situation.
If the encryption's sufficiently hard, and the key length is sufficiently long (and with even a 256MB USB key, that shouldn't be too difficult), then maybe the military or intelligence communities would be interested, but I'm guessing they already have something similar.
I'm reminded of something similar with SQL Server, something like you don't have to rebuild index files, it's automatic. However, if one get corrupted, you don't have the option/tools to rebuild it. Such a deal!
I'd imagine not. Companies that use Oracle usually wind up buying enough hardware that they establish good relationships with hardware vendors, and need enough expertise that they hire their own DBAs. They're not likely to dump their vendor relationships and lay off highly critical employees without some serious guaranteed payback.
The only companies who would be interested in such things would be companies transtioning from small to mid-sized status. I doubt there's enough of them to generate enough revenue to make those offerings profitable for Oracle.
Sounds like somebody just wanted some new toys, and used Oracle upgrades as an excuse to get them...
If they provided (or closely specced) the hardware they could even get old-school on it and hand-write assembler code for the critical parts of the disk driver or network stack. Anyone know if Oracle's raw device I/O is faster than a good RAID set-up?
I got to a client site early Saturday afternoon to move my app into their QA environment. I installed the app, but when I tried to configure the database connection, I couldn't get through. I noticed I also couldn't get through from the test app, which had been working fine the day before. So I try to connect via the command line — nada. Ping? Nope. Called Operations, told them I couldn't find BBCPL0BT (whatever...) and could he check on it? Guy says he'll call me back. Fifteen minutes later, he does.
Op: Found your box.
Me: Great, when will it be back up? I need to get this app set up for $GROUP so it'll be ready Monday.
Op: It should be up by, ummmm...Tuesday, it says here.
Me: !!! TUESDAY !!! Why so long?
Op: It's on a truck.
Turns out they were migrating their data center and the group I was working for hadn't been paying attention to which machines were moving when. At least the explanation was good enough for the client, who was very serious about holding me to a fixed delivery date and not so serious about agreeing to a fixed feature set...
If they want their own OS, they're probably going to want something that'll support clustering and a fast file system. Currently GPFS is the top dog in that area, and it's only available (currently) for AIX and Linux. It'd probably make more sense to put effort into improving this than porting it to Solaris.
Agreed that Solaris would provide more enterprise-grade (<—marketing term) features than Linux, although zones are becoming less compelling given the rise of virtualization, and I hear that ZFS doesn't provide the performance boost on SANs that it does on JBODs.
Thompson's criticism also presumes that the judge merely sat passively and watched. Who's to say he wasn't sitting there telling the player "Hey, that kid with the glasses! Punch him! Again! Kick him! Take his lunch money! YEAAAH!"
I was wondering if it was a subtle reference to "Zima", another allegedly genre-busting product that made an ineffective splash and has all but faded into the woodwork.
Actually, while I was writing this comment, I was thinking this would be a fun toy to write for an informal network. Just collect a group of friends, and pass all "suspicious" traffic (say any HTTP request that doesn't refer to the original page's server, or any graphics file requests) out to one of them randomly. They make the request, and pass the result back to you. You might even get civic-minded citizens setting up public servers to handle this sort of traffic.
Of course there's all kinds of separate issues this brings up, but if you really suspect you need protection from this sort of thing, it wouldn't be too hard to set up.
It's not because you want to.</clue> I pulled a couple of these at the worst job I ever had. Problem was, it was during the post-bubble meltdown, and the unemployment rate for software developers in my region was somewhere around 30%. So when you get laid off and your first baby's on the way, you take whatever you can get. The company worked us that hard because they could. They knew that as soon as the economy recovered, we'd be gone in a heartbeat (especially at the wages they were paying), so they got as much work out of us as they could. I finally left when they threatened to fire me if I took the vacation I had told them about two months before.
I guess I got off a little more lightly than some of my cow-orkers, though. They kept getting more and more stock options "in recognition of their hard work and dedication". Then the company was sold to (Earthlink?) and all their options became worthless. Kind of sums it up.
I'm not so sure that this is a trivial matter. One of the great advantages of Gmail is that I can access it from anywhere (anywhere I can get net access with a browser, anyway). Your assumption that a suitably-clued Thunderbird or other MUA is available isn't valid. And in this day and age of people e-mailing video clips and multi-megabyte PDFs about, the storage implications would be considerable, even given Google's capacity.
I think a better solution would be for each URL in every message to be rewritten to prepend "http://gmail.google.com/proxyFetch?uri=..." to the front. This would at least cause the request in the log file to point back to Google, instead of the user's actual machine. It doesn't address the issue of URLs with unique identifiers, but does at least cloak the user's location.
Is it just me, or is Gaiman just OK since he started writing novels? His stuff is interesting, but (IMHO) nowhere near as good as his work on the Sandman series. I've read Stardust and more recently American Gods (which I was really looking forward to), but neither of them were something I'd rush to recommend to anyone. He has interesting characters and good plot devices, but the writing itself just kind of lays there.
Maybe it's just me, American Gods did take home the Hugo...
More importantly, Sun had been around for awhile, and there was a lot of special software available for it. Linux was still really new, and Windows 95 hadn't come out yet (yet alone NT), so if you needed to do high-end circuit simulation or run the top-end software engineering toolkits, you bought a Sun. We used ours largely because they ran FrameMaker, a seriously wicked document creation tool. This software availability kept people buying Suns (and Auspex servers...) long after their performance was eclipsed by mainstream PCs.
That was my first thought: this OLPC is going to be bigger than Windows or cell phones? Pretty major hubris for an outfit that hasn't even started shipping in quantity yet!
You kids with your microprocessors and your CRTs and computers you can lift — GET OFF MY ROCKS!
I think my WRT-54G allows me to force a disconnect for a particular connection. Don't most WAPs? ISTR I had to use this when I was trying to get a new laptop configured, and I wound up eating all my DHCP leases (I have it set to five to discourage mass leeching).
Offtopic question: do any consumer-grade WAPs support both WPA and WEP simultaneously? It's a hassle when my dad drops by, as his old laptop only handles WEP, and I've got all our systems at home configured to use WPA. Is there some technical (protocol-related?) reason why this can't be done?
Nope, won't work. /dev/dsk entries are just symlinks to the actual block or character special files, and are definitely not readable or writable by user-mode programs.
/dev/dsk/c0t0d0s2 /dev/dsk/c0t0d0s2 -> ../../devices/pci@0,0/pci-ide@1f,1/ide@0/sd@0,0:c /devices/pci@0,0/pci-ide@1f,1/ide@0/sd@0,0:c /devices/pci@0,0/pci-ide@1f,1/ide@0/sd@0,0:c /dev/dsk/c0t0d0s2 /dev/dsk/c0t0d0s2: Permission denied
oracle@fm-orc01 09:33:30> ls -l
lrwxrwxrwx 1 root root 49 May 4 13:40
oracle@fm-orc01 09:33:51> ls -l
brw-r----- 1 root sys 55, 2 Oct 12 09:34
oracle@fm-orc01 09:34:17> od -x
od: cannot open
Hmmm, I think I see your point, though — I should have said non-privileged user mode programs can't get access to the raw disk. Obviously, users running as root can open the raw drive and do whatever they want. Is that the catch under Windows, too — can only programs run as an Administrator open the raw disk? Obviously this is much more common than users running as root on Unix, but it'd still be nice if they restricted access at least that much.
Jesus, is this for real? User mode programs get raw access to the disk? WTF? I'm not buying the "Well, it would break disk editors" argument, if you're using a disk editor and have raw disk access, you can modify any on-disk data structures, which makes it merely an interesting exercise to bypass file security. Seriously, I studied OS design almost twenty years ago, and this was a known issue even back then. You don't allow users raw disk access, you don't allow user mode writes to system files, and you don't allow users to access devices they don't have rights to. I really can't believe they didn't fix this in NT, at the very least.
I didn't even bother to read the rest of the article. If a regular non-privileged user's got raw disk access, then they can patch and plug whatever the hell they want. Game over, just make a copy of NTOSKRNL.SYS, patch the fsck out of it (or into it, if that's yer karma), then just do a raw disk transfer: read a block, write a block, repeat.
Man, I'm glad I'm on Solaris....