it does... well, everything. From inside it I can...
0) read Web Tools Review 1) check email 2) autorespond to email 3) compile stuff 4) debug stuff in DDD 5) check stuff out of/into CVS, and 6) play Tetris.
what else would you use a computer for? it takes some getting used to, however. (XEmacs)
Shortly I should be upgrading my hoary old dual PPro to them and upping the RAM to 128MB. I decided the cost of getting old memory is less than that of a new computer, and the box is a ROCK... no complaints, EVER.
The reason you might not want to cluster is that you have to buy (n * largest memory image you run) where n is the number of nodes. Of course, you get n-fold availability if you monitor and failover the nodes, but you pay for it.
With an SMP box, you buy your RAM for all n processors in the box. The OS and hardware had better be up to snuff if you're going to run large-scale SMP; ensuring this is not cheap either.
AFAIK the cheapest way to do 4-way SMP is with PPros. You may be able to substitute PII-Overdrive processors on a PPro mobo and drive down the price since the PII-OD is effectively a slow Xeon. But then you might have to shop for (expensive and slow) nonstandard memory.
A compilation of the results for the first week or so, plotted by movement and initial score, would be interesting... more so if there were some easy way to correlate movement up or down (baseline or average score) by user and plot that too.
> I've installed Redhat, Win3.1, Win95, Win98, > WinNT. Redhat loses by a wide margin.
You're weird. For nonstandard hardware, NT is much more painful to install.
I wouldn't know about Win95 or Win98 because it's been so long since I used an OS without preemptive multitasking that my memories would be inaccurate.
But I don't come to the same conclusion. My father, for example, really likes how fast Linux is and how he can choose how it should look by picking a different window manager. Yet he's having a hell of a time installing LinuxPPC with ApplixWare. Now, granted, it's probably best to wait for R5 or get R5beta at this point, and yeah, Apple loves to put crappy nonstandard hardware in their machines, but I'm a little pissed that I can't remember exactly what all I had to do to get stuff peachy on my PPC boxes. Of course I'm installing MkLinux on a PowerBook 5300ce shortly so it's sure to come rushing back, accompanied by waves of nausea. And it won't see my network!
Anyways, the point is that MkLinux and LinuxPPC installs suck rotting crotch. They're awful. The only people who end up running Linux on PPCs are either hardcore, or are willing to sweat blood while they become hardcore. I've met many, many system administrators -- competent ones -- who would just as soon run NT as go through that.
That's really *NOT* the way to get support for running such a system at work.
What is better is a gentle but firm reminder to people that ask really, really stupid questions to RTFM, and an intuitive way to present said FM. I hate dealing with users, and I like writing elegant code that does cool and/or magical things. But in order to get the users to %$@! off, I have to put stuff where they can understand it.
That's really a lot more enjoyable way to have things, set up so that ignorant people can educate themselves and lazy people can be gently scolded into helping themselves.
Neal Stephenson's essay had a section detailing the usage of the Hole Hawg drill. I refer you to said passage and then ask, "Do you want your grandmother trying to wield a Hole Hawg, or would it be better if there was a safe way to use that power?". Be has such a system, though it doesn't really work (yet). Linux ought to have one too, not just because more people will be able to use it, but because we can probably get it done.
And it will really piss off die-hard Windows users when everything becomes as easy to do on Linux as on Windows. At that point Linux will have become the Universal Operating System and everyone can go work on more interesting stuff.
Only problem is, when you split development costs 15 million ways, it's pretty cheap to maintain an OS, but when you only split it among 1000 or 100000 people, it's much more expensive. Same thing for the chips.
Personally I don't care so much about the chip as the OS; SGI has always ruled because of raw, seething, furious I/O and memory bandwidth. If they support Linux on their machines (preferably all of them) we all win -- an alternative to poopy NT and kick-ass I/O at (relatively) low prices.
SGI no longer enjoys their position as "The" graphics workstation and hence can't gouge for them as much, so they can't reasonably support low-end development. If someone buys an Origin or an Onyx, suddenly there's a lot of money to spend on development again.
So Linux on the low end (as soon as there is support for hardware-accelerated X) and Irix on the high end is about all SGI can be expected to offer, given the razor-thin margins for lower-end machines. I mean, do you see Dell doing any OS development? IBM (on the low end)? HP (low end)?
Linux is about all the Unix workstation makers can afford these days;-).
Oh, ps. I have a couple of PPC boxes and will probably buy an old Sparc or Alpha one of these days. Maybe even a real old MIPS R3000, that's the only kind of assembler I can stomach. (well, x86, but that's a necessary and rare evil)
Only problem is, when you split development costs 15 million ways, it's pretty cheap to maintain an OS, but when you only split it among 1000 or 100000 people, it's much more expensive. Same thing for the chips.
Personally I don't care so much about the chip as the OS; SGI has always ruled because of raw, seething, furious I/O and memory bandwidth. If they support Linux on their machines (preferably all of them) we all win -- an alternative to poopy NT and kick-ass I/O at (relatively) low prices.
SGI no longer enjoys their position as "The" graphics workstation and hence can't gouge for them as much, so they can't reasonably support low-end development. If someone buys an Origin or an Onyx, suddenly there's a lot of money to spend on development again.
So Linux on the low end (as soon as there is support for hardware-accelerated X) and Irix on the high end is about all SGI can be expected to offer, given the razor-thin margins for lower-end machines. I mean, do you see Dell doing any OS development? IBM (on the low end)? HP (low end)?
Linux is about all the Unix workstation makers can afford these days;-).
piss on RFC 1945, vote for RFC 761!
on
Tuesday Quickies
·
· Score: 2
The original TCP standard... don't leave home without it. Besides, Postel authored it.
used Indigos for $2K and SparcBooks for $600
on
Tuesday Quickies
·
· Score: 1
what's an ex-Unix-lover to do?
I saw somewhere that older NeXT cubes were going for $700 or less... that's kind of cool. I think it was in Performance Computing (nee Unix Review or something like that) where they had a big article on running older Unix systems for cheap.
I think I'd rather just have the monitor from one of the Maximum Impact systems and an adapter card to use all 1600x1200 pixels... mmmm, decadence.
And who is this lunatic talking about "lesser CPUs" and "inferior operating systems"? Feel free to buy an Origin 2000 with 128 R10K's, and support the development of Irix. What's that? You can't afford to? You can only buy an NT box like everyone else?
Hmm, what a coincidence -- SGI's going where the buyers are. Granted, they seem to be rather slow with the hardware-accelerated X support, but we'll have to wait and see... it is rather pointless to start flaming them this early on.
Any site that has a "support" address and a form to register on is just begging for a feedback loop. One of these days I'm going to write a bot to search out and "enlighten" these silly people.
Either that or I'll just compile a list of places that ask me for my email address and pound them with a bot for a week.
I can't think of a better way to get the point across ("look, if I wanted your junk mail I would have said so!") than to let them deal with their own. Zones.com will probably be first to go...
CMU, MIT, Cornell, and NCSU all run Kerberos for authenticating students, and that's just the ones I know of. If you want to use Kerberos people really need to know why their tickets expire, how to resolve problems with the Kerberos servers, etc., but it is a good large-scale system IMHO. Kerberos encrypts all authenticated communications so a sniffer won't do much damage if you get everyone to authenticate via kerberos. Same sort of principle as SSH, different implementation.
As someone who wandered out of the big-corporate, big-university environment into a small company, I have recently begun to deal with the (pathetic) security precautions which seem to be commonplace outside of large computing centers. One irritating thing is that if something's too technical or inconvenient, no one will use it, and if it blocks access to something important, they'll also complain. So you have to weigh the amount of worry time against the amount of support time you want to spend (IMHO). I can't guarantee that everyone's data is secure against every attack, but I can at least point out obvious holes and good ways to plug them (thanks, IBM).
nmap and SAINT are awesome tools, btw. There is a nice article on using them in this month's SysAdmin magazine, and O'Reilly even has a book on customizing SATAN for your own needs.
I want to take a look at the Deception toolkit and Secure Mailer as well; people have repeatedly stated that these are great. (i.e. securemailer + qmail as a relay + gateway, versus sendmail as both; and dtk just in general) But I haven't had time yet.
Anyone using a Zip drive and a printer on par0?
on
Linux 2.2.4
·
· Score: 1
Nah, I'm running 2.2.4, it's just that I have a local printer passed-through my zip drive on the same port, and the printer isn't working. It's not a big deal I guess, I can just print and backup at different times.
Thanks!
ps. 2.2.4 is noticeably snappier for me, and I haven't even re-upgraded my pgcc-ized version of XFree86 since I switched to Debian... I should do that too. Hell, right now I'm on the phone with Oracle to see whether they'll support our server if I put 2.2.4 on there too.
Anyone using a Zip drive and a printer on par0?
on
Linux 2.2.4
·
· Score: 1
I've had some confusion getting the printer to print and the zip drive to zip, although I thought I followed the directions pretty carefully (rmmod lp, insmod ppa, insmod lp). The zip drive works but I can't print.
If anyone else chains their printer onto their zip drive's pass-through port, I'd greatly appreciate any bonehead tips ("one at a time, bozo!") or other advice. I'd like to be able to print and use the zip drive without constantly shuffling cables, and the documentation indicates that I should be able to do so... but I can't so far.
Is it just a matter of editing conf.modules? (I wasn't sure whether the IRQ and io arguments in parport.txt were generic or "[your_IRQ]...")
and is published / will be published by O'Reilly... y'know, Right Tool For The Job. GTK was written for GUI programming in C. Gtk-- is at 1.0 and I even installed it, but I hain't tried writing anything funky with it.
I was monkeying around with the FLTK a while ago and it seemed wonderful, too. PyGTK is my new favorite RAD language, though. Slap things together in Glade, use glade.py to Pythonize them, and bind your events to Python functions. Neat!
My feeling is that it is very important. I run FreeBSD, Debian, and RedHat; I'd very much like for RedHat and Debian to be more compatible, as I imagine many people would like Slackware to be, etc. There's no reason for gratuitous incompatibility -- RedHat is a bunch of nice people, but sometimes their decisions for where to put things and what to include make little sense.
A common starting point would be very good for Linux, and would really stuff it to the "fragmentation FUDrakers" once and for all.
Slashdot Mirror
yeah, there's one called XEmacs...
it does... well, everything. From inside it I can...
0) read Web Tools Review
1) check email
2) autorespond to email
3) compile stuff
4) debug stuff in DDD
5) check stuff out of/into CVS, and
6) play Tetris.
what else would you use a computer for?
it takes some getting used to, however. (XEmacs)
It's a Pentium II OD and they're around $400-500.
;-)
That's a significant savings over a Xeon...
Shortly I should be upgrading my hoary old dual PPro to them and upping the RAM to 128MB. I decided the cost of getting old memory is less than that of a new computer, and the box is a ROCK... no complaints, EVER.
The reason you might not want to cluster is that you have to buy (n * largest memory image you run) where n is the number of nodes. Of course, you get n-fold availability if you monitor and failover the nodes, but you pay for it.
With an SMP box, you buy your RAM for all n processors in the box. The OS and hardware had better be up to snuff if you're going to run large-scale SMP; ensuring this is not cheap either.
AFAIK the cheapest way to do 4-way SMP is with PPros. You may be able to substitute PII-Overdrive processors on a PPro mobo and drive down the price since the PII-OD is effectively a slow Xeon. But then you might have to shop for (expensive and slow) nonstandard memory.
A compilation of the results for the first week or so, plotted by movement and initial score, would be interesting... more so if there were some easy way to correlate movement up or down (baseline or average score) by user and plot that too.
> I've installed Redhat, Win3.1, Win95, Win98,
> WinNT. Redhat loses by a wide margin.
You're weird. For nonstandard hardware, NT is much more painful to install.
I wouldn't know about Win95 or Win98 because it's been so long since I used an OS without preemptive multitasking that my memories would be inaccurate.
that would be SGI then, I assume?
Since their servers are mostly O2Ks last time I checked.
> Perhaps there is a moral in there somewhere.
Every moral has a story, every story has an end...
What was I saying?
(eg. MkLinux, minimal sysadmin duties, etc)
But I don't come to the same conclusion. My father, for example, really likes how fast Linux is and how he can choose how it should look by picking a different window manager. Yet he's having a hell of a time installing LinuxPPC with ApplixWare. Now, granted, it's probably best to wait for R5 or get R5beta at this point, and yeah, Apple loves to put crappy nonstandard hardware in their machines, but I'm a little pissed that I can't remember exactly what all I had to do to get stuff peachy on my PPC boxes. Of course I'm installing MkLinux on a PowerBook 5300ce shortly so it's sure to come rushing back, accompanied by waves of nausea. And it won't see my network!
Anyways, the point is that MkLinux and LinuxPPC installs suck rotting crotch. They're awful. The only people who end up running Linux on PPCs are either hardcore, or are willing to sweat blood while they become hardcore. I've met many, many system administrators -- competent ones -- who would just as soon run NT as go through that.
That's really *NOT* the way to get support for running such a system at work.
What is better is a gentle but firm reminder to people that ask really, really stupid questions to RTFM, and an intuitive way to present said FM. I hate dealing with users, and I like writing elegant code that does cool and/or magical things. But in order to get the users to %$@! off, I have to put stuff where they can understand it.
That's really a lot more enjoyable way to have things, set up so that ignorant people can educate themselves and lazy people can be gently scolded into helping themselves.
Neal Stephenson's essay had a section detailing the usage of the Hole Hawg drill. I refer you to said passage and then ask, "Do you want your grandmother trying to wield a Hole Hawg, or would it be better if there was a safe way to use that power?". Be has such a system, though it doesn't really work (yet). Linux ought to have one too, not just because more people will be able to use it, but because we can probably get it done.
And it will really piss off die-hard Windows users when everything becomes as easy to do on Linux as on Windows. At that point Linux will have become the Universal Operating System and everyone can go work on more interesting stuff.
64 bits beats 32 bits any day...
;-).
Only problem is, when you split development costs 15 million ways, it's pretty cheap to maintain an OS, but when you only split it among 1000 or 100000 people, it's much more expensive. Same thing for the chips.
Personally I don't care so much about the chip as the OS; SGI has always ruled because of raw, seething, furious I/O and memory bandwidth. If they support Linux on their machines (preferably all of them) we all win -- an alternative to poopy NT and kick-ass I/O at (relatively) low prices.
SGI no longer enjoys their position as "The" graphics workstation and hence can't gouge for them as much, so they can't reasonably support low-end development. If someone buys an Origin or an Onyx, suddenly there's a lot of money to spend on development again.
So Linux on the low end (as soon as there is support for hardware-accelerated X) and Irix on the high end is about all SGI can be expected to offer, given the razor-thin margins for lower-end machines. I mean, do you see Dell doing any OS development? IBM (on the low end)? HP (low end)?
Linux is about all the Unix workstation makers can afford these days
Oh, ps. I have a couple of PPC boxes and will probably buy an old Sparc or Alpha one of these days. Maybe even a real old MIPS R3000, that's the only kind of assembler I can stomach. (well, x86, but that's a necessary and rare evil)
64 bits beats 32 bits any day...
;-).
Only problem is, when you split development costs 15 million ways, it's pretty cheap to maintain an OS, but when you only split it among 1000 or 100000 people, it's much more expensive. Same thing for the chips.
Personally I don't care so much about the chip as the OS; SGI has always ruled because of raw, seething, furious I/O and memory bandwidth. If they support Linux on their machines (preferably all of them) we all win -- an alternative to poopy NT and kick-ass I/O at (relatively) low prices.
SGI no longer enjoys their position as "The" graphics workstation and hence can't gouge for them as much, so they can't reasonably support low-end development. If someone buys an Origin or an Onyx, suddenly there's a lot of money to spend on development again.
So Linux on the low end (as soon as there is support for hardware-accelerated X) and Irix on the high end is about all SGI can be expected to offer, given the razor-thin margins for lower-end machines. I mean, do you see Dell doing any OS development? IBM (on the low end)? HP (low end)?
Linux is about all the Unix workstation makers can afford these days
It was merely the DoD standardizing of TCP.
But Postel wrote it up anyways.
The original TCP standard... don't leave home without it. Besides, Postel authored it.
what's an ex-Unix-lover to do?
I saw somewhere that older NeXT cubes were going for $700 or less... that's kind of cool. I think it was in Performance Computing (nee Unix Review or something like that) where they had a big article on running older Unix systems for cheap.
I think I'd rather just have the monitor from one of the Maximum Impact systems and an adapter card to use all 1600x1200 pixels... mmmm, decadence.
Well, that, and stereo-in-a-window on Linux.
I liked the old one way better!
And who is this lunatic talking about "lesser CPUs" and "inferior operating systems"? Feel free to buy an Origin 2000 with 128 R10K's, and support the development of Irix. What's that? You can't afford to? You can only buy an NT box like everyone else?
Hmm, what a coincidence -- SGI's going where the buyers are. Granted, they seem to be rather slow with the hardware-accelerated X support, but we'll have to wait and see... it is rather pointless to start flaming them this early on.
>>...we may someday find ourselves with nobody to hate.
Forgive me for being a wide-eyed hippy kook, but:
1) there's no shortage of Bad Guys -- see also Oracle
2) if we have nobody to hate, is that not a GOOD thing?
I thought the point was to have an enjoyable, reliable platform that everyone could share.
Any site that has a "support" address and a form to register on is just begging for a feedback loop. One of these days I'm going to write a bot to search out and "enlighten" these silly people.
Either that or I'll just compile a list of places that ask me for my email address and pound them with a bot for a week.
I can't think of a better way to get the point across ("look, if I wanted your junk mail I would have said so!") than to let them deal with their own. Zones.com will probably be first to go...
Yeah, isn't the NT BootMangler precious?
Love that interoperability.
>I don't really think that we need to see the proof of it.
Yeah, when I have a contest for a million bucks, I'll be sure and hand it out to whoever says they won. I mean, they could have, right?
Not only did Microsoft not publish a benchmark for the whole query, they routinely lie, cheat, and otherwise manipulate results.
I wouldn't trust a Microserf if they were pinned down by a UE10K. Which, incidentally, ought to clobber an 8-way Xeon without any trouble.
Methinks Microsoft is (as usual) heavily constipated.
CMU, MIT, Cornell, and NCSU all run Kerberos for authenticating students, and that's just the ones I know of. If you want to use Kerberos people really need to know why their tickets expire, how to resolve problems with the Kerberos servers, etc., but it is a good large-scale system IMHO. Kerberos encrypts all authenticated communications so a sniffer won't do much damage if you get everyone to authenticate via kerberos. Same sort of principle as SSH, different implementation.
As someone who wandered out of the big-corporate, big-university environment into a small company, I have recently begun to deal with the (pathetic) security precautions which seem to be commonplace outside of large computing centers. One irritating thing is that if something's too technical or inconvenient, no one will use it, and if it blocks access to something important, they'll also complain. So you have to weigh the amount of worry time against the amount of support time you want to spend (IMHO). I can't guarantee that everyone's data is secure against every attack, but I can at least point out obvious holes and good ways to plug them (thanks, IBM).
nmap and SAINT are awesome tools, btw. There is a nice article on using them in this month's SysAdmin magazine, and O'Reilly even has a book on customizing SATAN for your own needs.
I want to take a look at the Deception toolkit and Secure Mailer as well; people have repeatedly stated that these are great. (i.e. securemailer + qmail as a relay + gateway, versus sendmail as both; and dtk just in general) But I haven't had time yet.
Nah, I'm running 2.2.4, it's just that I have a local printer passed-through my zip drive on the same port, and the printer isn't working. It's not a big deal I guess, I can just print and backup at different times.
Thanks!
ps. 2.2.4 is noticeably snappier for me, and I haven't even re-upgraded my pgcc-ized version of XFree86 since I switched to Debian... I should do that too. Hell, right now I'm on the phone with Oracle to see whether they'll support our server if I put 2.2.4 on there too.
I've had some confusion getting the printer to print and the zip drive to zip, although I thought I followed the directions pretty carefully (rmmod lp, insmod ppa, insmod lp). The zip drive works but I can't print.
If anyone else chains their printer onto their zip drive's pass-through port, I'd greatly appreciate any bonehead tips ("one at a time, bozo!") or other advice. I'd like to be able to print and use the zip drive without constantly shuffling cables, and the documentation indicates that I should be able to do so... but I can't so far.
Is it just a matter of editing conf.modules?
(I wasn't sure whether the IRQ and io arguments in parport.txt were generic or "[your_IRQ]...")
not everyone who might be interested in kernel updates reads linux today.
It's a lot more interesting to me than hearing about the latest start wars or whatever.
and is published / will be published by O'Reilly... y'know, Right Tool For The Job. GTK was written for GUI programming in C. Gtk-- is at 1.0 and I even installed it, but I hain't tried writing anything funky with it.
I was monkeying around with the FLTK a while ago and it seemed wonderful, too. PyGTK is my new favorite RAD language, though. Slap things together in Glade, use glade.py to Pythonize them, and bind your events to Python functions. Neat!
LSB Home Page
My feeling is that it is very important. I run FreeBSD, Debian, and RedHat; I'd very much like for RedHat and Debian to be more compatible, as I imagine many people would like Slackware to be, etc. There's no reason for gratuitous incompatibility -- RedHat is a bunch of nice people, but sometimes their decisions for where to put things and what to include make little sense.
A common starting point would be very good for Linux, and would really stuff it to the "fragmentation FUDrakers" once and for all.
I'm at work, using Linux on the server and on my desktop. So tone it down, ok? Not everyone has to be a pushover to get a job.