What was "state of the art" in the '80s is now ubiquitous and hidden from the end user.
Ever found a bug in your portable Nomad MP3 player (The flash-based ones, not any of the disk-based ones - Although the disk ones are still pretty strong)
Has your car ever shut down because its computer crashed? (Note: Hardware failure doesn't count, although automotive ECU failure is RARE unless you've done something to screw with its cooling.)
What about your VCR?
These are all cases of coding like you described - Fitting as much as possible into as little space as possible. In Lucent's (now Avaya) business communications division, there was (maybe still is) a raging debate on whether the usability benefits of using two LEDs rather than one justified the *pennies* of extra cost on an item that sold for a few hundred dollars. In a cost-cutting environment that intensive, you're not going to spec a processor with 8k of flash and 2k RAM when a processor with 2k/256 bytes will do. (Note - Popular microcontroller such as the Atmel AVR, Microchip PIC, Motorola 68HC11, etc. all are in this range.)
And it's quite easy for a single programmer to do all of this. I've seen CD-based MP3 players developed in a few weeks by a team of two college students for their Microcontrollers course taking 3-4 other classes at the same time.
I think I know why fewer people showed than I expected.
The location of the funeral ingame, Caer Benowyc, is a keep on the Albion frontier.
There are relatively nasty critters wandering around the area - Dangerous for people as high as 37 when solo, and probably dangerous for anyone under 25-30 even when escorted. (My guild was congratulating me for making it there solo w/o dying - Not easy.)
It's also the farthest keep out for Albion players, through some really dangerous terrain.
The only safe place to have the funeral would've been in Midgard, but that would've prevented anyone from Albion or Hibernia from participating.
I said the engine *SUPPORTS* high-res textures, not that he actually *USED* such textures in his download, and that the textures included with the download are, in fact, ugly. (Low-res and the bumpmaps/glossmaps are overdone)
I tried using some high-res textures/bumpmaps I found and the difference was amazing - MUCH better.
Maximum level in DAoC is 50. And I don't think they would've held a funeral near a Lv99 mob like the Dragon.
That said - I wish I'd known about this. I play DAoC and knew nothing. Of course, it WAS on a different server.
I'm honestly surprised at how few people showed up in that picture - That's on par with a single realm's RvR zerg, and word tends to spread REALLY fast throughout a realm in DAoC. If I'd been on the server in question (Pellinor, I play on Lancelot), I would've been there.
Strange thing was, I was at Beno last night... For the first time ever.
One other odd thing to note: It looks like the screenshot was made by someone not from the guild/realm holding the funeral, as the shot is almost all Midgard players but they have the generic "not from your realm" names. I also see a couple of generic Albion tags, so the shot must've been taken by a Hibernian.
The lighting effects are (in general) much neater looking.
In addition, he also implemented bumpmapping/glossmapping. Unfortunately the bumpmaps/glossmaps he's using in the demo are kind of... ugly. But the engine DOES support high-res TGA textures/bumpmaps/glossmaps. I tried downloading a few high-res textures/bumpmaps from a texture archive and slapping them into a test map - The results were... Amazing. The engine is NOWHERE close to its full potential.
(I'd post the testmap/textures somewhere but my cable modem connection is apparently down so I can't access my home box from work.)
The problem has already been worked around, although in a reverse manner.
The autofocus and metering systems in modern cameras do NOT take kindly to polarized light.
So how do you use a polarizing filter? A circular polarizer. Basically, it consists of a linear polarizing layer followed by a quarter-wave retardation plate of some material that will take the polarized light passing through the filter and convert it to circular polarization. (Not sure of the exact mechanics of how it works, but in short they somehow convert a linear polarization to circular.)
So one would simply need to do this in reverse with an LCD projector - In fact, one might even be able to use circ polarizers designed for cameras.
It will probably have higher loss (reduced brightness) than a system with a DLP projector, but it should probably still work well.
I'll try it tonight with my 72mm circ and my projector.
As some have pointed out, sometimes live human beings aren't much better, and in fact are worse.
I know it's a little different than a tech support line, but the company where I work was recently bought out by a larger company.
The company's policy was that no one has voicemail for external calls - It all goes to an answering service.
The problem is that while there were issues with voicemail that caused it to have a bad rep when it first came out, answering services in general are regarded as being low-quality.
Especially here - We're a tech company. We can't have suppliers/customers calling and getting some stupid teenager or old lady who is going to munge the details of whatever the message is. Plus, most of our existing customer base is used to being able to leave a voicemail message that goes *directly* to the intended recipient, not through some middle-person. (Let's not even get into issues regarding proprietary information here...)
Fortunately, the two locations that just merged in have been given an exception - We're now allowed to use voicemail again, but our messages must say, "Please press 0 to talk to a person". Overall, I'm impressed with the sensibilities of the new company. (But all the new red tape sucks...)
"Look at all of the SCSI vendors out there and tell me which gives better support for Linux?"
Um, BusLogic? Who supported Linux LONG before Adaptec did? In fact, the very drivers that LNZ maintained? Whose cards had mature Linux drivers LONG before you even acknowledged Linux?
It's a good thing Leonard has done such an excellent job with the drivers - They're mature and won't need TOO much fixing. But who will update them to new kernels/etc? Hopefully someone will take up the slack.
I have a FlashPoint controller. It started acting up after upgrading to a K6-2/500 (from a 300) - It would randomly hang the system. (Timing problem in the driver)
Seems like a total of 3-4 people on the planet had a similar hw combo (A Flashpoint in a higher-speed K6/2) and problems, nonetheless, Leonard had a patch out to fix the problem within a week of the first reports of anything happening. A role model for any other software developer...
Since such projectors are likely to have a lot of compromises.
On the other hand, it's theoretically possible to make projectors more suited to a task.
Finally - Search hard enough and you can get used LCD projectors for $1000. Resolution will be a bit low tho. I'm assuming 90%+ of that $12k was for the projectors. (Maybe only 80% - They did use a Quadro4...)
Agree. The Ciscos are VERY nice, as long as you're careful to get one that doesn't decide to peep anyway in RFMon mode.
Dunno how Orinocos compare to the Ciscos sensitivity-wise. They're definately much better than Prisms (We have 3 Prism2-based cards in my house and one Orinoco, the Orinoco gets much better range.)
100 mW transmit isn't much of an advantage for wardriving. Is very nice otherwise though.:)
As I mentioned in a previous post - Many people just do it out of curiosity to see what's out there. They never DO anything with the information except for plot it, and in mant cases, laugh at the morons who leave their APs wide-open factory-default.
In many cases, these people go wardriving just out of curiosity.
I find it very interesting that in a short drive around my area I found 45 networks (I was NOT expecting that many, esp. since I wasn't using an external antenna), and over a third of them were factory default. (Not just unencrypted, but completely unchanged factory default units.)
I haven't actually DONE anything with those APs though.
Most cards have Linux drivers that allow them to be put into "RF Monitor" mode, which is completely passive. This is the default mode of operation for Kismet (http://www.kismetwireless.net/)
Supported cards include: Prism2 with the linux-wlan-ng drivers Orinoco cards with a slightly patched driver from http://airsnort.shmoo.net/ SOME Cisco cards. While they all happily go into RF Monitor mode when asked, SOME OF THEM KEEP BROADCASTING.
So all in all, if you *absolutely* don't want to be detected, Cisco is the least safe choice for wardriving. Orinoco is probably the best bet, even though you will have to downgrade your Orinoco firmware for compatibility (8.10 is severely broken for RFMon usage). Prism2s have the best compatibility, but are generally known for crappy receivers and most don't allow external antennas. Almost all Orinoco-based cards have much better receivers and support external antennas. The Cisco hardware is the best (100 mW transmit, not like that matters if you're trying to stay silent, some have dual MMCX jacks for diversity antennas), but you can't trust it to stay silent in RF Monitor mode.
Orinoco silver, no ext antenna, laptop inside the car (lots of nice metal shielding)
Probably 1/3 of the networks heard (45 found in a relatively short loop) were factory default Linksys boxes.
There are a total of *3* 802.11 networks near my house.
One on Ch11 with a custom SSID (mine - No WEP, I don't really care. I'm in the boonies and not much damage someone could do) Two on Ch6, one factory default Linksys, one listed as by Kismet. Needless to say, those two weren't going to be getting max performance.:)
Interesting little thing about Kismet - Apparently Netstumbler is not entirely passive (Otherwise it wouldn't be detectable). Unless your driver is bugged or you have an unsupported card, Kismet is purely passive. Even better, while NS only works with Orinoco (and maybe Aironet) cards, Kismet works with Prism2 cards.
That said - With the exception of the last of the 3 utilities, most of them seem to be pretty similar to Netstumbler.
Apparently Kismet currently (for whatever reason) seems to ignore Netstumbler packets for some reason, but this is considered to be a bug. Implementing Netstumbler detection is apparently not far off.
Reputation, quality of program, broadness (Well-rounded engineer with extra knowledge in wireless vs. wireless-only engineer who is screwed if the industry goes tits-up, or gets dragged down by the fact that most of the equipment manufacturers were also involved heavily in optical networking), and facilities.
If you ask someone "in the industry", i.e. someone who is hiring, which they would rather hire - Either one of these Auburn "wireless" engineers or a GaTech EE that concentrated in wireless, they'll probably say GaTech because of its reputation and the fact that a lot of the major players do large amounts of business with GaTech.
Some devices (microwave RF transistors) tend to be more sensitive than others. (Atmel AVR microcontrollers, generally regarded as nearly invincible.)
It's probably not critical for a small-scale development lab.
As soon as you're manufacturing, that's a different story. Note the one guy that worked for VisionTek and said their return rates plummeted after they implemented strict ESD procedures.
In addition, if you're a large supplier to another company, they may audit your quality control processes. (There are a number of ISO quality control certifications that are pretty much required to be on the supplier's list for many companies.) If you don't have ESD control in your manufacturing facility, you're NOT going to pass the audit.
Many universities have EE programs that require a concentration. What's a common concentration in such programs? You guessed it, wireless. Even Cornell, which admittedly is not a "strong" school in wireless despite a top-notch EE program since the main physical-layer wireless guy was hired away by Illinois, has a pretty good wireless concentration. (Due to the fact that most of the domain of "wireless" can be covered quite well by the DSP, Information Theory, and the radar people in Space & Plasma Physics, all of which are fields where Cornell is top-notch) All in all, you'll get a much broader exposure to signal theory and RF in general than you would in a "Wireless" degree.
Whatever this program is, I'm sure it pales in comparison to the EE programs at Georgia Tech and the University of Illinios (They have two of the top wireless programs in the country - It's all under the EE umbrella.) I believe GaTech has an antenna testing range and numerous other facilities that rival that of most corporations in the field.
If you want to do wireless, go to Georgia Tech or the University of Illinios. I hear Ohio State is pretty good too, as are UCSD and probably Caltech. If you want to go to a wannabe program that won't get you a broad exposure that'll leave you with backup if wireless dries up, go to Auburn.
It worked fine with my old RH7.0/Ximian box, but since reformatting and moving to a fresh RH7.3 install with Ximian GNOME installed, it's crapola anymore.
Doesn't make sense, if anything the fresh install should work better.
The PSM is installed, it's in the same directory as all of the other Mozilla library files, it still just Doesn't Work.
The whole issue of a seperate PSM is just plain stupid - It (obviously) introduces an additional failure point that's not needed as compared to integrated SSL support like any other browser, now that export restrictions on SSL have been pretty much removed.
In my school district, there were two guys responsible for my high school's network/computers/etc.
First there was Mr. M (Name hidden), the head of the business department. Nice guy, reasonably knowledgeable (Knew Netware, which was a key skill around then, but not too much about Unices and the Internet), but KNEW HIS LIMITS, and most importantly, was a good judge of character. In return for a relaxation of school rules (During an independent study AP CS class, my friend Ross was playing GTA. Mr. M walked in, commented, "Amazing graphics programming you've done there." We also ran a Quake server on our webserver in the evenings, and the only students with school email.), about 5-6 students in the school assisted him with setting up the school's cool new webserver, IP Masquerading, and general network/machine maintenance.
Then there was Mr. S. also known as Elvis because you saw Elvis more often than you saw him - The "Technology Coordinator" for the school district.
Moron.
Treated students like The Enemy.
Thought he knew everything.
The September after I graduated high school, the network was FUBARed for over a month, because Elvis decided he was going to install Win98 across the board on every single machine without testing it on one first. ooops. Win98 and the machines didn't get along. Poor Mr. M was left to fix it with most of his assistants off to college.
Slashdot Mirror
You just don't see it.
What was "state of the art" in the '80s is now ubiquitous and hidden from the end user.
Ever found a bug in your portable Nomad MP3 player (The flash-based ones, not any of the disk-based ones - Although the disk ones are still pretty strong)
Has your car ever shut down because its computer crashed? (Note: Hardware failure doesn't count, although automotive ECU failure is RARE unless you've done something to screw with its cooling.)
What about your VCR?
These are all cases of coding like you described - Fitting as much as possible into as little space as possible. In Lucent's (now Avaya) business communications division, there was (maybe still is) a raging debate on whether the usability benefits of using two LEDs rather than one justified the *pennies* of extra cost on an item that sold for a few hundred dollars. In a cost-cutting environment that intensive, you're not going to spec a processor with 8k of flash and 2k RAM when a processor with 2k/256 bytes will do. (Note - Popular microcontroller such as the Atmel AVR, Microchip PIC, Motorola 68HC11, etc. all are in this range.)
And it's quite easy for a single programmer to do all of this. I've seen CD-based MP3 players developed in a few weeks by a team of two college students for their Microcontrollers course taking 3-4 other classes at the same time.
Rather than replying.
I think I know why fewer people showed than I expected.
The location of the funeral ingame, Caer Benowyc, is a keep on the Albion frontier.
There are relatively nasty critters wandering around the area - Dangerous for people as high as 37 when solo, and probably dangerous for anyone under 25-30 even when escorted. (My guild was congratulating me for making it there solo w/o dying - Not easy.)
It's also the farthest keep out for Albion players, through some really dangerous terrain.
The only safe place to have the funeral would've been in Midgard, but that would've prevented anyone from Albion or Hibernia from participating.
I've driven in DC. Except for downtown (Which, like Arlington, doesn't seem to be contained entirely in Euclidean space), it's not that bad.
Boston, on the other hand... YIKES.
I said the engine *SUPPORTS* high-res textures, not that he actually *USED* such textures in his download, and that the textures included with the download are, in fact, ugly. (Low-res and the bumpmaps/glossmaps are overdone)
I tried using some high-res textures/bumpmaps I found and the difference was amazing - MUCH better.
Maximum level in DAoC is 50. And I don't think they would've held a funeral near a Lv99 mob like the Dragon.
That said - I wish I'd known about this. I play DAoC and knew nothing. Of course, it WAS on a different server.
I'm honestly surprised at how few people showed up in that picture - That's on par with a single realm's RvR zerg, and word tends to spread REALLY fast throughout a realm in DAoC. If I'd been on the server in question (Pellinor, I play on Lancelot), I would've been there.
Strange thing was, I was at Beno last night... For the first time ever.
One other odd thing to note: It looks like the screenshot was made by someone not from the guild/realm holding the funeral, as the shot is almost all Midgard players but they have the generic "not from your realm" names. I also see a couple of generic Albion tags, so the shot must've been taken by a Hibernian.
The lighting effects are (in general) much neater looking.
In addition, he also implemented bumpmapping/glossmapping. Unfortunately the bumpmaps/glossmaps he's using in the demo are kind of... ugly. But the engine DOES support high-res TGA textures/bumpmaps/glossmaps. I tried downloading a few high-res textures/bumpmaps from a texture archive and slapping them into a test map - The results were... Amazing. The engine is NOWHERE close to its full potential.
(I'd post the testmap/textures somewhere but my cable modem connection is apparently down so I can't access my home box from work.)
The problem has already been worked around, although in a reverse manner.
The autofocus and metering systems in modern cameras do NOT take kindly to polarized light.
So how do you use a polarizing filter? A circular polarizer. Basically, it consists of a linear polarizing layer followed by a quarter-wave retardation plate of some material that will take the polarized light passing through the filter and convert it to circular polarization. (Not sure of the exact mechanics of how it works, but in short they somehow convert a linear polarization to circular.)
So one would simply need to do this in reverse with an LCD projector - In fact, one might even be able to use circ polarizers designed for cameras.
It will probably have higher loss (reduced brightness) than a system with a DLP projector, but it should probably still work well.
I'll try it tonight with my 72mm circ and my projector.
As some have pointed out, sometimes live human beings aren't much better, and in fact are worse.
I know it's a little different than a tech support line, but the company where I work was recently bought out by a larger company.
The company's policy was that no one has voicemail for external calls - It all goes to an answering service.
The problem is that while there were issues with voicemail that caused it to have a bad rep when it first came out, answering services in general are regarded as being low-quality.
Especially here - We're a tech company. We can't have suppliers/customers calling and getting some stupid teenager or old lady who is going to munge the details of whatever the message is. Plus, most of our existing customer base is used to being able to leave a voicemail message that goes *directly* to the intended recipient, not through some middle-person. (Let's not even get into issues regarding proprietary information here...)
Fortunately, the two locations that just merged in have been given an exception - We're now allowed to use voicemail again, but our messages must say, "Please press 0 to talk to a person". Overall, I'm impressed with the sensibilities of the new company. (But all the new red tape sucks...)
A human is not necessarily the best option...
"Look at all of the SCSI vendors out there and tell me which gives better support for Linux?"
Um, BusLogic? Who supported Linux LONG before Adaptec did? In fact, the very drivers that LNZ maintained? Whose cards had mature Linux drivers LONG before you even acknowledged Linux?
It's a good thing Leonard has done such an excellent job with the drivers - They're mature and won't need TOO much fixing. But who will update them to new kernels/etc? Hopefully someone will take up the slack.
I have a FlashPoint controller. It started acting up after upgrading to a K6-2/500 (from a 300) - It would randomly hang the system. (Timing problem in the driver)
Seems like a total of 3-4 people on the planet had a similar hw combo (A Flashpoint in a higher-speed K6/2) and problems, nonetheless, Leonard had a patch out to fix the problem within a week of the first reports of anything happening. A role model for any other software developer...
Since such projectors are likely to have a lot of compromises.
On the other hand, it's theoretically possible to make projectors more suited to a task.
Finally - Search hard enough and you can get used LCD projectors for $1000. Resolution will be a bit low tho. I'm assuming 90%+ of that $12k was for the projectors. (Maybe only 80% - They did use a Quadro4...)
Agree. The Ciscos are VERY nice, as long as you're careful to get one that doesn't decide to peep anyway in RFMon mode.
:)
Dunno how Orinocos compare to the Ciscos sensitivity-wise. They're definately much better than Prisms (We have 3 Prism2-based cards in my house and one Orinoco, the Orinoco gets much better range.)
100 mW transmit isn't much of an advantage for wardriving. Is very nice otherwise though.
As I mentioned in a previous post - Many people just do it out of curiosity to see what's out there. They never DO anything with the information except for plot it, and in mant cases, laugh at the morons who leave their APs wide-open factory-default.
Note that Kismet (http://www.kismetwireless.net/ is purely passive - You NEVER associate with the AP or broadcast it to any way.
So how is that intrusion?
In many cases, these people go wardriving just out of curiosity.
I find it very interesting that in a short drive around my area I found 45 networks (I was NOT expecting that many, esp. since I wasn't using an external antenna), and over a third of them were factory default. (Not just unencrypted, but completely unchanged factory default units.)
I haven't actually DONE anything with those APs though.
Most cards have Linux drivers that allow them to be put into "RF Monitor" mode, which is completely passive. This is the default mode of operation for Kismet (http://www.kismetwireless.net/)
Supported cards include:
Prism2 with the linux-wlan-ng drivers
Orinoco cards with a slightly patched driver from http://airsnort.shmoo.net/
SOME Cisco cards. While they all happily go into RF Monitor mode when asked, SOME OF THEM KEEP BROADCASTING.
So all in all, if you *absolutely* don't want to be detected, Cisco is the least safe choice for wardriving. Orinoco is probably the best bet, even though you will have to downgrade your Orinoco firmware for compatibility (8.10 is severely broken for RFMon usage). Prism2s have the best compatibility, but are generally known for crappy receivers and most don't allow external antennas. Almost all Orinoco-based cards have much better receivers and support external antennas. The Cisco hardware is the best (100 mW transmit, not like that matters if you're trying to stay silent, some have dual MMCX jacks for diversity antennas), but you can't trust it to stay silent in RF Monitor mode.
OK, cool, thanks.
Been using Kismet for a few days and it's *great*, other than the fact that the -L option to gpsmap (labeling) is busted.
I went wardriving this past weekend.
:)
Orinoco silver, no ext antenna, laptop inside the car (lots of nice metal shielding)
Probably 1/3 of the networks heard (45 found in a relatively short loop) were factory default Linksys boxes.
There are a total of *3* 802.11 networks near my house.
One on Ch11 with a custom SSID (mine - No WEP, I don't really care. I'm in the boonies and not much damage someone could do)
Two on Ch6, one factory default Linksys, one listed as by Kismet. Needless to say, those two weren't going to be getting max performance.
I've been using it for a little myself.
Interesting little thing about Kismet - Apparently Netstumbler is not entirely passive (Otherwise it wouldn't be detectable). Unless your driver is bugged or you have an unsupported card, Kismet is purely passive. Even better, while NS only works with Orinoco (and maybe Aironet) cards, Kismet works with Prism2 cards.
That said - With the exception of the last of the 3 utilities, most of them seem to be pretty similar to Netstumbler.
Apparently Kismet currently (for whatever reason) seems to ignore Netstumbler packets for some reason, but this is considered to be a bug. Implementing Netstumbler detection is apparently not far off.
Reputation, quality of program, broadness (Well-rounded engineer with extra knowledge in wireless vs. wireless-only engineer who is screwed if the industry goes tits-up, or gets dragged down by the fact that most of the equipment manufacturers were also involved heavily in optical networking), and facilities.
If you ask someone "in the industry", i.e. someone who is hiring, which they would rather hire - Either one of these Auburn "wireless" engineers or a GaTech EE that concentrated in wireless, they'll probably say GaTech because of its reputation and the fact that a lot of the major players do large amounts of business with GaTech.
If you want to concentrate very specifically on a field in depth (i.e. wireless), that's what graduate school is for.
I regret concentrating too much on RF as an undergrad, despite having taken a few courses outside of RF in DSP and information theory.
Some devices (microwave RF transistors) tend to be more sensitive than others. (Atmel AVR microcontrollers, generally regarded as nearly invincible.)
It's probably not critical for a small-scale development lab.
As soon as you're manufacturing, that's a different story. Note the one guy that worked for VisionTek and said their return rates plummeted after they implemented strict ESD procedures.
In addition, if you're a large supplier to another company, they may audit your quality control processes. (There are a number of ISO quality control certifications that are pretty much required to be on the supplier's list for many companies.) If you don't have ESD control in your manufacturing facility, you're NOT going to pass the audit.
This isn't a first, in any way whatsoever.
Many universities have EE programs that require a concentration. What's a common concentration in such programs? You guessed it, wireless. Even Cornell, which admittedly is not a "strong" school in wireless despite a top-notch EE program since the main physical-layer wireless guy was hired away by Illinois, has a pretty good wireless concentration. (Due to the fact that most of the domain of "wireless" can be covered quite well by the DSP, Information Theory, and the radar people in Space & Plasma Physics, all of which are fields where Cornell is top-notch) All in all, you'll get a much broader exposure to signal theory and RF in general than you would in a "Wireless" degree.
Whatever this program is, I'm sure it pales in comparison to the EE programs at Georgia Tech and the University of Illinios (They have two of the top wireless programs in the country - It's all under the EE umbrella.) I believe GaTech has an antenna testing range and numerous other facilities that rival that of most corporations in the field.
If you want to do wireless, go to Georgia Tech or the University of Illinios. I hear Ohio State is pretty good too, as are UCSD and probably Caltech. If you want to go to a wannabe program that won't get you a broad exposure that'll leave you with backup if wireless dries up, go to Auburn.
It worked fine with my old RH7.0/Ximian box, but since reformatting and moving to a fresh RH7.3 install with Ximian GNOME installed, it's crapola anymore.
Doesn't make sense, if anything the fresh install should work better.
The PSM is installed, it's in the same directory as all of the other Mozilla library files, it still just Doesn't Work.
The whole issue of a seperate PSM is just plain stupid - It (obviously) introduces an additional failure point that's not needed as compared to integrated SSL support like any other browser, now that export restrictions on SSL have been pretty much removed.
And treating ALL students like The Enemy.
In my school district, there were two guys responsible for my high school's network/computers/etc.
First there was Mr. M (Name hidden), the head of the business department. Nice guy, reasonably knowledgeable (Knew Netware, which was a key skill around then, but not too much about Unices and the Internet), but KNEW HIS LIMITS, and most importantly, was a good judge of character. In return for a relaxation of school rules (During an independent study AP CS class, my friend Ross was playing GTA. Mr. M walked in, commented, "Amazing graphics programming you've done there." We also ran a Quake server on our webserver in the evenings, and the only students with school email.), about 5-6 students in the school assisted him with setting up the school's cool new webserver, IP Masquerading, and general network/machine maintenance.
Then there was Mr. S. also known as Elvis because you saw Elvis more often than you saw him - The "Technology Coordinator" for the school district.
Moron.
Treated students like The Enemy.
Thought he knew everything.
The September after I graduated high school, the network was FUBARed for over a month, because Elvis decided he was going to install Win98 across the board on every single machine without testing it on one first. ooops. Win98 and the machines didn't get along. Poor Mr. M was left to fix it with most of his assistants off to college.