In this vein I propose an alternate theory, that in fact the world is spherical. No need to thank me. Just tell everyone that you heard it first on slashdot.
This is just a random press release from a company selling boring old wireless kit. There isn't anything special about getting 5km of range with off the shelf wireless gear,
If anyone cares the math is not that hard to follow:
You take the transmitter output power, subtract any loss in the cable between the transmitter and the antenna, add the gain of the anteana.
http://db.osoal.org.nz/eirp-calculator.html
And finally you add the recieve gain of the receiving antenna, subtract the loss from the cable on the recieving end and compare with the receive sensitivity of your wireless receiver. I have bundled the rest of the calculation into this bit of javascript.
For example, if I have two 2.4ghz radios that output 15dbm (32mw) that have a recieve sensitivity of -83db that are in waterproof boxes on the antenna mounting connected to two 22db antenna's 5km apart very roughly.
15db - 1db + 22db = 36db or ~3981mw (just under the 4 watt max).
- 121.65db for our free space loss
+ 22db - 1db = -64.65
-64.65 is the strength of the signal received at the other end, fortunately the receiver has a receive sensitivity of -83 so we are in business. There is a link margin of ~19db to account for a little bit of noise, fade, solar flares, alien abductions etc.
If you want more range, increase the power of the transmitter or the gain of your antenna. The government limits ( 4 watts for 2.4ghz, 250mw for 5.3ghz and 4 watts for 5.8Ghz in New Zealand ) are going to determine your maximum range barring some magical new wireless gear that has a better receive sensitivity.
Well I still read slashdot, mostly these days just to post in the Low UID threads =) I guess Taco must read slashdot occasionally too, he's got a pretty low UID. I was reading slashdot a while before user accounts existed and I'm guessing there are still a heap of people from way back then.
The parent was talking about Firefox, you are talking about Mozilla. Mozilla is definitely a whole lot slower and the last time I used it a whole lot more broken than Firefox.
I haven't given much thought to which cut and paste buffers in action because I've never had any trouble cut and pasting between anything I've wanted to before.
I don't use anything stunning in terms of applications, 98% of the time just konsoles and firefox. I still read mail in pine in an ssh session on a different box. I used openoffice once to write a letter to get off a parking ticket.
Most of the real work I do involves tcpdump, mtr, traceroute, ping and other assorted bits of perl, all of which are actually being run on other machines.
I have no idea what release of Linux you are using, but I figured I'd try out a couple of things you mentioned in your post. Our desktop machines here at work run SuSE, they haven't been upgraded in a while so mine is still running SuSE 9.2.
I went and hunted about in the server room here at work for a USB mouse, and found an old microsoft one in a box of junk, I plug it into the front USB ports on my PC. A dialog box pops up which I have taken a screenshot of here:
The mouse is actually working at this point, and I can use it to click on the "Yes" button.
Okay how about changing resolutions, I click on the "Screen resize" applet in the tray and choose a resolution, it changes and a second later I'm looking at 800x600 rather than 1600x1200.
I agree adding USB devices and changing resolutions used to suck, it doesn't anymore.
I used to be quite mystified by people complaining about copy and paste on Linux until I went and used a windows machine for a few days. People in windows have to highlight, right click and select copy ( or the corresponding keyboard shortcut ). I have been using Linux for about 10 years now and for the whole time I have been highlighting text to copy and clicking the middle mouse button to paste. It works in just about everything on Linux. I can see that the windows method is totally busted across most Linux applications.
Configuring printers I agree is rubbish in Linux I have no problem with editing the cups config to add stuff, but the GUI frontends SuSE provide had me absolutely bamboozled the last time I tried.
Oh I agree that cycles per second is not a useful measurement of performance. The point of my post was that even if your application performs two or three times better per clock on a given architecture you are going to get beaten if you competition does four mhz for every mhz you do at a fraction of the price.
Reading this article evokes a bit of a shock for me, so I go to the sgi site to check out what they are doing these days. I look at their workstations because thats what people used to buy, SGI workstations right?
700mhz or 800mhz MIPS processors?! 800mhz dual or quad processor boxes?! I see they have a whopping 4mb of cache these days but 800mhz MIPS processors were available in 2002. Even if $scientific_calculation function you want runs two or three times faster on MIPS than on a P4 you can get a 3ghz P4 for $285NZD these days. This is why SGI stock is tanking.
The only saving grace is that SGI still has a greater market cap than SCO ( 148.48M vs 70.44M ).
I accept that windows has been fairly decent as a desktop box since 2000.
I have been running Linux on my desktop machine for almost 10 years now however and it has been decent for alot longer ( not as pretty for as long though ).
Almost certainly weirdo problems you were encountering were peculiar to your hardware / bios versions on the release of Linux you were running. Linux as a whole ( some GUI elements excluded ) is disturbingly stable and has been for a very long time.
So what you say? This is the third time its been around the uptime counter thats approx 1010 days ( 2.7 years ) of uptime. The boot before this one, it was up for over two years before the rather elderly SCSI disk in the box failed. Just about every machine I can find around here excluding those that have been replaced with new ones recently have more than 180 days of uptime.
With any operating system you are going to have oddities where a specific driver or bit of hardware is flakey, it doesn't need to be an excuse to conclude that the platform is immature or not ready in general.
> Remeber your ancient TNT graphics card that had 16MB of memory?
I don't think anything with 16mb qualifies as ancient. I still have cards like that in use in firewalls etc. Ugh, people put video cards in PCI slots before AGP, and ISA slots before that ( I'm sure other weird and wonderful buses before that ). Anyone else remember how many characters per second their old beasts could do?
> In all cases, including "wardriving", there is no > legitimate reason to collect the information or > listen in. It's none of your goddamn business.
I operate a bunch of 802.11b 2.4ghz access points in my area ( somewhere in the order of 6 ), a couple of connections commercially but mostly for employees / acquaintances of our company with an assortment of antennas and gear.
We also operate a fair bit of Trango gear in the 5.3Ghz and 5.8Ghz spectrums, the fundamentals are the same.
Scoping out who is running networks and where they are pointing and roughly what EIRP they run is absolutely essential to ensure that I don't stomp on other peoples networks and that I run my own networks in channels that receive the least interferance.
I am completely uninterested in the data that crosses other peoples networks, and I am not defending people who are into snooping the _traffic_. But from a RF point of view, this is absolutely my business, as this is the only way that you can be a 'good citizen' in unlicenced spectrum.
The solution that I introduced to the company I work for a few years ago was to run an old box with Windows 2000 Server on it as a terminal server. We had a guy running VMWare and putting up with trying to get it to work again with every kernel update and this eliminated all the hassle.
I'm sure there are commercial terminal services clients for Linux, but we run rdesktop.
Since we started using it rdesktop has included support for RDP5 which supports 16 bit colour, so with a Windows 2003 server ( we have upgraded ) you get a reasonably nice looking windows desktop. Audio seems to go mostly too, not that its needed for a couple of minutes worth of checking some html renders in IE or talking someone through how to setup outlook express.
If a windows only accounts package or similar is keeping you from running Linux on your desktop this could be a good solution, the only negative is possibly the Windows server licensing is a bit steep for some situations.
Actually we are pretty damn well connected for a couple of little islands in the middle of nothing.
In Wellington city you can get a 10/100/1000 mbit connection off citylink as featured in this slashdot article, and from there you can buy bandwidth off a number of providers which often are reselling southern cross bandwidth ( around 240 gigabit of mostly unused capacity ). Victoria University ( who are hosting the paper linked ) are just up the hill from the citylink offices.
Judging by the plea for assistance from a worried sounding VUW staff member posted to the nznog mailing list this afternoon, their web server is a fraction too slow to handle a slashdotting.
I have been in a situation with an aironet network where I have flushed the SSID and wep key of the card, and noticed while flicking bettween consoles that there was traffic from another network floating past. This is with a little ( quite directional ) parabolic grid antennae facing about half way bettween two of our own sites.
As these cards get cheaper and more people use them, the fixed set of frequency's that the frequency hopping cards use are going to become more and more useless with high gain aerials.
Even without the security implications, each site within 'earshot' are going to end up sharing the realistic 500k/s or so that the 11 megabit cards provide.
An issue I have with binary distributions that I have encountered in the wild is that in most cases there is little indication of which headers and libraries the binaries were built against.
An rpm may have been built against the original Redhat 6.2 release, RedHat 6.2 with a couple of the ( rather extensive ) updates applied, redhat 6.2 with glibc 2.2.x installed from source, redhat 7.0, mandrake etc..
At least a source rpm can be rebuilt against the libraries and headers on *your* machine.
It might be worth noting that this source release is not the complete source for Q3.
This release contains the source for all of the 'interesting' parts of the code, the game logic, and anything that Carmack/id decided mod makers would find interesting.
Alot of people have been confused by previous source releases in this way.
I have a linux 2.0.36 machine which has a current uptime of 112 days, 112 days ago ( while the machine was on around 100-ish days ) it needed rebooting so its power source could be relocated.
More interestingly I have a whole horde of 2.2.x machines most of which are running the stock kernel that was installed with which ever distribution seemed good at the time, these seem to crash a fair bit, I occasionally find them completely locked up, not responding to the keyboard, pings etc.. most of them have good motherboards, a mix of scsi, ide disks, some SMP, some uniprocessor.
I have upgraded the kernel on a couple of the machines ( guinea pigs if you like ) to remove alot of the useless things and try to get some rock solid reliability back.
I absolutely love showing off good uptimes to clients on machines that do alot of work but I just havent managed to make it happen with the current stable kernel.
I have an intel 486DX-33 machine running at 50mhz with a current uptime of 114 days ( the last time it rebooted was due to a power cut ).
This speed was achieved by putting a plain old 486 heatsink on it, no fan, and in fact it had run previously at 50mhz with out the heatsink. 50mhz is as fast as the board will do. The vesa local bus video completely failed to work at all with the bus running that fast, and it has an isa card in it currently.
I have posted some thoughts and screenshots of it running under Linux at http://www.osoal.org.nz/ in the comments section. Basically its very cool, seems to run ok in 1024x768 on my celery 333 ( with 2 8mb diamond voodoo II's ).
Slashdot Mirror
In this vein I propose an alternate theory, that in fact the world is spherical. No need to thank me. Just tell everyone that you heard it first on slashdot.
This is just a random press release from a company selling boring old wireless kit. There isn't anything special about getting 5km of range with off the shelf wireless gear,
If anyone cares the math is not that hard to follow:
You take the transmitter output power, subtract any loss in the cable between the transmitter and the antenna, add the gain of the anteana. http://db.osoal.org.nz/eirp-calculator.html
Then you have to subtract the free space loss of your path ( the loss you get by putting the signal across the air ): http://db.osoal.org.nz/freespace-loss-calculator.h tml
And finally you add the recieve gain of the receiving antenna, subtract the loss from the cable on the recieving end and compare with the receive sensitivity of your wireless receiver. I have bundled the rest of the calculation into this bit of javascript.
For example, if I have two 2.4ghz radios that output 15dbm (32mw) that have a recieve sensitivity of -83db that are in waterproof boxes on the antenna mounting connected to two 22db antenna's 5km apart very roughly.
15db - 1db + 22db = 36db or ~3981mw (just under the 4 watt max).
- 121.65db for our free space loss
+ 22db - 1db = -64.65
-64.65 is the strength of the signal received at the other end, fortunately the receiver has a receive sensitivity of -83 so we are in business. There is a link margin of ~19db to account for a little bit of noise, fade, solar flares, alien abductions etc.
If you want more range, increase the power of the transmitter or the gain of your antenna. The government limits ( 4 watts for 2.4ghz, 250mw for 5.3ghz and 4 watts for 5.8Ghz in New Zealand ) are going to determine your maximum range barring some magical new wireless gear that has a better receive sensitivity.
Well I still read slashdot, mostly these days just to post in the Low UID threads =)
I guess Taco must read slashdot occasionally too, he's got a pretty low UID. I was reading slashdot a while before user accounts existed and I'm guessing there are still a heap of people from way back then.
The parent was talking about Firefox, you are talking about Mozilla. Mozilla is definitely a whole lot slower and the last time I used it a whole lot more broken than Firefox.
I haven't given much thought to which cut and paste buffers in action because I've never had any trouble cut and pasting between anything I've wanted to before.
I don't use anything stunning in terms of applications, 98% of the time just konsoles and firefox. I still read mail in pine in an ssh session on a different box. I used openoffice once to write a letter to get off a parking ticket.
Most of the real work I do involves tcpdump, mtr, traceroute, ping and other assorted bits of perl, all of which are actually being run on other machines.
I guess I'm just not that fussy.
I have no idea what release of Linux you are using, but I figured I'd try out a couple of things you mentioned in your post. Our desktop machines here at work run SuSE, they haven't been upgraded in a while so mine is still running SuSE 9.2.
i on.jpg i on-after.jpg
I went and hunted about in the server room here at work for a USB mouse, and found an old microsoft one in a box of junk, I plug it into the front USB ports on my PC. A dialog box pops up which I have taken a screenshot of here:
http://db.osoal.org.nz/screenshots/new-mouse.jpg
The mouse is actually working at this point, and I can use it to click on the "Yes" button.
Okay how about changing resolutions, I click on the "Screen resize" applet in the tray and choose a resolution, it changes and a second later I'm looking at 800x600 rather than 1600x1200.
http://db.osoal.org.nz/screenshots/change-resolut
http://db.osoal.org.nz/screenshots/change-resolut
I agree adding USB devices and changing resolutions used to suck, it doesn't anymore.
I used to be quite mystified by people complaining about copy and paste on Linux until I went and used a windows machine for a few days. People in windows have to highlight, right click and select copy ( or the corresponding keyboard shortcut ). I have been using Linux for about 10 years now and for the whole time I have been highlighting text to copy and clicking the middle mouse button to paste. It works in just about everything on Linux. I can see that the windows method is totally busted across most Linux applications.
Configuring printers I agree is rubbish in Linux I have no problem with editing the cups config to add stuff, but the GUI frontends SuSE provide had me absolutely bamboozled the last time I tried.
Anyway just trying to add some more data points.
Oh I agree that cycles per second is not a useful measurement of performance. The point of my post was that even if your application performs two or three times better per clock on a given architecture you are going to get beaten if you competition does four mhz for every mhz you do at a fraction of the price.
And anyway I have a lower slashdot ID, so I win.
Reading this article evokes a bit of a shock for me, so I go to the sgi site to check out what they are doing these days. I look at their workstations because thats what people used to buy, SGI workstations right?
700mhz or 800mhz MIPS processors?! 800mhz dual or quad processor boxes?! I see they have a whopping 4mb of cache these days but 800mhz MIPS processors were available in 2002. Even if $scientific_calculation function you want runs two or three times faster on MIPS than on a P4 you can get a 3ghz P4 for $285NZD these days. This is why SGI stock is tanking.
The only saving grace is that SGI still has a greater market cap than SCO ( 148.48M vs 70.44M ).
I accept that windows has been fairly decent as a desktop box since 2000.
I have been running Linux on my desktop machine for almost 10 years now however and it has been decent for alot longer ( not as pretty for as long though ).
Almost certainly weirdo problems you were encountering were peculiar to your hardware / bios versions on the release of Linux you were running. Linux as a whole ( some GUI elements excluded ) is disturbingly stable and has been for a very long time.
I submit exibit A:
2:05pm up 16 days, 12:12, 4 users, load average: 0.18, 0.12, 0.11
So what you say? This is the third time its been around the uptime counter thats approx 1010 days ( 2.7 years ) of uptime. The boot before this one, it was up for over two years before the rather elderly SCSI disk in the box failed. Just about every machine I can find around here excluding those that have been replaced with new ones recently have more than 180 days of uptime.
With any operating system you are going to have oddities where a specific driver or bit of hardware is flakey, it doesn't need to be an excuse to conclude that the platform is immature or not ready in general.
> Remeber your ancient TNT graphics card that had 16MB of memory?
I don't think anything with 16mb qualifies as ancient. I still have cards like that in use in firewalls etc. Ugh, people put video cards in PCI slots before AGP, and ISA slots before that ( I'm sure other weird and wonderful buses before that ). Anyone else remember how many characters per second their old beasts could do?
> In all cases, including "wardriving", there is no > legitimate reason to collect the information or
> listen in. It's none of your goddamn business.
I operate a bunch of 802.11b 2.4ghz access points in my area ( somewhere in the order of 6 ), a couple of connections commercially but mostly for employees / acquaintances of our company with an assortment of antennas and gear.
We also operate a fair bit of Trango gear in the 5.3Ghz and 5.8Ghz spectrums, the fundamentals are the same.
Scoping out who is running networks and where they are pointing and roughly what EIRP they run is absolutely essential to ensure that I don't stomp on other peoples networks and that I run my own networks in channels that receive the least interferance.
I am completely uninterested in the data that crosses other peoples networks, and I am not defending people who are into snooping the _traffic_. But from a RF point of view, this is absolutely my business, as this is the only way that you can be a 'good citizen' in unlicenced spectrum.
The solution that I introduced to the company I work for a few years ago was to run an old box with Windows 2000 Server on it as a terminal server. We had a guy running VMWare and putting up with trying to get it to work again with every kernel update and this eliminated all the hassle.
I'm sure there are commercial terminal services clients for Linux, but we run rdesktop. Since we started using it rdesktop has included support for RDP5 which supports 16 bit colour, so with a Windows 2003 server ( we have upgraded ) you get a reasonably nice looking windows desktop. Audio seems to go mostly too, not that its needed for a couple of minutes worth of checking some html renders in IE or talking someone through how to setup outlook express.
If a windows only accounts package or similar is keeping you from running Linux on your desktop this could be a good solution, the only negative is possibly the Windows server licensing is a bit steep for some situations.
Ha! And I refute you. ( are we getting silly yet? )
Actually we are pretty damn well connected for a couple of little islands in the middle of nothing.
In Wellington city you can get a 10/100/1000 mbit connection off citylink as featured in this slashdot article, and from there you can buy bandwidth off a number of providers which often are reselling southern cross bandwidth ( around 240 gigabit of mostly unused capacity ). Victoria University ( who are hosting the paper linked ) are just up the hill from the citylink offices.
Judging by the plea for assistance from a worried sounding VUW staff member posted to the nznog mailing list this afternoon, their web server is a fraction too slow to handle a slashdotting.
Well seeing as everyone else is jumping off the cliff....
Me too!
I'd like to add that I'm on top.
Looking at the usernames of people with 50 UID's yields alot of people of repute ( and some complete nobodies like myself. ).
I have a Linux 2.2.x machine which has just crossed into its second year of uptime.
Windows 98 is not 'stable'.
I have been in a situation with an aironet network where I have flushed the SSID and wep key of the card, and noticed while flicking bettween consoles that there was traffic from another network floating past. This is with a little ( quite directional ) parabolic grid antennae facing about half way bettween two of our own sites.
As these cards get cheaper and more people use them, the fixed set of frequency's that the frequency hopping cards use are going to become more and more useless with high gain aerials.
Even without the security implications, each site within 'earshot' are going to end up sharing the realistic 500k/s or so that the 11 megabit cards provide.
You are going to have to underclock your slashdot id some more dude.
An issue I have with binary distributions that I have encountered in the wild is that in most cases there is little indication of which headers and libraries the binaries were built against.
An rpm may have been built against the original Redhat 6.2 release, RedHat 6.2 with a couple of the ( rather extensive ) updates applied, redhat 6.2 with glibc 2.2.x installed from source, redhat 7.0, mandrake etc..
At least a source rpm can be rebuilt against the libraries and headers on *your* machine.
It might be worth noting that this source release is not the complete source for Q3.
This release contains the source for all of the 'interesting' parts of the code, the game logic, and anything that Carmack/id decided mod makers would find interesting.
Alot of people have been confused by previous source releases in this way.
I have a linux 2.0.36 machine which has a current uptime of 112 days, 112 days ago ( while the machine was on around 100-ish days ) it needed rebooting so its power source could be relocated.
More interestingly I have a whole horde of 2.2.x machines most of which are running the stock kernel that was installed with which ever distribution seemed good at the time, these seem to crash a fair bit, I occasionally find them completely locked up, not responding to the keyboard, pings etc.. most of them have good motherboards, a mix of scsi, ide disks, some SMP, some uniprocessor.
I have upgraded the kernel on a couple of the machines ( guinea pigs if you like ) to remove alot of the useless things and try to get some rock solid reliability back.
I absolutely love showing off good uptimes to clients on machines that do alot of work but I just havent managed to make it happen with the current stable kernel.
I have an intel 486DX-33 machine running at 50mhz with a current uptime of 114 days ( the last time it rebooted was due to a power cut ).
This speed was achieved by putting a plain old 486 heatsink on it, no fan, and in fact it had run previously at 50mhz with out the heatsink. 50mhz is as fast as the board will do. The vesa local bus video completely failed to work at all with the bus running that fast, and it has an isa card in it currently.
It appears that the 'stuff' url on the short description should be ftp://ftp.troll.no/qt/source/qt-2.00.ta r.gz
I have posted some thoughts and screenshots of it running under Linux at http://www.osoal.org.nz/ in the comments section. Basically its very cool, seems to run ok in 1024x768 on my celery 333 ( with 2 8mb diamond voodoo II's ).