I think it's a kludge that kilo means 1024 in kilobyte, while it means 1000 in kilometer, kilogram, kilonewton, kilovolt etc. It's time we give it up, and accept correct terms. I agree kibi sounds ridiculous, but that's just a matter of habit. We'll get used to it.
I think it's a kludge that hacker means criminal in the media, while it has very good connotations in our circles. It's time we give it up and accept correct terms. I agree cracker sounds ridiculous, but that's just a matter of habit. We'll get used to it.
It's brilliant from a stratigic viewpoint. Kind of like giving a little bit of money to a bunch of ignorant Arab terrorists to keep the Russians from taking over a certain country.
Further, the Patent *specifically* states 'a "unistroke" is a single, unbroken stroke', whereas graffiti has a number of multiple-stroke characters.
Which Grafitti characters are multiple-stroke, exactly? Every single one I can think of (including the funky characters) are all single stroke -- you lift the pen off the screen and the PDA attempts to recognize the stroke you just made. (I do not count charset shifts as multiple strokes...)
The LinuxDA PDA, OTOH, does have multiple-stroke characters, and many of them. Pisses me off, actually, because I'm so used to the Palm Grafitti.
The OS allows access to raw sockets and, therefore, the entire kernel.
Go read it again. Raw sockets is not a security flaw. Unix (including Linux and OSX) has them too. All it means is that it's easy to spoof packets. That's it.
a) KOffice gets a Word and Excel Input/Output doc filter which works reliably.
While I don't have much trouble with Word/Excel doc import (big fancy ones yes, but straightforward ones no), I don't know why straight RTF isn't supported in KWord. Crazy.
b) Konqueror speeds up - I loved it and then its startup time seemed to slow down drastically.
I hear that Konq has totally rewritten their JScript interpreter. I hope the hell they fixed the popup problem... popups normally get a prompt action for me (i.e. "this site is trying ot use a popup. Allow?") but for Flash sites the popup never ever gets prompted, which drives me insane. Especially when 8 or 10 windows pop up because the JScript interpreter doesn't provide the right answer. UGH!
Speedups will be good though. I wonder if they were able to speed up any further than the 2.2.2 and prelinking. Startup time is still ugly for most KDE apps. That is one thing I noticed right away. Every time I start up xchat, it's onscreen almost immediately after I click the button. Konq, KWord, KMail... ~3-5s pause. Prelinked. On a Cel300 @1024x768x24 with 256M of RAM and no swap. Shouldn't be this slow.
I want to install Nvidia drivers on a Dual PIII.
Mandrake 8.1
X 4.1 something.
I dunno about Mandrake but it was very easy under Slackware (dual Cel466, Slack 8.0, X 4.1.0) -- found out about #nvidia from #linuxhelp. Went there, chanbot introduced itself and told me to !files to see what was available without my asking. Found the 1241 drivers, grabbed, compiled, insmod'd and modified XF86Config to use the new driver. Boom, it was up in under 30 minutes.
Sometimes it can be complicated to install something on Windows.
Sometimes it's just impossible under X on Linux.
And Johnny Lawnmover WON'T have the same patience as me and will install a Bootleg Windows within 30 minutes.
True enough; But don't forget the other side of those first two statements: Sometimes things are hard to do in Windows that are trivial under Linux (X). Sometimes things are just impossible to do under Windows.
And as far as Johnny Lawnmower is concerned you're absolutely right. But let's look at people other than the average schmuck who only wants to surf porn and listen to mp3s: Let's talk the technical crowd, the academia and people like me who are sick of being trapped in the Windows world where pretty much everything you see and touch is covered in a thin layer of Works but if you scratch through most things you find out there really are monkeys controlling the machine. I'm sick of having to pay for support, pay for updates and pay to hear "just reinstall." For people like me, Linux is a godsend. I can get done what I want and make my computer do what I want instead of moulding my wants and needs to some predetermined way of how things have to work.
You can have war stories about any OS and any hardware/driver. Try to keep that in mind.
The way it's setup here, it works with anything that can print. We do use AutoCAD LT 98 with it too. It's quite transparent for the users, and for the application.
nonono... this is for an automated system in our intranet. i.e. customer wants to see/play with drawing without downloading the entire thing. Yeah I know Acrobat creator can print to pdf; that's not quite what we need (unless it is perhaps possible to script it out from a network connection...)
The problem with the readers (voloview et al) is that they only work for Windows; there's no decent dwg viewer for Linux that I have been able to find, so I'm trying to find a good convertor.:-)
Because some partners and clients who are still using Office 95/97 (and even 2k itself) sometimes couldn't open our Office files, we now only send stuff in pdf (automatic conversion through a linux virtual shared printer and ps2pdf).
Know of anything that will work for AutoCAD 14.dwg files and AutoDesk Inventor files? (either to DXF or to.pdf) -- I'd love to be able to provide that same functionality to people who use our drawings. Yeah you can save to.dxf but that's a pain for the draftsmen who have a zillion drawings around WITHOUT having dupes in.dxf format.
One is called Deleted Item Retention Time. You set the number of days and when a user deletes an email it's not really deleted for the specified days. If he realizes he made a mistake restoration is from the Outlook client and takes seconds saving the admin time from going to the back up tapes. For businesses like law firms it's a life saver since they are required to keep records and emails for five years or so. They simply buy a lot of storage and set a deleted item retention item of 1600 days or so and it's a secondary back up.
Just out of curiosity, how do you get an email back when it's deleted when the only thing on tape is a collection of large proprietary binary files which make up the entire private mail store?
Then third party back up programs have a feature called brick level back up's where you can back up individual mailboxes. If you delete on by accident restoration is simple. Exchange 2000 has this feature out of the box.
Answered my own question, I suppose.:-)
And the global address book is great. Users don't have to keep their own huge address book and greatly minimizes the calls to the admin of I sent out this email but it came back returned and asking you to track down an email address.
Perhaps I'm using it totally wrong but at least in Exchange Server 5.5 the GAL is for local mailboxes only. This was years ago but we had to create a public folder and use Outlook to store contacts in it.
Sure you can cobble together a few products for most of the functionality and perform some of the usability features manually, but you'll spend more time while the CEO is asking you to restore an email from a year ago.
I'll be the first to admit that Exchange Server and Outlook are one killer combination. However like I have stated several times over the years, I'll be god-damned if I'm going to lock up my company's data like that. The data store is proprietary. The access tools are proprietary. Maybe I'm getting old and crochety but I've been bit before (too many times in fact) to just let it go. I want to be able to get the format of the data files and protocols so that when I want to do X I can go do it, hire someone to do it or otherwise do what the hell I want with my data, when I want and without some motherly giant cooing "Now now, dear. You don't really need to do it like that."
Buy an always-on UPS. It will take the "unclean" power and re-create it for your stereo etc.
Tellyawhat: You show me any kind of high frequency getting past the 50/60Hz transformer in the audio power supply and/or the switchmode power supply secondary inductor (DC choke) and the filter caps sprinkled all over the PCB and I'll be surprised. Show the same noise appearing in the output of said audio device and I'll buy you an always-on UPS.
The kind of noise you describe getting through is such utter bullshit. The DC power supplies in practically all electronic equipment are capable of filtering this crap out and the measures for A/V devices are double. Remember that in a linear power supply you have a huge laminated-core transformer that will absorb high frequency noise as eddy current losses. In switchers you have a high-frequency (usually 60kHz and up) oscillator and the chokes and filters on the output are designed to give you as-close-to-zero ripple as possible. And after that you have high frequency, low-ESR caps across every IC and tons of filtering on the audio inputs and outputs to keep things sane.
Always-on UPSes are useful for really shitty lines and equipment which is sensitive to the fast switchover of traditional UPSes. It's once again proven that you can get superb advice from an AC.
"Flag any sentence contianing the phrase "pheremone" as potential bullshit"
Alright, I'm no biologist, but I thought that pheremones were well proven in the insect world. I thought it was also proven that smells can induce responses in the brain on subconscious levels.
You *could* use a spectrum analyzer... or you could make creative use of directional antennas and the abilities of some off-the-shelf equipment to do site surveys.
A spectrum analyzer without the directional antenna is as useless as tits on a bull in this scenario. I would have done it the same way as the fellow you describe did. Use a directional antenna with a device that can tell you the signal strength and point it around. Hell get a couple and triangulate the position, then take a high power rifle with a decent scope and take it out.:-)
2.4 is now into maintenance mode, where it should be bug fixes and new drivers etc only. That does not mean they should be assumed to work for everyone - if you dont have the time (and the patience) to test them to ensure that they work for you then you should be using whatever Debian/Redhat/whoever chooses to test and ship.
Don't get me wrong; I totally agree with you. I was taking exception to what you had said in your previous post that they only added things to the kernel that you could opt-in to on the even-middle-number kernels.
The core kernel will remain stable, new drivers will be added where they have no effect on the rest of the kernel - so its the users choice if they want the new drivers or not.
Like the vm changes? That was just a driver change that anyone could swap in or out of the stable 2.4 kernel, eh?
Yes the anonymous cowards are a pain but sometimes they do make a good point.
The F-CPU project has been around for a good long while, and is supposed to be 64-bit. The idea of an opensource CPU is a neat idea, I must admit. Unfortunately it hasn't been updated in a while.
Marlon isn't a Linux/Unix guru, but his workshop is that of a "hands on" guy; littered with parts, antenna wires, soldering irons and tools. No spectrum analyzers, digital scopes, or automated test beds... Marlon uses the built-in site survey capabilities of some wireless radios to do 75% of what that equipment would do at 1% of the cost.
Umm... as a distributor/sales, you would never use spectrum analyzers, digital scopes or ATE; you buy your products and if they break, send 'em back under warranty. To do otherwise is folly. Now if Marlon were manufacturing this equipment and didn't have the test equipment mentioned, he'd also be stupid. He'd be wasting a lot of time guessing what was wrong.
How do I know? I'm in both positions. I design equipment but I also resell other equipment. If you've bought a warranty, you better damn well use it.
This sort of "can do" thinking is why I think that the spirit of America shines brightest in its small towns and rural areas and not in the corporate hallways. Marlon learned enough about microwave systems to be able to design workable wireless networks (and DSL systems too, although the article doesn't speak to that), sell them, install them, connect them, and keep them working.
Strong will and common sense go a hell of a long way anywhere but you're right. Corporate environments are very good at discouraging this type of practise. They want you to do sim and sim and report and sim and before all that have an M.Everything where you spent a shitload of years doing just the same thing instead of the "country boy" way of doing things: make it work and worry about the details as they come at you. Country-engineering is almost always over-engineered. Cost isn't taken out because it doesn't have to be right then. It's not as small or as efficient as it can be because that's not important at the start. The corporate engineered way dictates that you get it to market both fast and efficiently. You're making a zillion so you need cost out up front.
It takes a special kind of person to make something out of nothing, and a different kind of person to take that something and engineer it so that it's not bigger, faster, stronger or power-hungry than it has to be.
Selling a T1 and giving 128K is illegal. And very shady.
Yes, if you are saying you are providing a T1's worth of bandwidth. However the world is full of burstable rates, where you are guaranteed "up to" 1.5Mbps. I don't recall you saying "have a frac T1 and sell a T1", rather "sell it for $300" IIRC.
You cant get digital lines as cheaply in your basement, you need extra hardware and install costs per line can be very expensive.
Again, things may be different here in Canada but T1s are trivial to terminate anywhere. The AS5248s we use terminate a T1 directly so there isn't any need for a separate CSU/DSU; just plug in a bit of CAT5 from the telco's DSU/CSU's DSX-1 port to one of the T1 controller ports on the NAS and away you go. In fact, Portmasters are exactly the same: the telco terminates the T1s with their own equipment and you plug yours in. No more expensive than if it were located at the CO. Less expensive, in fact, since the telco has very strict rules about colocation. Separate entrances, cages, the works.
Bigger companies dont lie as much, they try to offer a TOS and stick to it.
Now the truth comes out. I took offense at your last post because the tone was "little company crook; big company priestlike." We offer a decent ToS with our DSL clients and do everything we can to honour it. Yes, we sell burstable bandwidth and yes we overcommit on the bandwidth we guarantee, but not to the point that we can't bring in another T1 to correct the situation within 20 days. I challenge you to find a "big" company who doesn't do this. It's exactly like monitors: they say 17" but you know damn sure that it's about an inch less unless it's an LCD.
And BTW, I was a founding member of the Spokane ISP Association. I still talk to other people in the field, and have friends working for bbnplanet, uunet, version, att, etc.. Alot of people start at smaller ISPs and tech support. It is a common theme that shady business practices occur.
At the risk of this coming across WAY worse than I really intend it, good for you. At least to me, being a member of some super-high-level tech association means about as much to me as being a member of the local chamber of commerce. It looks good on paper and you have a lot of meetings and discuss important things, but in the end not a whole hell of a lot changes. You still do what you have to to stay in business, and you still do what you can with what you have to keep your customer base happy and encourage it to grow. And I am serious about the challenge. Bigger companies have more incentive to honour the ToS because their customers (often more business than residential customers) are more apt to challenge them and raise a stink.
There's no magic formula that says that big businesses are better than little ones. That was my gripe with your original post.
But your experiences may very, life isnt in absolutes.
Starting at a small BBS turned ISP (cet.com [cet.com]), I seen how the owner would sell a full T1 and split it a dozen ways, scam customers on software packages, replace broken hardware that wasnt broken. I moved onto another ISP, and saw how the salesmen reminded me of car salesmen, "Let me talk to the manager..."
You worked for a crook, not a small ISP/computer store. Don't confuse the two. I've been the admin for a small ISP for four years now and the only thing we do that could be considered shady is overcommit the bandwidth.
People dont see whats going on behind the scenes, how the young kids are working thier ass off to keep the servers up cause they cant hire professional admins. The systems are always having outages and they blame the larger telcos as a "network problem..."
Our servers are rock-solid, even back when we were using the 1.x (2.0.x?) kernels and the then-new iproute2 to prioritize/mark lowlatency traffic because we had a satellite downlink and four-33k6-bonded-modem uplink. In my ealrier post I made mention of a helper script that propped up MySQL every time it fell over; that was the extent of our server problems and with the script the users rarely noticed.
Compare that to today where our DSL uplink (UUNet) does go down and doesn't tell us, despite the service agreement. In your world I bet the big guys never make mistakes.
Smaller ISP's have to cut costs too, I remember when all the ISPs in Spokane moved into the tel-west building so they could cut out the local exchange. Save 200 bux on federal taxes and transport fees. A T1 that costs 900 bux wholesale could be bought for 500, since all they had to do was run some cable down the hallway (overhead).
I don't see anything illegal at all about what I quoted of you. It's ingenious, even. We're going to lease an entire DS3 instead of 28 PRIs to save money; that must be shady to you since it's a small ISP trying to save a buck.
BTW, you don't need to be in the telco to run digital lines; just get the lines terminated at your location. Our AS5248s aren't at a telco, as an example.
Sell the T1 (frac) to 10 people paying you 300 bux, and they pay thier own costs, you could out bid. And then charge them for any hourly work needed.
What's wrong with that? It's called overcommit and it's done by every ISP on the planet, large and small. And if you don't buy into an SLA, you will get billed per hour for every little change you ask for. Sounds pretty above-board and level to me.
There's a certain amount of salesing involved in any business. If you don't ask for the best price you will get screwed by any business. Not so much when you're dealing with joe consumer but B2B transactions (ordering lines, bandwidth, etc.) ALWAYS work this way. List prices are often called the idiot tax, and if your customer doesn't realize that $300/mo for a frac T1 is crazy then it's not your fault. It's called business. If you want to buy a $30000 lemon, they won't stop you.
I think most slashdotters can confirm the shady side of the ISPs. How some run out of computer stores in the back, or BBS's that turned ISP.
how is this shady?? It's called a small business for a reason, for Christ's sake! Was Apple for H-P shady for working out of their garage/shed??
Been there done that, now I work for a major wireless telco, millions of customers, and I never have to be shady.
Ok, big companies aren't shady, but then
Sometimes I just shake my head and say "Umm, if our stockholders only knew....)
Interesting. So as long as it's making money for the stockholders it's okay? How about "I don't have to do the shady shit, so big business is okay. I can install equipment I know works half-assed but I am not the one ordering it to be installed so it's okay."
Oh and a T-1 for every modem rack of 24 incoming lines. you HAVE to buy 24 phonelines at a time as that is how a T-1 is sold. This T-1 is more expensive than the above T-1, as you get charged per call on it.. incoming or outgoing. incoming is cheap, 0.01 per call but it does add up.
You're not far off the mark at all. We get two PRIs at a time to fill each access server. We use the EOL'd Cisco AS5248s because the just work, are dirt cheap and plentiful as hell; we've never had bad hardware and the software problems seem to have levelled out before they were EOL'd. We did get thrown for a loop when we received a shipment of them with DC power supplies, but now we can handle either.:-)
To make things a little easier, our incoming PRIs have a monthly cost but there are no costs per connect. We'd love to use CAS signalling but that's just not available anymore. So instead of 24 lines per T1 we now have 23 (damned ISDN OOB switching). 8-bit clean DS0s are useless for dialup, so ISDN gains us nothing here. Fortunately we can use NFAS on the AS5248s and gain one extra line per NAS by using a single D channel for both PRIs. And yes, when you're a mom 'n pop, every line counts. We'd gain more lines by using AS5300s (4/8 PRIs per box instead of 2) but the equipment is a lot more expensive.
The other thing that took a while to hammer down was the user:line ratios. Over the years we have discovered that 7:1 is pretty much optimal. Busies are relatively infrequent. 6.9 and lower is wasteful (nobody cares if there are free lines after they're on) and 7.1 and up gets nasty real fast.
Software wise, we use GNU-RADIUSd and PostgreSQL. GNU-RADIUSd has never failed us, has SNMP suppport and runs circles around Cistron and all the others. PostgreSQL can handle the (very) large connect database (600M), not bog down when we do some funky queries and doesn't require a helper script to prop it up when it decides to gack (hello MySQL)! Back in the beginning we used Cistron and MySQL -- it barely held up with 300 customers, let alone what we've got now. I really can't say enough good about GNU-RADIUSd; it's an excellent program with a responsive developer. And Postgres? Well I use it everywhere for just about everything. Nary a complaint.
So how are we doing? Not bad. Our price is the lowest in the area (yeah we're winning against the much larger corps, including the telco, for now) and our userbase is growing about 5-7% a month which is keeping us plenty busy. We are profitable and have been for some time now, even when we dropped to $9.95/mo from $24.95/mo (CAD). I just bought an M13 (DS3 demux; terminates a single DS3 to 28 DS1s) from Ebay, as we're rapidly approaching 28 PRIs and we can get a major price break if we buy an entire DS3 instead of individual PRIs. There seem to be plenty of AS5248s on the used market for dirt cheap and we've got the test and the config pretty much nailed down. Five AS5248s per/24 for IP allocation, slightly under 2Mbps peak bandwidth usage for each/24... It's down to a science now and we're now focussing on getting that 5% ramped up since the backend is working 100% and the growth and costing is falling into place.
Kind of... but for photons. A google search for "photomultiplier tube" or "pmt" spits back tons of physics experiments but I was hoping for something like a howstuffworks description to link to in here.
And... can you chain one to a fiberoptic cable?
That's just a physical connection; I don't see why not.
The problem with the detector is that it's possible to build detectors that register single photons, it just requires that someone builds one, and that shouldn't be impossible either.
Umm... I was under the impression that photomultiplier tubes were regularly built this sensitive.
For example, lets look at a situation where I need to tranfer data from the east coast to the west coast and my round trip time is 70 ms. If I have a 32 Mbps link, I can send 64K in about 2 ms. So I have to wait 68 ms until I send another 64K. This gives an effective throughput of less than 1 Mbps. Throw in the slightest amount of congestion and things go even further downhill real fast.
Thank you for that detailled explanation. It seems that my shit detector is a little trigger-happy.:-) Your comment on the "magic algorithm" is spot-on. I could think of a way to pre-determine an algorithm which recodes 64k chunks of data by precomputing all possible 64k datastreams and then simply selecting the datastream which represents that 64k (or whatever size). Anyway the other comments in this article have already attacked the idea on its use of UDP and its lack of net-friendliness. I predict a number of angry customers who paid $300k for a pair of these and whos data is getting dropped/throttled like crazy by their ISPs or the ISPs in the middle.:-)
Slashdot Mirror
I think it's a kludge that kilo means 1024 in kilobyte, while it means 1000 in kilometer, kilogram, kilonewton, kilovolt etc. It's time we give it up, and accept correct terms. I agree kibi sounds ridiculous, but that's just a matter of habit. We'll get used to it.
I think it's a kludge that hacker means criminal in the media, while it has very good connotations in our circles. It's time we give it up and accept correct terms. I agree cracker sounds ridiculous, but that's just a matter of habit. We'll get used to it.
So you're saying that ten transmeta processors put out as much heat as a single P3 card solution? What exactly is the problem?
This also makes it a lot safer to work with, i.e. you won't burn your eyes out if you accidently look into it."
The site says the EIRP is 10kW -- you will most certainly be hurt if you stare into this thing!
The letter 'x' is a multiple stroke letter in grafitti (if you look at the grafiti manual)
Hmm, I've never done an X like that. Backwards K for me. :-) But even punctuation and others are still one-stroke.
It's brilliant from a stratigic viewpoint. Kind of like giving a little bit of money to a bunch of ignorant Arab terrorists to keep the Russians from taking over a certain country.
+10 User doesn't have his head up his arse!
Further, the Patent *specifically* states 'a "unistroke" is a single, unbroken stroke', whereas graffiti has a number of multiple-stroke characters.
Which Grafitti characters are multiple-stroke, exactly? Every single one I can think of (including the funky characters) are all single stroke -- you lift the pen off the screen and the PDA attempts to recognize the stroke you just made. (I do not count charset shifts as multiple strokes...)
The LinuxDA PDA, OTOH, does have multiple-stroke characters, and many of them. Pisses me off, actually, because I'm so used to the Palm Grafitti.
The OS allows access to raw sockets and, therefore, the entire kernel.
Go read it again. Raw sockets is not a security flaw. Unix (including Linux and OSX) has them too. All it means is that it's easy to spoof packets. That's it.
a) KOffice gets a Word and Excel Input/Output doc filter which works reliably.
While I don't have much trouble with Word/Excel doc import (big fancy ones yes, but straightforward ones no), I don't know why straight RTF isn't supported in KWord. Crazy.
b) Konqueror speeds up - I loved it and then its startup time seemed to slow down drastically.
I hear that Konq has totally rewritten their JScript interpreter. I hope the hell they fixed the popup problem... popups normally get a prompt action for me (i.e. "this site is trying ot use a popup. Allow?") but for Flash sites the popup never ever gets prompted, which drives me insane. Especially when 8 or 10 windows pop up because the JScript interpreter doesn't provide the right answer. UGH!
Speedups will be good though. I wonder if they were able to speed up any further than the 2.2.2 and prelinking. Startup time is still ugly for most KDE apps. That is one thing I noticed right away. Every time I start up xchat, it's onscreen almost immediately after I click the button. Konq, KWord, KMail... ~3-5s pause. Prelinked. On a Cel300 @1024x768x24 with 256M of RAM and no swap. Shouldn't be this slow.
I want to install Nvidia drivers on a Dual PIII.
Mandrake 8.1
X 4.1 something.
I dunno about Mandrake but it was very easy under Slackware (dual Cel466, Slack 8.0, X 4.1.0) -- found out about #nvidia from #linuxhelp. Went there, chanbot introduced itself and told me to !files to see what was available without my asking. Found the 1241 drivers, grabbed, compiled, insmod'd and modified XF86Config to use the new driver. Boom, it was up in under 30 minutes.
Sometimes it can be complicated to install something on Windows.
Sometimes it's just impossible under X on Linux.
And Johnny Lawnmover WON'T have the same patience as me and will install a Bootleg Windows within 30 minutes.
True enough; But don't forget the other side of those first two statements:
Sometimes things are hard to do in Windows that are trivial under Linux (X).
Sometimes things are just impossible to do under Windows.
And as far as Johnny Lawnmower is concerned you're absolutely right. But let's look at people other than the average schmuck who only wants to surf porn and listen to mp3s: Let's talk the technical crowd, the academia and people like me who are sick of being trapped in the Windows world where pretty much everything you see and touch is covered in a thin layer of Works but if you scratch through most things you find out there really are monkeys controlling the machine. I'm sick of having to pay for support, pay for updates and pay to hear "just reinstall." For people like me, Linux is a godsend. I can get done what I want and make my computer do what I want instead of moulding my wants and needs to some predetermined way of how things have to work.
You can have war stories about any OS and any hardware/driver. Try to keep that in mind.
The way it's setup here, it works with anything that can print. We do use AutoCAD LT 98 with it too. It's quite transparent for the users, and for the application.
nonono... this is for an automated system in our intranet. i.e. customer wants to see/play with drawing without downloading the entire thing. Yeah I know Acrobat creator can print to pdf; that's not quite what we need (unless it is perhaps possible to script it out from a network connection...)
The problem with the readers (voloview et al) is that they only work for Windows; there's no decent dwg viewer for Linux that I have been able to find, so I'm trying to find a good convertor. :-)
Because some partners and clients who are still using Office 95/97 (and even 2k itself) sometimes couldn't open our Office files, we now only send stuff in pdf (automatic conversion through a linux virtual shared printer and ps2pdf).
Know of anything that will work for AutoCAD 14 .dwg files and AutoDesk Inventor files? (either to DXF or to .pdf) -- I'd love to be able to provide that same functionality to people who use our drawings. Yeah you can save to .dxf but that's a pain for the draftsmen who have a zillion drawings around WITHOUT having dupes in .dxf format.
One is called Deleted Item Retention Time. You set the number of days and when a user deletes an email it's not really deleted for the specified days. If he realizes he made a mistake restoration is from the Outlook client and takes seconds saving the admin time from going to the back up tapes. For businesses like law firms it's a life saver since they are required to keep records and emails for five years or so. They simply buy a lot of storage and set a deleted item retention item of 1600 days or so and it's a secondary back up.
Just out of curiosity, how do you get an email back when it's deleted when the only thing on tape is a collection of large proprietary binary files which make up the entire private mail store?
Then third party back up programs have a feature called brick level back up's where you can back up individual mailboxes. If you delete on by accident restoration is simple. Exchange 2000 has this feature out of the box.
Answered my own question, I suppose. :-)
And the global address book is great. Users don't have to keep their own huge address book and greatly minimizes the calls to the admin of I sent out this email but it came back returned and asking you to track down an email address.
Perhaps I'm using it totally wrong but at least in Exchange Server 5.5 the GAL is for local mailboxes only. This was years ago but we had to create a public folder and use Outlook to store contacts in it.
Sure you can cobble together a few products for most of the functionality and perform some of the usability features manually, but you'll spend more time while the CEO is asking you to restore an email from a year ago.
I'll be the first to admit that Exchange Server and Outlook are one killer combination. However like I have stated several times over the years, I'll be god-damned if I'm going to lock up my company's data like that. The data store is proprietary. The access tools are proprietary. Maybe I'm getting old and crochety but I've been bit before (too many times in fact) to just let it go. I want to be able to get the format of the data files and protocols so that when I want to do X I can go do it, hire someone to do it or otherwise do what the hell I want with my data, when I want and without some motherly giant cooing "Now now, dear. You don't really need to do it like that."
Buy an always-on UPS. It will take the "unclean" power and re-create it for your stereo etc.
Tellyawhat: You show me any kind of high frequency getting past the 50/60Hz transformer in the audio power supply and/or the switchmode power supply secondary inductor (DC choke) and the filter caps sprinkled all over the PCB and I'll be surprised. Show the same noise appearing in the output of said audio device and I'll buy you an always-on UPS.
The kind of noise you describe getting through is such utter bullshit. The DC power supplies in practically all electronic equipment are capable of filtering this crap out and the measures for A/V devices are double. Remember that in a linear power supply you have a huge laminated-core transformer that will absorb high frequency noise as eddy current losses. In switchers you have a high-frequency (usually 60kHz and up) oscillator and the chokes and filters on the output are designed to give you as-close-to-zero ripple as possible. And after that you have high frequency, low-ESR caps across every IC and tons of filtering on the audio inputs and outputs to keep things sane.
Always-on UPSes are useful for really shitty lines and equipment which is sensitive to the fast switchover of traditional UPSes. It's once again proven that you can get superb advice from an AC.
"Flag any sentence contianing the phrase "pheremone" as potential bullshit"
Alright, I'm no biologist, but I thought that pheremones were well proven in the insect world. I thought it was also proven that smells can induce responses in the brain on subconscious levels.
As I said though, IANAB.
You *could* use a spectrum analyzer... or you could make creative use of directional antennas and the abilities of some off-the-shelf equipment to do site surveys.
A spectrum analyzer without the directional antenna is as useless as tits on a bull in this scenario. I would have done it the same way as the fellow you describe did. Use a directional antenna with a device that can tell you the signal strength and point it around. Hell get a couple and triangulate the position, then take a high power rifle with a decent scope and take it out. :-)
2.4 is now into maintenance mode, where it should be bug fixes and new drivers etc only. That does not mean they should be assumed to work for everyone - if you dont have the time (and the patience) to test them to ensure that they work for you then you should be using whatever Debian/Redhat/whoever chooses to test and ship.
Don't get me wrong; I totally agree with you. I was taking exception to what you had said in your previous post that they only added things to the kernel that you could opt-in to on the even-middle-number kernels.
The core kernel will remain stable, new drivers will be added where they have no effect on the rest of the kernel - so its the users choice if they want the new drivers or not.
Like the vm changes? That was just a driver change that anyone could swap in or out of the stable 2.4 kernel, eh?
Yes the anonymous cowards are a pain but sometimes they do make a good point.
The F-CPU project has been around for a good long while, and is supposed to be 64-bit. The idea of an opensource CPU is a neat idea, I must admit. Unfortunately it hasn't been updated in a while.
Marlon isn't a Linux/Unix guru, but his workshop is that of a "hands on" guy; littered with parts, antenna wires, soldering irons and tools. No spectrum analyzers, digital scopes, or automated test beds... Marlon uses the built-in site survey capabilities of some wireless radios to do 75% of what that equipment would do at 1% of the cost.
Umm... as a distributor/sales, you would never use spectrum analyzers, digital scopes or ATE; you buy your products and if they break, send 'em back under warranty. To do otherwise is folly. Now if Marlon were manufacturing this equipment and didn't have the test equipment mentioned, he'd also be stupid. He'd be wasting a lot of time guessing what was wrong.
How do I know? I'm in both positions. I design equipment but I also resell other equipment. If you've bought a warranty, you better damn well use it.
This sort of "can do" thinking is why I think that the spirit of America shines brightest in its small towns and rural areas and not in the corporate hallways. Marlon learned enough about microwave systems to be able to design workable wireless networks (and DSL systems too, although the article doesn't speak to that), sell them, install them, connect them, and keep them working.
Strong will and common sense go a hell of a long way anywhere but you're right. Corporate environments are very good at discouraging this type of practise. They want you to do sim and sim and report and sim and before all that have an M.Everything where you spent a shitload of years doing just the same thing instead of the "country boy" way of doing things: make it work and worry about the details as they come at you. Country-engineering is almost always over-engineered. Cost isn't taken out because it doesn't have to be right then. It's not as small or as efficient as it can be because that's not important at the start. The corporate engineered way dictates that you get it to market both fast and efficiently. You're making a zillion so you need cost out up front.
It takes a special kind of person to make something out of nothing, and a different kind of person to take that something and engineer it so that it's not bigger, faster, stronger or power-hungry than it has to be.
Even if you dont quote me right.
Where didn't I quote you right?
Selling a T1 and giving 128K is illegal. And very shady.
Yes, if you are saying you are providing a T1's worth of bandwidth. However the world is full of burstable rates, where you are guaranteed "up to" 1.5Mbps. I don't recall you saying "have a frac T1 and sell a T1", rather "sell it for $300" IIRC.
You cant get digital lines as cheaply in your basement, you need extra hardware and install costs per line can be very expensive.
Again, things may be different here in Canada but T1s are trivial to terminate anywhere. The AS5248s we use terminate a T1 directly so there isn't any need for a separate CSU/DSU; just plug in a bit of CAT5 from the telco's DSU/CSU's DSX-1 port to one of the T1 controller ports on the NAS and away you go. In fact, Portmasters are exactly the same: the telco terminates the T1s with their own equipment and you plug yours in. No more expensive than if it were located at the CO. Less expensive, in fact, since the telco has very strict rules about colocation. Separate entrances, cages, the works.
Bigger companies dont lie as much, they try to offer a TOS and stick to it.
Now the truth comes out. I took offense at your last post because the tone was "little company crook; big company priestlike." We offer a decent ToS with our DSL clients and do everything we can to honour it. Yes, we sell burstable bandwidth and yes we overcommit on the bandwidth we guarantee, but not to the point that we can't bring in another T1 to correct the situation within 20 days. I challenge you to find a "big" company who doesn't do this. It's exactly like monitors: they say 17" but you know damn sure that it's about an inch less unless it's an LCD.
And BTW, I was a founding member of the Spokane ISP Association. I still talk to other people in the field, and have friends working for bbnplanet, uunet, version, att, etc.. Alot of people start at smaller ISPs and tech support. It is a common theme that shady business practices occur.
At the risk of this coming across WAY worse than I really intend it, good for you. At least to me, being a member of some super-high-level tech association means about as much to me as being a member of the local chamber of commerce. It looks good on paper and you have a lot of meetings and discuss important things, but in the end not a whole hell of a lot changes. You still do what you have to to stay in business, and you still do what you can with what you have to keep your customer base happy and encourage it to grow. And I am serious about the challenge. Bigger companies have more incentive to honour the ToS because their customers (often more business than residential customers) are more apt to challenge them and raise a stink.
There's no magic formula that says that big businesses are better than little ones. That was my gripe with your original post.
But your experiences may very, life isnt in absolutes.
Amen. Absolutely. :-)
Starting at a small BBS turned ISP (cet.com [cet.com]), I seen how the owner would sell a full T1 and split it a dozen ways, scam customers on software packages, replace broken hardware that wasnt broken. I moved onto another ISP, and saw how the salesmen reminded me of car salesmen, "Let me talk to the manager..."
You worked for a crook, not a small ISP/computer store. Don't confuse the two. I've been the admin for a small ISP for four years now and the only thing we do that could be considered shady is overcommit the bandwidth.
People dont see whats going on behind the scenes, how the young kids are working thier ass off to keep the servers up cause they cant hire professional admins. The systems are always having outages and they blame the larger telcos as a "network problem..."
Our servers are rock-solid, even back when we were using the 1.x (2.0.x?) kernels and the then-new iproute2 to prioritize/mark lowlatency traffic because we had a satellite downlink and four-33k6-bonded-modem uplink. In my ealrier post I made mention of a helper script that propped up MySQL every time it fell over; that was the extent of our server problems and with the script the users rarely noticed.
Compare that to today where our DSL uplink (UUNet) does go down and doesn't tell us, despite the service agreement. In your world I bet the big guys never make mistakes.
Smaller ISP's have to cut costs too, I remember when all the ISPs in Spokane moved into the tel-west building so they could cut out the local exchange. Save 200 bux on federal taxes and transport fees. A T1 that costs 900 bux wholesale could be bought for 500, since all they had to do was run some cable down the hallway (overhead).
I don't see anything illegal at all about what I quoted of you. It's ingenious, even. We're going to lease an entire DS3 instead of 28 PRIs to save money; that must be shady to you since it's a small ISP trying to save a buck.
BTW, you don't need to be in the telco to run digital lines; just get the lines terminated at your location. Our AS5248s aren't at a telco, as an example.
Sell the T1 (frac) to 10 people paying you 300 bux, and they pay thier own costs, you could out bid. And then charge them for any hourly work needed.
What's wrong with that? It's called overcommit and it's done by every ISP on the planet, large and small. And if you don't buy into an SLA, you will get billed per hour for every little change you ask for. Sounds pretty above-board and level to me.
There's a certain amount of salesing involved in any business. If you don't ask for the best price you will get screwed by any business. Not so much when you're dealing with joe consumer but B2B transactions (ordering lines, bandwidth, etc.) ALWAYS work this way. List prices are often called the idiot tax, and if your customer doesn't realize that $300/mo for a frac T1 is crazy then it's not your fault. It's called business. If you want to buy a $30000 lemon, they won't stop you.
I think most slashdotters can confirm the shady side of the ISPs. How some run out of computer stores in the back, or BBS's that turned ISP.
how is this shady?? It's called a small business for a reason, for Christ's sake! Was Apple for H-P shady for working out of their garage/shed??
Been there done that, now I work for a major wireless telco, millions of customers, and I never have to be shady.
Ok, big companies aren't shady, but then
Sometimes I just shake my head and say "Umm, if our stockholders only knew....)
Interesting. So as long as it's making money for the stockholders it's okay? How about "I don't have to do the shady shit, so big business is okay. I can install equipment I know works half-assed but I am not the one ordering it to be installed so it's okay."
Oh and a T-1 for every modem rack of 24 incoming lines. you HAVE to buy 24 phonelines at a time as that is how a T-1 is sold. This T-1 is more expensive than the above T-1, as you get charged per call on it.. incoming or outgoing. incoming is cheap, 0.01 per call but it does add up.
You're not far off the mark at all. We get two PRIs at a time to fill each access server. We use the EOL'd Cisco AS5248s because the just work, are dirt cheap and plentiful as hell; we've never had bad hardware and the software problems seem to have levelled out before they were EOL'd. We did get thrown for a loop when we received a shipment of them with DC power supplies, but now we can handle either. :-)
To make things a little easier, our incoming PRIs have a monthly cost but there are no costs per connect. We'd love to use CAS signalling but that's just not available anymore. So instead of 24 lines per T1 we now have 23 (damned ISDN OOB switching). 8-bit clean DS0s are useless for dialup, so ISDN gains us nothing here. Fortunately we can use NFAS on the AS5248s and gain one extra line per NAS by using a single D channel for both PRIs. And yes, when you're a mom 'n pop, every line counts. We'd gain more lines by using AS5300s (4/8 PRIs per box instead of 2) but the equipment is a lot more expensive.
The other thing that took a while to hammer down was the user:line ratios. Over the years we have discovered that 7:1 is pretty much optimal. Busies are relatively infrequent. 6.9 and lower is wasteful (nobody cares if there are free lines after they're on) and 7.1 and up gets nasty real fast.
Software wise, we use GNU-RADIUSd and PostgreSQL. GNU-RADIUSd has never failed us, has SNMP suppport and runs circles around Cistron and all the others. PostgreSQL can handle the (very) large connect database (600M), not bog down when we do some funky queries and doesn't require a helper script to prop it up when it decides to gack (hello MySQL)! Back in the beginning we used Cistron and MySQL -- it barely held up with 300 customers, let alone what we've got now. I really can't say enough good about GNU-RADIUSd; it's an excellent program with a responsive developer. And Postgres? Well I use it everywhere for just about everything. Nary a complaint.
So how are we doing? Not bad. Our price is the lowest in the area (yeah we're winning against the much larger corps, including the telco, for now) and our userbase is growing about 5-7% a month which is keeping us plenty busy. We are profitable and have been for some time now, even when we dropped to $9.95/mo from $24.95/mo (CAD). I just bought an M13 (DS3 demux; terminates a single DS3 to 28 DS1s) from Ebay, as we're rapidly approaching 28 PRIs and we can get a major price break if we buy an entire DS3 instead of individual PRIs. There seem to be plenty of AS5248s on the used market for dirt cheap and we've got the test and the config pretty much nailed down. Five AS5248s per /24 for IP allocation, slightly under 2Mbps peak bandwidth usage for each /24... It's down to a science now and we're now focussing on getting that 5% ramped up since the backend is working 100% and the growth and costing is falling into place.
Aren't photomultiplier tubes akin to... opamps?
Kind of... but for photons. A google search for "photomultiplier tube" or "pmt" spits back tons of physics experiments but I was hoping for something like a howstuffworks description to link to in here.
And... can you chain one to a fiberoptic cable?
That's just a physical connection; I don't see why not.
The problem with the detector is that it's possible to build detectors that register single photons, it just requires that someone builds one, and that shouldn't be impossible either.
Umm... I was under the impression that photomultiplier tubes were regularly built this sensitive.
For example, lets look at a situation where I need to tranfer data from the east coast to the west coast and my round trip time is 70 ms. If I have a 32 Mbps link, I can send 64K in about 2 ms. So I have to wait 68 ms until I send another 64K. This gives an effective throughput of less than 1 Mbps. Throw in the slightest amount of congestion and things go even further downhill real fast.
Thank you for that detailled explanation. It seems that my shit detector is a little trigger-happy. :-) Your comment on the "magic algorithm" is spot-on. I could think of a way to pre-determine an algorithm which recodes 64k chunks of data by precomputing all possible 64k datastreams and then simply selecting the datastream which represents that 64k (or whatever size). Anyway the other comments in this article have already attacked the idea on its use of UDP and its lack of net-friendliness. I predict a number of angry customers who paid $300k for a pair of these and whos data is getting dropped/throttled like crazy by their ISPs or the ISPs in the middle. :-)