Homebrew S/ADSL

schvin writes: "ISP Planet has a brief article about rolling your own DSL connection. One person in Washington state has extensive information on how he got his home-brewed SDSL fully functional." This is great, and I wish I knew about it before I had all this hooked up here. Save money! Do it yourself! It's the GNU way!

I'm using this method.. works great

I live in a rural town as well, population 4000, and at least 60 miles to a town with more than 2 red lights. and i have SDSL here, its fucking sweet.

my isp has a DSLAM, southwestern bell charged $175 for the dry pair installation, and im not sure what the monthly rate is, its included in the isp cost. I pay $75/month for 400kbps+5ips frame relay, not bad at all since i never thought I would see higher than 28.8 a month ago. The router is a Lucent Pipeline ($600), its pretty nice but is having some problems mentioned by someone else here.. massive packet loss at times and the connection eventually dropping. rebooting seems to have cleared it up though, who knows..

ascend% show revision
SCPipe system revision: l.s15 7.10.5

apparantly the telco was hesitant about allowing the dry pair to be used for this purpose at first but they are required by law to offer the service, no skin off my back.

fastest transfer rate i've seen is 47K/sec and quake3 pings are normally sub-50.

Re:cable v. dsl

With a DSL line, the connection between you, and your service provider, is a constant, flat rate. No matter what your neighbors do, you never slow down. Theoretically. There is a potential bottleneck tho, your service provider, has to have a big enough pipe from their office, to the internet to support all the users. But since most of the time it's a phone company, they can certainly afford the pipe(s) out to the net.

Unfortunately most (all?) providers overcommit. The level of overcommit is actually able to go up with the more customers they get because the whole thing shares really well if people are just browsing. Overcommits on dialup-class service can easily hit 100:1 on bandwidth and 7:1 on lines. It works just fine as long as you don't have a two-digit percentage of the people you're overcommitting doing some kind of long high-bandwidth transfer.

I'm not sure what the numbers are like for DSL/Cable but they can't be much better.. That's a wad of bandwidth they're giving each person and DS3s are neither cheap nor enough to service a lot of 1-meg customers doing fulltime downloads.

Re:Cheap?

This is low - he is getting symmetric 768kbps DSL for about 20 bucks a month, with a 100% CIR. That's bascially a fractional T1 in all but name. It would cost probably $8000-$10000 per year including ISP service in the real world. I'd say he's getting a steal.

--

Re:Interesting...

The circuit he is talking about is not a 56k circuit. The 56k P-t-P you are talking about is a DS0 w/ 8kb reserved for signaling (64kb - 8kb).

An alarm pair is (IIRC) a lower-bandwidth P-t-P circuit that is routed differently through the Telco network. Wish I remember more specifics about it, but I've never ordered one myself. There's no QoS guarantee for circuit noise, etc. Caveat Emptor.

Regarding the actual implementation of this, there was a discussion on this topic on both the ISP-EQUIPMENT and Cisco GroupStudy mailing lists. The general consensus was this:

1) Make sure the distances from the PoP to the CO is less than 12,000 feet for the best chance.
2) Not all Bells will allow you to get these unloaded copper pairs, and those that do will often not give you the highest quality circuits.

I've never done it myself, but there seems to be a fair amount of success in the matter in the marketplace.

Matt


Matthew J Zito, CCNA

Re:Finally a Solution for Rural Areas

Most likely this won't do much for your mom. The problem is, who's gonna be at the other end of the pair? This is not a DSL line to the internet. It's a high speed connection between 2 places. It's basically a cheap way to set up a MAN (Metropolitan Area Network).

$20/mo. dry pair?

$20/mo. for a dry pair from telco is an incredible deal. Especially for the distances stated (20,000 feet). I have seen customers pay up to $500-$1000 per month to Pacific Bell for an unconditioned dry pair of less then a mile. I suppose the shortage of copper facilities in most areas of No. California may have something to do with this...

Anyhow, sounds like a great solution at a good price, but readers from various areas should beware that they may find the cost of getting a dry pair through their local telco to be significantly more expensive.

bjsvec

Re:Dry pair?

And T1 runs over balanced synchronous lines.. known usually (I believe) as HDSL, and they consist of two pairs instead of one, and are rigorously tested for quality.

T1 is still T1. Frame is still Frame. You're describing different layers.
Though most ISP's that sell you 'T1' access really just sell you T1 access to a frame cloud, then do whatever from there on out.

Alarm circuits are single pair copper, exactly what is used in your house for your analog phone, but wihtout the dialtone/switching gear hooked to it.

this also works with T1 CSU/DSUs

I have used SDSL or HDSL equipment on both private and telco dry pairs in several installations. Usually, it just fires right up -- especially if one of the end points is literally across the street from the CO! I have used both PairGain and Tut Systems equipment. The PairGain stuff is nicer, but it is way more expensive.

Some may be surprised to know that you can also do this sort of thing with standard T1 CSU/DSU units, wired up in a cross-over fashion. Set the "line build out" setting to the maximum, and as long as the cabling is less than ~3500 feet, it will work. If the line is of marginal quality you will see some framing and CRC errors. But, it is usually nothing much worse than an average point-to-point telco T1. You can get used CSU/DSUs and routers cheaper than xDSL gear at the moment, too. This also has the advantage of letting you combine voice and data on that "circuit", if you are trying to connect two business locations together as cheaply as possible.

SDSL will give you 2MBit/second on a single pair of dry copper. HDSL gives you 768kbit/second on a single pair of copper, which can be combined for a 4 wire connection at 1.5Mbit/second (common with PairGain stuff).

Another little secret of the telcos is that these days almost all "T1" type circuits (including frame relay ports and ISDN PRIs) are actually delivered to you on PairGain or Paradyne type HDSL equipment. These are the so called "smart jacks", the white/gray boxes that you plug your CSU/DSU into on your end. The telco doesn't have to put all of that time into engineering a true "repeatered T1" anymore -- they can just grab some pairs and fire up the HDSL and turn up a T1 circuit in a matter of hours instead of days or weeks. They do this all of the time for their overpriced dedicated circuits, but for some reason they have to spend years "trialing" DSL for mass consumer use. uh-huh.

Re:dry copper?

What exactly makes the line between me, and say, my friend's house unsuitable for DSL?

The fact that there isn't a line between you and your friend's house. Simplifying it a bit (a lot?), there's a line from your house to the CO, and a line from your friend's house to the CO. When you dial your friend's number, the CO switch connects your line to his line. When you hang up, the lines are no longer connected.

Ok, so you can't use them because there's no dedicated (nailed) connection. It's only there when you ask for it. So, what if you ask for it all the time? Well, you don't have copper running straight from building to building, and at some point your analog signal (you talking into your phone) gets encoded into a digital signal. And only 64 or 56k is allocated for that signal. The lines run from the CO to your house COULD support a lot more than 56k -- but back before phone lines were used for anything more than phones, there was no reason to, and every reason not to: aggregating 24 64k channels = 1 T1. What if each channel was, well, 768k?

One more thing that should be pointed out for this article is... the "roll your own DSL" is somewhat confusing. The guy that did it could do it, because he already had a connection to the internet. The "roll your own dsl" part was the fact that he was then able to connect other people to him, via phone lines / DSL modems. Just don't expect to buy some DSL modems, order a dry circuit, and magically have internet connectivity..

dry copper?

Could someone who knows way more than me explain what the difference between "dry" copper and the existing structure of a line between me and the phone company. What exactly makes the line between me, and say, my friend's house unsuitable for DSL?

Sorry for the basic question, but I have DSL in my house and still don't quite get it... Is this similar to the various "home network over your existing phone line" offerings?

Want to work at Transmeta? Hedgefund.net? AT&T?

Cheap?

I think a project on this scale is for people who know what they're doing with routers and networks.

The article suggests that the cost of one of these 'Do-It-Yerself' DSL lines is low compared to that of the phone companies charges. Where I live, Installation and Modem are free, and its $50.00/month for ISP & Line Service.(GTE) You can't beat a deal like that...(especially since the modems he's talking about are in the $500-$1000 range..) This price provides a constant 768k/128k down/upstream, and makes me perfectly happy.

--
Homer: "No beer, No TV make Homer something something";
Marge: "Go crazy?";
Homer: "Don't mind if I do!"

ok, this is just plain wrong

these people are RIPPING OFF THE PHONE COMPANY!!! that is very immoral of them. how would you feel if someone was ripping you off by going around your rules??? the phone company charges like $1000+ a month for this type of stuf!!!! YOU ARE STEALING if you do this!!!!!!!!!

plus its bad too because there probabley isnt a TOS with this that bans servers!!!! so people would be able to run servers!!!! that is ripping off the corprate sites on the internet!!!! if you want to run a site you gotta pay lots of money to a hosting company!!!! otherwise youre STEALING.

you ppl need to quit STEALING. IT IS WRONG TO STEAL FROM THE PHONE COMPANY!!!!!!!!!!!!!!!!!!!!!!

Two local loops, added distance

Running short-haul modems (and now, point-to-point DSL equipment) over solid copper circuits has always been an interesting activity. This kind of thing has been getting done for decades in campus type situations, and for organizations that own multiple adjacent buildings.

The problem with deploying an ISP-style rollout of this technology is that the distance between the telco's central office and the ISP's POP now has to be added to the total length of every circuit. That can make a difference -- as anyone who lives on the outskirts of a town will tell you, with DSL every foot of cable counts!

I'd like to see this kind of thing tried with IDSL. IDSL uses 2B1Q encoding rather than the G.Lite technology typically used in an ADSL circuit -- it's basically an un-channelized ISDN line that terminates in a DSLAM instead of at the telco switch. The top speed it runs at is only 144kbps, but it can run nearly twice the distance that other DSL technologies can run -- and unlike ADSL and SDSL, the telco can extend IDSL circuits using U-loop repeaters. I'm considering ordering an IDSL circuit for my home.

Considering that the article we're currently commenting on discusses deployment of broadband in sparsely populated towns, it's likely that the extra distance will come in handy. I wonder if anyone is manufacturing peer IDSL modems?

--

Re:Dry pair?

The thing is, when the telco wires into an area, they don't just run a few pairs, they run a *Lot* of pairs.

When you order a dry pair, essentially they just figure out what they have to do to get a connection from point A to point B - this usually goes through any number of cables that converge at various points on regular punchdown blocks.

The problem is, in many areas, you can't get a dry pair for love or money.

Well, that's not true, in USWest country, you can get a dry pair for one heck of a lot of money. Classically, yes, they are $20/mo. As soon as USWest figured out people were using them for data, they (legally) priced them as data-capable circuits for several hundred dollars a month, so that they wouldn't hurt their frame relay sales.

Basically, they now charge about as much for a dry pair as they do for frame relay. So why bother.

This sounds like he's investing about $1200 per connection in hardware - he might have been able to pull off the same thing using off the shelf wireless gear and directional antennae for about the same money. Wavelan cards with appropriate signal amps and antennae can go 3 or 4 miles easily.

The real problem with wireless, within a city, tho, is that non-licensed bands require a very clear line of sight.

The common misconception is that you essentially need a clear area about the size of the transciever element to get from here to there. This is completely, depressingly untrue.

Pretty much all the unlicensed gear is 2.4ghz - that sounds great, doesn't it? It's not.

At a 7 mile distance, a 2.4ghz signal needs a corridor fourty feet wide with no obstructions. And just one tree branch in the way can screw with your signal.

They call this the fresnel zone. 2.4ghz is especially sensitive to obstruction because the leaves of most trees make wonderful quarterwave antennae for the signal, and you get lots of reflection off of solid objects like brick or cement.

If you're serious about a point to point connection, licensed bands are the way to go. a 23ghz (note the lack of a decimal point) link over the same distance requires more like a 12 foot corridor.

That's the problem with wireless - the vendors hype it up, but in reality it's actually problematic.

It's late, and I don't know where I'm going with this. G'nite.

Doing HDSL In The UK

You can do this in Britain as well - you need to ask British Telecom for a "Key-line Baseband" line. They'll only do it if both ends are in the same exchange and they don't like selling it, but Mailbox Internet [mailbox.net.uk] sell it as a product in the Fulham area of London.

Using the Pairgain kit you can drive a single pair up to about 1.1meg, or if you use 4-wire circuits you can get about 2meg. The greater the distance though, the lower the signal, you have to remember that.

As I recall, Demon [demon.net] used to give it away to their staff as a perk - Mailbox [mailbox.net.uk] seem to be doing that as well now.

FWIW, we put a link into the local pub, The Southern Cross.

It's not necessarily as good as it sounds

As an expat Aussie living in NYC I can tell you that the bandwidth is nice when it works, but the local Telco here (Bell Atlantic) are at least 2 orders of magnitude worse than Telecom was as a monopoly at customer service.

They have Lied out right to me 3 times (that I caught)

Have inept network techs, but they are level three techs

To get to a level 3 tech you need to be put on hold for around 3 hours by a level 2 tech.

Level 2 techs probably have DSL at home, but if you mention Linux, BeOS, Windows 2000, SMP, Home Networks, any setup where there are more cards in your box than a video card and NIC, or any TLA that confuses them you will be put on the not supported treadmill.

To get to a level 2 tech you need to be put on hold for 4-6 hours (seriously EVERY TIME) by a level 1 tech.

Level 1 techs may understand english.

In the past week I have had to manually resynch my modem 5-6 times a day.

They schedule maintenance where they take down and entire suburbs residential connections, and then bring them back for 1 minute of every 3 for an entire weekend, without ANY publically accessible pre-warning.

Have no newsgroup

Have mail and news servers that are thoroughly overloaded.

And to top it all off, sometimes the bandwidth is slower than a 56k connection, AND I'm paying for the second tier of service.

It is almost as if they are trying their level best to beat the previous bench mark for the worst customer service ever.

I've got dibbs on the first lawsuit.

It's only a matter of minutes before the suits jump on this one!

tcd004
LostBrain [lostbrain.com]

Some Line Lore (slightly OT)

It's been a while since I ordered lines (and that for radio broadcast use), but this should cover the basics: The important thing about the "dry pair" is that it is solid copper from one end to the other, with no intervening power supplies or electronics. They're sometimes called "alarm circuits" because they're most frequently used for burglar alarm (or other telemetry) where it's necessary to be able to tell if the line has been cut or shorted. (Applying constant DC and monitoring the current lets you do this easily. Yes, there are other ways, but this is old technology.) Dry pairs can usually be ordered between two points served by a single exchange, but they're almost impossible to come by if you're trying to get from one part of a metro area to another. (The telco would rather multiplex your service onto one of their wideband interexchange circuits.) In the case you're asking about (between your house and your neighbor) you should be able to get it. Hopefully you can talk the installer into doing the hookup at the neighborhood terminal block, rather than going all the way to the exchange and back. (Big performance difference!)

One of the things that may be lurking on your line (and that used to give us fits when we were trying to set up remote broadcasts) is a "loading coil". These are put on long circuits to even out the audio frequency response, but unfortunately that means "even" for 10-15khz and below, and only if you're using ancient low-impedance equipment. You certainly don't want them there for digital use, but as they may be installed at intervals between the exchange and you, it's often difficult to convince the telco to remove them. (Short story: In one installation we were feeding audio to a transmitter site. All the available lines were in the same cable with 6 pair loaded and 4 unloaded. The telco first tried to put us on the loaded circuits, but we couldn't crank enough high frequencies out of them to make proof-of-performance standards. After a lot of arguing, we finally got them to move us over to the unloaded ones, which worked fine. Later on they moved the interexchange part of the circuit from a dedicated analog circuit to a channel on a T-1: Their mod/demod equipment added so much garbage that we finially gave up and put in a dedicated 2-hop microwave. Sigh!) Also in some areas you may not even be copper all the way back to your exchange. A number of companies are starting to install neighborhood boxes that do the copper to fiber conversion out on a pole or in a manhole somewhere near you.

Other posters have noted that you may also run into problems with tariffs. Rates from the major telcos have the interesting wrinkle of charging different prices for different applications that use the same kind of circuits. If your company gets huffy, you may have to do some fast talking ("Sure it's an alarm... it's sending security telemetry from my house to my office!"). Whatever you do, don't mention anything about "computer networks" or "digital modems" :-) Also, be careful to watch your signal levels. Telcos tend to get pissed if a bunch of subscribers start calling up with complaints about your signal bleeding through on their phones.

It looks to me like this guy's in about the best situation: He's served by a locally-owned small-town company with only one exchange building, and that doesn't mind him bending the rules (or may not even have any), and is willing to cooperate with his experimenting. Just don't think you can get similar cooperation from Ameritech or PacBell!

Re:CO-to-CO hops?

DSL connections do not work between COs for the simple reason that there is no unloaded copper between COs. Most COs are hooked together via frame clouds/ATM/SONET/what have you. Once the individual pairs hit a CO it's concentrated into your carrier lines: DS1-3 and piped either via a single copper/fiber/RF channel (usually copper but the new ones are all fiber I believe).

24 phone lines can be carried in a single T1 line. That's were you get your 56k from. The DS1 spec is (simple form): 193 bits per frame, 1 for frame sync, the other 192 for 24 8-bit channels. 8000 frames per second gives you 8000*193 or 1544000 bits/sec, with 1536000 bits/sec actual useable data.

Now about those 24 8-bit channels. They need to be encoded because you want to maintain a net 0 voltage on the line. so too many 1s or too many 0's gets you into trouble, not to mention confuses the clock recovery circuitry. Enter AMI. Alternate Mark Inversion just reversed the polarity of a '1' every other time it occurred. This still didn't solve the problem of too many 0s though, but with the 8kHz PCM modulated voice data it wasn't a problem because any kind of little noise would send an LSB to the '1' state and keep the clock recovery happy.

(I'm a little hazy here, and I can't look up my old /. comment (about 6 months ago) that described this in eerily gory details, if someone can tell me how to search OLD OLD COMMENTS (not just stories) email me PLEASE!) Somewhere in the past B8ZS (Bit 8 Zero Set) came along and, with AMI, cleaned up everything. Basically if too many zeros were sent a '1' was injected with the same polarity as the last '1'. The clock recovery circuitry didn't care, but the data recovery circuit would notice that the mark had the same polarity and change it to a zero to keep everything 'right'.

Anyway. To be able to put 24 phone conversations and have them all be kept track of properly, the telcos decided to band 12 of these frames together (12 frames, each with 24 channels) and use the LSB of each conversation for switching information. I believe it was the 6th and 12th frames of the Super Frame (SF) which carried this info, call them bit A and B.

Bit A and B were the line status bits. They told the telco equipment whether each line was on-hook, off-hook, busy or ringing. Now with 8kHz PCM conversations going on, losing a bit 1/6 of the time meant nothing important and it was hardly missed. For data conversations though that reduced your 24 64k channels to 24 56k channels since the end modems have no idea when that bit will be lost. There's your limit to analog modems. ISDN gives you a 64k channel because the equipment doesn't screw with the data frames every 6th frame.

Anyway the SF format was put away and the ExtendedSupaFormat (ESF) method was brought in. All it is is instead of using 12 193-bit frames, it uses 24 and continues the "I need the LSB of every channel every 6th frame" to give the telco 4 line status bits, A, B, C and D. The other two bits aren't used yet but it's for future expansion. They typically just mirror the A and B bits for now.

T1 is actually both a Layer 1 and a Layer 2 spec. It originated from the old Bellcore D1 spec in the 60s and details both the actual line voltages and the framing of data to get to those line voltages. There are no such things as T2s and T3s, only DS2s and DS3s.

quick summary:

• DS0 - 64k channel
• DS1 - (T1 data spec) - 24 DS0s with framing
• DS2 - 4 DS1s with framing and filling
• DS3 - 7 DS2s with framing and filling

Now the DS2 framing gets REALLY weird because the DS1 lines feeding them are not likely to be in sync so there are variable fill rates and such to maintain an intermediate bit rate which isn't really standardized... it sits between 1.544mbps and 1.56mbps I believe and is tuned at each installation since the DS1s feeding the DS2s are more or less unique. Simiarly, the DS2s feeding the DS3 aren't necessarily in sync the DS3 too has an intermediate rate.

Anyway the whole point of this post is to show that once you hit a CO you are no longer on your own copper pair. You are part of the collective. You are just a 64k channel (with an involuntary loss of 1 bit, thus dropping you to a 56k channel) in a sea of millions and millions of bits. And the fact that it all works pretty much flawlessly is a tribute to the Bellcore engineers.

And no, I'm not a Bellcore engineer. :-)

Nominations For Geek Of The Year?

Are they taking nominations yet? I'm not saying that this guy would or even should win, but he definitely should be nominated. Let me outline why (gives me a chance to use that fancy <ol> sign, too)

1. Dedicated to his high-speed access
   Not many people would go through the lengths to get DSL that this guy did, especially considering his telco did tell them that they were planning on rolling out DSL in the future (just not the near-enough future).
   
2. Dedicated to perfection
   Most people would've stopped after getting the first connection and been like, "Whoop! I rule!" but this guy actually tested out more equipment because, well, damn, that first DSL connection wasn't good enough.
   
3. Donated efforts to local groups
   Granted, it's part of his work, but he took his understanding to the public and got them up and running as well. A geek is not selfish, nor is he greedy.
   
4. Published
   This is most important. A true geek feels the need to let everyone know not just that he did it, but how he did it, in the sort of detail that allows it to be repeated by one and all.

Kudos to this dude. He gets a nod for a nomination for geek of the year. Is there a beanie award for this? If not, there should be.

Some cautions in doing it yourself

I've been using an SDSL connection between my office and house for the past two months. Paid $800 for a pair of Flowpoint 2200 routers, and BA takes $30/mo for the copper loop ($100 install).

There are a few issues with doing it yourself. If there are are problems, you don't have an ISP to complain to. It can be difficult to track down if a problem is hardware related, in the copper loop, or in your own inside wiring. For the past two weeks, I had major problems (connection would develop massive packet loss and drop out at times), which I eventually figured out were the routers; just got them replaced yesterday after spending hours trying various possible fixes from Flowpoint. There are still some minor problems (had the routers randomly drop the link and reset on me once earlier tonight), so my next step is to redo all the inside wiring on both ends.

As far as hardware recommendations go, despite the problems I would still recommend the Flowpoints. They can run in routed or bridged mode and you can do fun stuff like set up multiple ATM PVCs with traffic shaping, so one user's traffic doesn't disturb another's. (Here's another issue; with the line saturated, the latency jumps from 3ms to 140ms; I set up one PVC for my roommates and one for myself, and set them both to 95% traffic max to get around this...)

At about 8500 feet, my SDSL runs at 1744kbit/sec; the max the Flowpoints will do is 2320 at shorter distances. Another option I looked at was the Netopia R7171; it wasn't out at the time (it should be now), but it'll do two bonded 1.5mbit/sec channels over a 2 pair loop, as opposed to a single pair for normal DSL.

So basically, if you're not prepared to have to fix problems yourself, go with a commercial DSL company. Otherwise, if you have somewhere to stick the other end (office or whatever), this is a pretty cheap alternative to going through an ISP; I doubt I could find an ISP to give me a large IP block and controllable revdns for so cheap.

New? And a caveat....

We've been doing this at my company for quite some time now, just using different equipment.

One of the main points to consider with this scheme is that the telco may be iffy on the use of an alarm pair for this use, especially if they offer their own DSL/Internet offering. We've been lucky so far, but we've heard reports of others trying this, and then getting shut down...

Just for your info, we've used (with great success) the Expresso units from Tut Systems [tutsys.com] and the XL-12000 series (for the longer haul stuff), also from Tut. BTW- the Tut 12000 units do a full 2mbit. ;-)

Heh

I wonder how long it's gonna be before the Slashdot Effect takes down the DSL line he worked so hard to wire up? :-)

Re:T1...

FWIW, I believe you can also slap a CSU/DSU and router on each end and have a "T1".

Yup, you sure can -- up to a limit of around 3500 feet, on 24 awg wire (at least, this has been my experience). If you use a heavier gauge of wire it will go further.

Of course, DSL routers are much less expensive than a Cisco 1600 and CSU/DSU.

True, but a new 1600 is overkill for something like a homebrew project anyway.

You can get used T1 CSU/DSUs (no frills units) for about $200 on eBay often, and then get a pair of used old Cisco routers (like a CGS/MGS or even an IGS) for between $100-150 each. Total cost around $600-800, unless you happen to have some old T1 capable routers sitting around (you'd be surprised at the sort of equipment that is being pulled out of service and scraped these days).

Yes, the older routers are not as fast, etc. etc. but for a simple end-point setup with a small number of users, even the 10 year old gear can keep up without any problems.

If one was right near a telco CO, you could run a small/cheap/hobby limited range ISP operation this way -- pick up an old AGS+ with scads of serial and ethernet ports for maybe $400 for the central site. They can still pump a respectable amount of data -- the AGS+ was often used as the network core router for entire campus or corporate networks "back in the day" (like with 12 ethernet ports, 8 T1s, and two FDDI ports).

The GNU way

Do it yourself! It's the GNU way!

Funny, I thought the GNU way was relying on people much smarter and harder working than oneself to produce software, without bothering to contribute working on open source projects of their own. Oh I forgot that I also thought the GNU way meant that you had to complain and whine about software that other people write and you don't pay for.