I use an Epia V system as my home mailserver/Easynews download box, works great. I put 512M of RAM in it and installed Trustix 2.1 with a custom initrd so that it loads the system from a tarfile into a tmpfs ramdisk.
Only time the drive spins up is when I'm downloading movies or the wife is playing MP3 audio via the Turtle Beach Audiotron.
External power supply and no CPU fan = silent system in the family room about 98% of the time, and music most likely playing when it isn't. Love it!
Anybody know anything about mid-to-early-80s Volkswagen vans and magnesium? I remember when I was a kid and one caught fire down the street, and something on it (underneath in the middle, as I recall) burned so damned white-hot I couldn't even look at it.
The firefighters who responded sat there dousing it with water for like an hour. It was wild.
I don't think magnesium in vehicles is such a hot idea. Or maybe it is!
Well, other than the fact that it was a very high-pressure job and I sorta did kinda enjoy it at the time....
I had a job in the early 90's programming IVR applications (i.e. call an 800 or 900 number and interact with a computer via touchtones). Most of these apps were the front-end interface for the psychic lines you see advertised on late-night TV. You call in and either choose to try and talk with either a specific 'psychic' or a random one. Either way, the service bureau's system calls a psychic working out of their house on the back side and connects the caller with the 'psychic'.
Now, the app language I used was called CLASS, and it allowed for a whole whopping 99 variables ($00 through $99, the first 30 of which were reserved) and looked a lot like a bastardized cross between BASIC and assembly language. Something like this:
start:
say.: c9100 h,b,a (Speak a voice prompt in the file 'c9100)
wait start: $46 (wait for a touchtone, if none go back to start)
if $46 = 0 presszero:
if $46 = 1 pressone:
goto start:
pressone:
You get the idea. Evil stuff.
So I'm writing apps in this crap all day. Not exactly the most maintainable code in the world, let me tell you.
And then there are lots of fun things like up-front limiting. This means that there are tons of freaking losers in the world who will gladly grind their fingers to the bone punching buttons to talk to a 'psychic' and ringing up $5000 phone bills every month, but then charging them back when they get their phone bill. ("It wasn't me! Somebody snuck in and spent 8 hours straight on my phone!") So you have to make sure and limit the amount they can use per month.
Also, I had one client whose 'psychic' pool were either "your personal angel" or "your salem witch" depending on which 900 number you called. We had to make damn sure and play a tone to the 'psychic' to tell them which one they were supposed to be for that call or hilarity would ensue, let me tell you!
Combine all this with the fact that the company I worked for was pretty much run by the clients, and you have a pretty sucky high-pressure job writing in a crappy language.
I'll be interested to see when (or rather, if) MSNBC reports this on their site - they've been quite good at reporting on the case lately, since SCO has been filing lawsuits, but now that the shoe is on the other foot, we'll see who was behind the reporting.
Seems to me they've been quite quick with the last couple of stories, but nothing about this one yet.
No, but I did get a $10 gift cert to ThinkGeek from a buddy at work (two Linux enthusiasts stuck in a Microsoft Certified Partner company) and it's going towards one...
Very much so. I'm a consultant and have actually spent quite a lot of time lately telling customers why their centralized Exchange deployments with Outlook 2000 suck *ss - it's the client, stupid!
Outlook 2000 massively blows, Outlook XP is a bit better but pops up annoying dialogs when the network gets slow, and Outlook 2003 finally has it right - it's the old "third time's a charm" cycle from MS rearing its ugly head again.
Outlook 2003 introduces a new semi-connected mode called "Cached mode" that caches messages locally and works great. It also supports (in conjunction with Exchange 2003 only, there's the rub!) a new remote transport, RPC over HTTP, that is frankly pretty amazingly cool and lets me run the full client remotely with no VPN, no hangs, and decent feedback as to what's going on. What a concept!
I'm sounding like a cheerleader here, and I'm not, but I do have to say that Exchange 2003 and Outlook 2003 are pretty much perfect poster children for the "third time's a charm" syndrome from Microsoft. They've finally gotten some of the problems through their thick skulls and, if not outright fixed them, at least started nicely down that road.
That's application level. You can shut it off. And if there comes a time when you can't, you're free to switch to a different browser, like, say, Opera.
And it doesn't result in mistakenly passed spam checks, email address leaks to Veri$ign, and general screwed-upedness like a wildcard DNS does.
Geez, does anybody get that "the web" is not a synonym for "the Internet" anymore?
Actually, I somewhat misspoke. It's worse than it appears, and the problem is sendmail, not fetchmail.
Basically, ANYONE who's running sendmail, most likely any sendmail, but definitely on RedHat 8.0, and has a bogus domain name configured on their server, is going to have problems with local mail delivery.
Say I have a server that I've configured with a local domain name of blarg.com, which doesn't exist. When someone on a shell account types "mail joeschmoe", the sendmail that gets started up doesn't deliver mail straight to a file like Sendmail did before the split into submission and delivery daemons.
Instead, it connects over port 25 to the host specified by MTAHost in submit.cf. By default, at least on RedHat 8.0, that setting is "[localhost]".
But guess what? Sendmail tacks on the domain name. And does DNS resolution before host table resolution, even if nsswitch.conf is set to check/etc/hosts first, because Sendmail does its own thing.
End result? You log into a shell, type "mail joeschmoe". The mail program then uses Sendmail as its delivery agent, which then connects to Verisign's mail plonker. No delivery.
The only solution I see is to set the MTAHost setting in submit.cf (I'm too lazy to figure out how to do it in submit.mc) to "[127.0.0.1]".
I run a home Linux machine using fetchmail to pull from my cable provider's POP server.
Today I logged in to check on mail processing and noticed that it wasn't delivering messages. Turns out fetchmail was connecting to "localhost.mydomain.com" which is in the box's host table as localhost.
But guess what? I have DNS first in the search order, and it was getting a response. That response led to a mailserver at Verisign that refuses all mail.
I think we should consider a class-action lawsuit. How much more stuff is breaking silently right now thanks to Verisign's invalid responses?
If Broadcom is compiling the code themselves on Linksys' behalf, then it wouldn't be a violation. However, if they're making a developer's kit available to Linksys, then they most certainly are in violation of the GPL and must release the source.
I do agree with the first point, though - I haven't looked at the Zebra source, but if the only changes made were to config files, then there are no source mods to release. If they changed a #define in a source config file, though, even that should be reflected in their distribution.
It's already happened - Windows Server 2003, Web Server Edition. MS was getting their butt kicked nine ways to Sunday in the hosting space, and that's the result. 400 bucks!
I wonder if the time zone difference might be seen as an advantage, i.e., as a way to have skilled, white-collar employees working on a problem 24/7 without having to pay them a premium for working overnight?
Hell yeah!
I work for a small (~120 employees, two divisions - software development and infrastructure) and we have an office in Pune, India to do software dev.
The software people here talk to the client all day, have meetings, and write specs. Then you know what they do at the end of the day?
They send an email with the specs to the guys in India. Then they go home for dinner, hang out with their families, and crash.
When they come in the next morning, they have an email from the coder slaves (sorry, I mean, "folks at the India office") that has the code. Done.
They spend the day demoing it to the client, having meetings, firming up the spec, and the cycle repeats. But only for about half as long as it would if the product were being developed here. And for 25% of the price.
Our India office wasn't very utilized when it first started up. Now I think the utilization in the first quarter of this year was above 90%.
That is a big reason this is so attractive, and India in particular.
Get involved in infrastructure work - all the suits left here aren't going to be communicating with the slave labor over there via the US Mail - they'll still need computing infrastructure here, like Internet access, email, and decision support systems.
I would say that you should look at computer security, but over the years I've tried to train a lot of people in it (my field) and I've become convinced that it just takes a certain kind of person to do it - you're pretty much always interested in it, or you're not and can't get that way by training.
I've got some karma to spend, so I'll say it - a certain amount of this will be good for the industry as a whole. A lot of the people getting weeded out by this outsourcing are the ones who took their classes to become a developer and "make the big bucks". Over time we'll realize that IT in the US will be left with the people who think up the cool stuff to do and leave it to the overseas grunts to actually execute.
That said, I also want to say that I'll be the first to laugh when one of these countries (probably not India, though) becomes the next Iraq and some US companies get put in a serious bind.
As for those saying that this will redistribute the wealth globally - get real! We're creating a new overseas worker class, not new overseas companies. The jobs are going there because the people will work for peanuts, not because they have great skills and deserve huge salaries.
Or you might want to consider an exercise that doesn't subject your joints to several G's of force at every step, like cycling.
And if you really want to have a good time, try one of these - they're mail-order only, but check out local bike shops or their message board to see if you can arrange a test drive on one or something like it.
I test-drove one, and though I'm married with kids and can't afford one right now, I can assure you I know where part of my tax refund is going next year.
There may be more possible addresses than atoms, but definitely not routable networks.
A big part of a reason that we're in the current situation we're in with IPv4 is that DARPA used to love to hand out Class A blocks like candy to anybody that asked.
Methinks that should provide you with a bit of a clue as to the inaccuracy of your statement.
They're available on some Pontiacs, at least - my boss's Grand Prix has a HUD that projects the speed on the windshield in front of the steering wheel.
I believe some Cadillacs also have HUDs for the night-vision system that's available.
Slashdot Mirror
Hey, my official title is "Solution Architect", you insensitive clod!
Which means I'm a consultant who has to have a cool-sounding title to command a higher hourly rate.
idot.com has them.
I use an Epia V system as my home mailserver/Easynews download box, works great. I put 512M of RAM in it and installed Trustix 2.1 with a custom initrd so that it loads the system from a tarfile into a tmpfs ramdisk.
Only time the drive spins up is when I'm downloading movies or the wife is playing MP3 audio via the Turtle Beach Audiotron.
External power supply and no CPU fan = silent system in the family room about 98% of the time, and music most likely playing when it isn't. Love it!
How about a script to tunnel in with SSH and then fake the host table? I believe all the Connector does is talk to port 80 on the Exchange server..
Anybody know anything about mid-to-early-80s Volkswagen vans and magnesium? I remember when I was a kid and one caught fire down the street, and something on it (underneath in the middle, as I recall) burned so damned white-hot I couldn't even look at it.
The firefighters who responded sat there dousing it with water for like an hour. It was wild.
I don't think magnesium in vehicles is such a hot idea. Or maybe it is!
Well, other than the fact that it was a very high-pressure job and I sorta did kinda enjoy it at the time....
.: c9100 h,b,a (Speak a voice prompt in the file 'c9100)
I had a job in the early 90's programming IVR applications (i.e. call an 800 or 900 number and interact with a computer via touchtones). Most of these apps were the front-end interface for the psychic lines you see advertised on late-night TV. You call in and either choose to try and talk with either a specific 'psychic' or a random one. Either way, the service bureau's system calls a psychic working out of their house on the back side and connects the caller with the 'psychic'.
Now, the app language I used was called CLASS, and it allowed for a whole whopping 99 variables ($00 through $99, the first 30 of which were reserved) and looked a lot like a bastardized cross between BASIC and assembly language. Something like this:
start:
say
wait start: $46 (wait for a touchtone, if none go back to start)
if $46 = 0 presszero:
if $46 = 1 pressone:
goto start:
pressone:
You get the idea. Evil stuff.
So I'm writing apps in this crap all day. Not exactly the most maintainable code in the world, let me tell you.
And then there are lots of fun things like up-front limiting. This means that there are tons of freaking losers in the world who will gladly grind their fingers to the bone punching buttons to talk to a 'psychic' and ringing up $5000 phone bills every month, but then charging them back when they get their phone bill. ("It wasn't me! Somebody snuck in and spent 8 hours straight on my phone!") So you have to make sure and limit the amount they can use per month.
Also, I had one client whose 'psychic' pool were either "your personal angel" or "your salem witch" depending on which 900 number you called. We had to make damn sure and play a tone to the 'psychic' to tell them which one they were supposed to be for that call or hilarity would ensue, let me tell you!
Combine all this with the fact that the company I worked for was pretty much run by the clients, and you have a pretty sucky high-pressure job writing in a crappy language.
I'll be interested to see when (or rather, if) MSNBC reports this on their site - they've been quite good at reporting on the case lately, since SCO has been filing lawsuits, but now that the shoe is on the other foot, we'll see who was behind the reporting.
Seems to me they've been quite quick with the last couple of stories, but nothing about this one yet.
No, but I did get a $10 gift cert to ThinkGeek from a buddy at work (two Linux enthusiasts stuck in a Microsoft Certified Partner company) and it's going towards one...
Very much so. I'm a consultant and have actually spent quite a lot of time lately telling customers why their centralized Exchange deployments with Outlook 2000 suck *ss - it's the client, stupid!
Outlook 2000 massively blows, Outlook XP is a bit better but pops up annoying dialogs when the network gets slow, and Outlook 2003 finally has it right - it's the old "third time's a charm" cycle from MS rearing its ugly head again.
Outlook 2003 introduces a new semi-connected mode called "Cached mode" that caches messages locally and works great. It also supports (in conjunction with Exchange 2003 only, there's the rub!) a new remote transport, RPC over HTTP, that is frankly pretty amazingly cool and lets me run the full client remotely with no VPN, no hangs, and decent feedback as to what's going on. What a concept!
I'm sounding like a cheerleader here, and I'm not, but I do have to say that Exchange 2003 and Outlook 2003 are pretty much perfect poster children for the "third time's a charm" syndrome from Microsoft. They've finally gotten some of the problems through their thick skulls and, if not outright fixed them, at least started nicely down that road.
Security concerns notwithstanding, I'm not sure I'd want the attention that comes these days with having a port 135 RPC server out on the net.
As someone who runs several IDS systems, I can tell you that there's active scanning for these systems going on, and it isn't from worms.
I'd put a VPN in front of it if I were you.
Okay, one more time...
That's application level. You can shut it off. And if there comes a time when you can't, you're free to switch to a different browser, like, say, Opera.
And it doesn't result in mistakenly passed spam checks, email address leaks to Veri$ign, and general screwed-upedness like a wildcard DNS does.
Geez, does anybody get that "the web" is not a synonym for "the Internet" anymore?
Woops! Time for a new subject, this one's done!
Actually, I somewhat misspoke. It's worse than it appears, and the problem is sendmail, not fetchmail.
/etc/hosts first, because Sendmail does its own thing.
Basically, ANYONE who's running sendmail, most likely any sendmail, but definitely on RedHat 8.0, and has a bogus domain name configured on their server, is going to have problems with local mail delivery.
Say I have a server that I've configured with a local domain name of blarg.com, which doesn't exist. When someone on a shell account types "mail joeschmoe", the sendmail that gets started up doesn't deliver mail straight to a file like Sendmail did before the split into submission and delivery daemons.
Instead, it connects over port 25 to the host specified by MTAHost in submit.cf. By default, at least on RedHat 8.0, that setting is "[localhost]".
But guess what? Sendmail tacks on the domain name. And does DNS resolution before host table resolution, even if nsswitch.conf is set to check
End result? You log into a shell, type "mail joeschmoe". The mail program then uses Sendmail as its delivery agent, which then connects to Verisign's mail plonker. No delivery.
The only solution I see is to set the MTAHost setting in submit.cf (I'm too lazy to figure out how to do it in submit.mc) to "[127.0.0.1]".
Well, isn't this just great!
I run a home Linux machine using fetchmail to pull from my cable provider's POP server.
Today I logged in to check on mail processing and noticed that it wasn't delivering messages. Turns out fetchmail was connecting to "localhost.mydomain.com" which is in the box's host table as localhost.
But guess what? I have DNS first in the search order, and it was getting a response. That response led to a mailserver at Verisign that refuses all mail.
I think we should consider a class-action lawsuit. How much more stuff is breaking silently right now thanks to Verisign's invalid responses?
Bill Joy's not dead, dude!
Oh, you read it through? You're ahead of most Slashdotters...
If Broadcom is compiling the code themselves on Linksys' behalf, then it wouldn't be a violation. However, if they're making a developer's kit available to Linksys, then they most certainly are in violation of the GPL and must release the source.
I do agree with the first point, though - I haven't looked at the Zebra source, but if the only changes made were to config files, then there are no source mods to release. If they changed a #define in a source config file, though, even that should be reflected in their distribution.
It's already happened - Windows Server 2003, Web Server Edition. MS was getting their butt kicked nine ways to Sunday in the hosting space, and that's the result. 400 bucks!
I wonder if the time zone difference might be seen as an advantage, i.e., as a way to have skilled, white-collar employees working on a problem 24/7 without having to pay them a premium for working overnight?
Hell yeah!
I work for a small (~120 employees, two divisions - software development and infrastructure) and we have an office in Pune, India to do software dev.
The software people here talk to the client all day, have meetings, and write specs. Then you know what they do at the end of the day?
They send an email with the specs to the guys in India. Then they go home for dinner, hang out with their families, and crash.
When they come in the next morning, they have an email from the coder slaves (sorry, I mean, "folks at the India office") that has the code. Done.
They spend the day demoing it to the client, having meetings, firming up the spec, and the cycle repeats. But only for about half as long as it would if the product were being developed here. And for 25% of the price.
Our India office wasn't very utilized when it first started up. Now I think the utilization in the first quarter of this year was above 90%.
That is a big reason this is so attractive, and India in particular.
Easy. Don't be a developer.
Get involved in infrastructure work - all the suits left here aren't going to be communicating with the slave labor over there via the US Mail - they'll still need computing infrastructure here, like Internet access, email, and decision support systems.
I would say that you should look at computer security, but over the years I've tried to train a lot of people in it (my field) and I've become convinced that it just takes a certain kind of person to do it - you're pretty much always interested in it, or you're not and can't get that way by training.
I've got some karma to spend, so I'll say it - a certain amount of this will be good for the industry as a whole. A lot of the people getting weeded out by this outsourcing are the ones who took their classes to become a developer and "make the big bucks". Over time we'll realize that IT in the US will be left with the people who think up the cool stuff to do and leave it to the overseas grunts to actually execute.
That said, I also want to say that I'll be the first to laugh when one of these countries (probably not India, though) becomes the next Iraq and some US companies get put in a serious bind.
As for those saying that this will redistribute the wealth globally - get real! We're creating a new overseas worker class, not new overseas companies. The jobs are going there because the people will work for peanuts, not because they have great skills and deserve huge salaries.
Or you might want to consider an exercise that doesn't subject your joints to several G's of force at every step, like cycling.
And if you really want to have a good time, try one of these - they're mail-order only, but check out local bike shops or their message board to see if you can arrange a test drive on one or something like it.
I test-drove one, and though I'm married with kids and can't afford one right now, I can assure you I know where part of my tax refund is going next year.
Cycling - try it out, you'll be glad you did.
It's one more reason that I'm wanting to get a job at Microsoft... (down the list, but still!)
Who wants to hit 'im first?
There may be more possible addresses than atoms, but definitely not routable networks.
A big part of a reason that we're in the current situation we're in with IPv4 is that DARPA used to love to hand out Class A blocks like candy to anybody that asked.
Methinks that should provide you with a bit of a clue as to the inaccuracy of your statement.
They're available on some Pontiacs, at least - my boss's Grand Prix has a HUD that projects the speed on the windshield in front of the steering wheel.
I believe some Cadillacs also have HUDs for the night-vision system that's available.
I could use some proofing!
Imagine a Beowulf cluster....