Retarded programmers

[Rococo Modem Basilisk]

Cavehamster, five bucks says the original programmer doesn't know how it works either.

The Enki school of programming: write the thing off the top of your head in an all-nighter without checking the reference manual. In a week, try to compile it and start debugging. By the end of the month, you should have something that you can understand -- or something that looks like the output of /dev/urandom.

[cavehamster]

Cavehamster, five bucks says the original programmer doesn't know how it works either.

The Enki school of programming: write the thing off the top of your head in an all-nighter without checking the reference manual. In a week, try to compile it and start debugging. By the end of the month, you should have something that you can understand -- or something that looks like the output of /dev/urandom.

Mostly, I have been working on integrating some systems.  I plan on having data reside on a iSCSI server.  I've setup the server, the client, setup redundant links, setup multipathing, and everything is humming along.

However, what no one posts about, is what happens when all the redundant links fail.  I lost a day testing and verifying until I resolved what happens and how to handle it gracefully.

And then I discovered the damn boot scripts try to connect to the iSCSI server everytime a network interface comes up, and it's not until the last one comes up that it actually connects, so... 7 minute wait time to get the system ready.  I finally poked around long enough to resolve that.

Only to hit another issue... and another... and... gah.  I have the subsystem up and running now, and am getting closer to actually getting work done, but it still irks me that it's been this annoying.

[Rococo Modem Basilisk]

Cavehamster, five bucks says the original programmer doesn't know how it works either.

The Enki school of programming: write the thing off the top of your head in an all-nighter without checking the reference manual. In a week, try to compile it and start debugging. By the end of the month, you should have something that you can understand -- or something that looks like the output of /dev/urandom.

Mostly, I have been working on integrating some systems.  I plan on having data reside on a iSCSI server.  I've setup the server, the client, setup redundant links, setup multipathing, and everything is humming along.

However, what no one posts about, is what happens when all the redundant links fail.  I lost a day testing and verifying until I resolved what happens and how to handle it gracefully.

And then I discovered the damn boot scripts try to connect to the iSCSI server everytime a network interface comes up, and it's not until the last one comes up that it actually connects, so... 7 minute wait time to get the system ready.  I finally poked around long enough to resolve that.

Only to hit another issue... and another... and... gah.  I have the subsystem up and running now, and am getting closer to actually getting work done, but it still irks me that it's been this annoying.

Have you gotten the debugging boner yet? Because, if this doesn't do it, you might want to try something a bit harder -- like OS dev, or writing a C compiler without using BNF.

[cavehamster]

Cavehamster, five bucks says the original programmer doesn't know how it works either.

The Enki school of programming: write the thing off the top of your head in an all-nighter without checking the reference manual. In a week, try to compile it and start debugging. By the end of the month, you should have something that you can understand -- or something that looks like the output of /dev/urandom.

Mostly, I have been working on integrating some systems.  I plan on having data reside on a iSCSI server.  I've setup the server, the client, setup redundant links, setup multipathing, and everything is humming along.

However, what no one posts about, is what happens when all the redundant links fail.  I lost a day testing and verifying until I resolved what happens and how to handle it gracefully.

And then I discovered the damn boot scripts try to connect to the iSCSI server everytime a network interface comes up, and it's not until the last one comes up that it actually connects, so... 7 minute wait time to get the system ready.  I finally poked around long enough to resolve that.

Only to hit another issue... and another... and... gah.  I have the subsystem up and running now, and am getting closer to actually getting work done, but it still irks me that it's been this annoying.

Have you gotten the debugging boner yet? Because, if this doesn't do it, you might want to try something a bit harder -- like OS dev, or writing a C compiler without using BNF.

I develop for microcontrollers a lot.  That does it for me, debugging on a device that might have a LED connected to tell you if things are working right or not...

[Rococo Modem Basilisk]

Ooh, so you eschew emulators and binary debuggers like a real man? Nice 

[Bebek Sincap Ratatosk]

I'm sure this is pretty close to the truth.

Trucrypt has come up with a solution for this which involves hidden encrypted partitions on the disk (the primary one protected by Password A, the secret one hidden by Password B). So you give the attacker password A and it unlocks the drive with the "data"... which is not the same as the data you actually care about (hidden behind B). Since there's no way to verify or determine if there is one or two partitions, the arttacker will not know that you have a second password.

Of course, I never use this and if anyone beats me with a wrench I will give them the right password... I promise....

[Rococo Modem Basilisk]

Mmm, wrenches... 

Well, personal preferences aside, there are far more irritating ways to be fucked over by undocumented code. I was the first person to write a working OS in D (the digitalmars D, not the various other languages with that name). It required excising about half the standard library, and rewriting 3/4ths of the remainder from scratch. That was about 80% of the project (which, by the way, had a file system, a curseslike gui, support for ram disks/hard drives/hard drive partitions, and its own executable format).

[cavehamster]

Mmm, wrenches... 

Well, personal preferences aside, there are far more irritating ways to be fucked over by undocumented code. I was the first person to write a working OS in D (the digitalmars D, not the various other languages with that name). It required excising about half the standard library, and rewriting 3/4ths of the remainder from scratch. That was about 80% of the project (which, by the way, had a file system, a curseslike gui, support for ram disks/hard drives/hard drive partitions, and its own executable format).

You know you are having a bad day when you start hitting compiler bugs. I had o e a few weeks ago, turned out to be a lighty fried CPU that would segfault when compiling, but not when running the OS, go figure.

[NotPublished]

Alternatively, hire a bunch of Asian programmers.

[Rococo Modem Basilisk]

Alternatively, hire a bunch of Asian programmers.

Or use rentacoder. It's the same thing, except that they are mostly Indian and it's cheaper.

[cavehamster]

Alternatively, hire a bunch of Asian programmers.

Or use rentacoder. It's the same thing, except that they are mostly Indian and it's cheaper.

When I was freelancing, I looked at rentacoder.  Holy cow, how does anyone make any money at all as a coder there?

I've still yet to see any quality code coming out of India.  I'm sure those guys are just as bright, but the constant hammer of doing it cheaper, faster is not helping anything.  It seems like to me if you send a project to India, you can be expected to take twice the time to fix it when it comes back.  Kind of like if you send stuff to be built in China, you can expect them to go ahead and build a run of knock-offs while they are at it to sell on the black market.

[Requia ☣]

Out of curiosity, just how much was an Indian rentacoder?

[cavehamster]

Out of curiosity, just how much was an Indian rentacoder?

It's been so long I don't have numbers, but over and over I saw jobs that would take me a solid 20-30 hours of work, someone would be bidding less than $100 for.

[NotPublished]

ugh, my boss hired some programmers in China to make a Crawler, worst choice ever; it was definantly something - and it crawled. He didn't want to burden me though, I thought it was nice.

[Requia ☣]

Out of curiosity, just how much was an Indian rentacoder?

It's been so long I don't have numbers, but over and over I saw jobs that would take me a solid 20-30 hours of work, someone would be bidding less than $100 for.

Hrm, I suppose it depends on how much of that they get, and how hard it is to be full time.  Median wage in India is only 1200-1600 (depending on city) so 3-4 dollars an hour would be pretty significant.