Random Tech (or IT Work!) Fail

[Purrrfect]

This is alas pretty common in my experience. And if they don't burn shitloads of cash on consulting services _before_ they go ask IT why their stupidity isn't working, you're amongst the lucky ones.
And it is _always_ ITs fault.

One time a few years ago at our old building, we had one of the HR people poke their head into our dept and ask why the water fountain in the production area around the corner was leaking.

[zoobernut]

This is alas pretty common in my experience. And if they don't burn shitloads of cash on consulting services _before_ they go ask IT why their stupidity isn't working, you're amongst the lucky ones.
And it is _always_ ITs fault.

Oh don't get me wrong that is happening too. We have wasted so much money on consultants that were not needed it is ridiculous. If you hire and pay for people with a certain skill why not use those people instead of paying someone else for the same work 100X more. Don't forget the individuals who decide to "quit" work give notice, leave as employees to come back and do the same job for the company at 10x the pay as an outside consultant.

[Yabby]

Don't let people know about that stuff...I do consulting on the side and it makes more money than my main job. Half the time they pay me more to tell them what they should do, rather than just pay me to do it for them. It is the greatest thing in the world.

[Guyverarmor]

Recently at my job, we signed SLA agreements for Carrier/Dark Fiber services for Cell Tower customers...without checking the actual latency between that location and the serving MSC. Then said people who got stuff signed get snappy with my team (Metro Core Engineering) when we aren't meeting SLA....

[Yuri-G]

grrrr


spent a while working on this assignment before I took a closer look and realized it's not my code that's wrong... it's the algorithm. we were supposed to do this in clojure:

Spoiler: show

Now here is an algorithm for computing √x with an accuracy of ε, for a real number x ≥ 0.0.

1. Initialize. Set G = x and H = 1.0.
2. Test. If |G − H| < ε then return G.
3. Update. Set G = (G + H) / 2.0 and H = x / G. Go to 2.

For example, if ε = 0.000001, then the algorithm says that √1.0 ≈ 1.0, √2.0 ≈ 1.41421, √3.0 ≈ 1.73205, and √4.0 ≈ 2.0.

try that with 3.0:
x = 3.0, g = 3.0, h = 1.0. |g - h| = 2.0, so recurse
x = 3.0, g = 2.0, h = 1.0. |g - h| = 1.0, so recurse
x = 3.0, g = 1.5, h = 1.5. |g - h| = 0.0 so clearly the square root of 3 is 1.5!!! ...or not


it works with literally any other number. 2.9999? works. 3.00001? works. 3 appears to be a special case. IDK why that bugs me so much, but it really does...

[orinthia]

Spoiler: show

grrrr


spent a while working on this assignment before I took a closer look and realized it's not my code that's wrong... it's the algorithm. we were supposed to do this in clojure:

Spoiler: show

Now here is an algorithm for computing √x with an accuracy of ε, for a real number x ≥ 0.0.

1. Initialize. Set G = x and H = 1.0.
2. Test. If |G − H| < ε then return G.
3. Update. Set G = (G + H) / 2.0 and H = x / G. Go to 2.

For example, if ε = 0.000001, then the algorithm says that √1.0 ≈ 1.0, √2.0 ≈ 1.41421, √3.0 ≈ 1.73205, and √4.0 ≈ 2.0.

try that with 3.0:
x = 3.0, g = 3.0, h = 1.0. |g - h| = 2.0, so recurse
x = 3.0, g = 2.0, h = 1.0. |g - h| = 1.0, so recurse
x = 3.0, g = 1.5, h = 1.5. |g - h| = 0.0 so clearly the square root of 3 is 1.5!!! ...or not


it works with literally any other number. 2.9999? works. 3.00001? works. 3 appears to be a special case. IDK why that bugs me so much, but it really does...

Walking through the update step, your new H value is based on the G update step being done, so you should have a new H value for the first loop. It should look like:

starting values: x = 3; g = x; h = 1.0, p(precision) = 0.000001.

x = 3, g = 3.0, h = 1.0. |g - h| = 2.0. 2.0 >= p? true, so continue:
update: g = (3+1.0)/2 = 2.0; h = 3/2.0(the new g value) = 1.5; run again.

x = 3, g = 2.0, h = 1.5. |g - h| = 0.5. 0.5 >= p? true, so continue:
update: g = (2.0+1.5)/2 = 1.75; h = 3/1.75 = 1.714285714285714; run again.

x = 3, g = 1.75, h = 1.714285714285714. etc, you get the idea


Code walkthroughs and writing problem-notes are important. Writing it out in plain-english, math, or pseudo-code is important too so you can see where you might have skipped a step or not did the algorithm right. Think about the task, plan it, and then see how the programming language can do it. And it'll be easier/faster to ask the prof on your problems.

[Yuri-G]

Walking through the update step, your new H value is based on the G update step being done, so you should have a new H value for the first loop. It should look like:

starting values: x = 3; g = x; h = 1.0, p(precision) = 0.000001.

x = 3, g = 3.0, h = 1.0. |g - h| = 2.0. 2.0 >= p? true, so continue:
update: g = (3+1.0)/2 = 2.0; h = 3/2.0(the new g value) = 1.5; run again.

x = 3, g = 2.0, h = 1.5. |g - h| = 0.5. 0.5 >= p? true, so continue:
update: g = (2.0+1.5)/2 = 1.75; h = 3/1.75 = 1.714285714285714; run again.

x = 3, g = 1.75, h = 1.714285714285714. etc, you get the idea


Code walkthroughs and writing problem-notes are important. Writing it out in plain-english, math, or pseudo-code is important too so you can see where you might have skipped a step or not did the algorithm right. Think about the task, plan it, and then see how the programming language can do it. And it'll be easier/faster to ask the prof on your problems.

Thanks for your reply. I did walk through it a bunch of times, but I was working on the basis that G and H can be updated in either order. Turns out you are correct - G must be updated before H. I ended up changing my code to

Spoiler: show

(def sqHelper (fn [x g h]
(if (< (abs (- g h)) epsilon)

g
(let
[g (/ (+ g h) 2.0)
h (/ x g)]

(sqHelper x g h)))))
(def squareRoot (fn [x] (sqHelper x x 1.0)))

and that worked just fine. I'm going to tweak it more though; I don't like having the sqHelper function separate and accessible by the user. Making it an internal function (terminology?) that's inaccessible outside of the squareRoot function is good practice, and something we'll be doing a lot later on. Plus this is my only class and we're moving slowly and I'm bored so... might as well.


On a related note, I'm not sure whether I love or hate functional programming languages... so confusing, so obscure, so nerdy - but there's a certain beauty to running code that does not resemble any sort of English syntax in the slightest.

[The Stig]

I would love to work in a world where the most important thing to do on a Monday morning is to change/update the internal intranet's "picture of the week". I bet they get paid more than me lol. Fuck....

[Hellfury]

At least I am not alone in my tech misery!

[Guyverarmor]

So logged in this morning....It's a Monday...so I knew I would get thrown at least SOMETHING stupid from a customer request...

Customer wants a dual circuit via DIA (Dedicated Internet Access) and was super paranoid about having everything diverse ( Diverse entrance at both ends of the building, diver routes from the start to end locs and serving hub nodes). Which is fine, costly but fine...but then tells us "Oh, but we only want it on the same CPE. We want to make sure neither of circuits go down, but we are totally fine with it being on the same Juniper Router !" Yeah, think you totally defeated the purpose of protected circuits.

[Coren]

Just saw part of an email chain from a user asking if we had copies of old faxes...
"..we lost all of our incoming faxes from last night and today due to our pregnant coworker cut them all in pieces."

You can't make this shit up.

[Ratatapa]

Just saw part of an email chain from a user asking if we had copies of old faxes...
"..we lost all of our incoming faxes from last night and today due to our pregnant coworker cut them all in pieces."

You can't make this shit up.

well do you?

[Coren]

well do you?

All analog, no copies. Tough shit!

[Melena]

So after months of fighting with county IT about our network performance and speeds that rivaled AOL Dial up. they finally admitted today it's not an individual user issue, it's a global issue somewhere. They swear it's at LAN level and not WAN, but at this point i'm ready to start making them tear shit out and replace it.

[Ratatapa]

So after months of fighting with county IT about our network performance and speeds that rivaled AOL Dial up. they finally admitted today it's not an individual user issue, it's a global issue somewhere. They swear it's at LAN level and not WAN, but at this point i'm ready to start making them tear shit out and replace it.

Time for port mirroing and wireshark!

[Melena]

They think it's on the wan side. They switched my vnet on my PC and everything looked gravy so they have to yell at AT&T

[TheDirtyHobo]

My dad's computer is shitting the bed so I told him I'd build him a PC out of my old parts for $400 so that I could use that money to upgrade my cpu/mobo/ram. After about 3 hours of deconstructing and reconstructing to get the two rigs squared away, re-installing windows on my new one, then going to install it on the old one, I realized the old one wouldn't turn on at all. Double checked all of the cables to make sure everything was right, wasn't fixing it. Pulled the power supply out of my new one to make sure that wasn't the problem. Put the other supply into the new one to make sure I didn't damage the mobo somehow while transfering it. Still nothing. Ripped out everything and swapped cases, and finally realized it was because of a faulty power switch on the case.

A good 5 hours of testing, disassembling, and reassembling with highly sensitive, delicate parts that easily could get ruined, all to find out it was being caused by the lowest-tech part. Fuck computers.

[The Stig]

A moderately cute girl who manages services comes over to me "Hey Stig is this ticket related to ERP?"

me: <twitches>
<beat>/<blinks>
"Come again?"

"ERP, I think enterprise resource planning?"

me: "Oh, that. Not sure, it wasn't considered when we constructed the ticket.. who asked?"

Yeah, I was momentarily confused. I thought I put an ad up for ERP unknowingly hahaha (erotic role play)

[Yuri-G]

made github account (shut up I'm very late, I know this)
verify account
enter password to verify
...
......


DAMNIT WHAT WAS MY PASSWORD AGAIN?


good lord. forgot password less than 60 seconds from creation. that is a new record for me.

[Yuri-G]

searching on our company job site, it times out "due to inactivity" AS I'M USING IT


fuuuuuuuuu can I just create a job, and the job will be "fixing this piece of shit website"?