WTF Bites

@Bulb said in WTF Bites:

@LaoC said in WTF Bites:

~~Perl 6~~Raku

Is there still anybody interested in that‽

Apparently. I mean, it's very baroque but also quite elegant. Definitely not for production but nerdy enough to give it a go.

Yeah, it doesn't matter how elegant it is; if it's broke, then it's not for production.

LaoC

@djls45 said in WTF Bites:

@LaoC said in WTF Bites:

@Bulb said in WTF Bites:

@LaoC said in WTF Bites:

~~Perl 6~~Raku

Is there still anybody interested in that‽

Apparently. I mean, it's very baroque but also quite elegant. Definitely not for production but nerdy enough to give it a go.

Yeah, it doesn't matter how elegant it is; if it's broke, then it's not for production.

coughC++cough

dkf

@topspin said in WTF Bites:

@dkf why anybody would still be intentionally using Python2 is beyond me.

A semi-legit answer would be "because we have some old software about". However, some idiots are still actively developing new software for 2.7 and that's just dumb as heck on ordinary computers.

MacOS seems to still be shipping Python 2.7 only.

Apple are in the process of trying to promote doing everything with Swift in XCode by a mix of low and underhanded tactics. Their support for everything else is getting gradually more and more shaky. Alas, it doesn't make Swift any good, much less its IDE, and portability to Apple means running on two versions of iOS...

dkf

@levicki said in WTF Bites:

This is ridiculous...

If the next line had been "Hammertime!" it would have been easy to forgive them.

Mason_Wheeler

@levicki said in WTF Bites:

Oh man

This is ridiculous...

Especially if the reason I am looking at the console is to figure out why your live chat is not working and why your live chat CDN is on uMatrix blacklist.

That's actually perfectly legitimate if that isn't the reason you're looking at the console. The main purpose of that message is to catch non-technical people who don't know what they're doing (ie. ordinary users, not you and me) who have been directed by a scammer to run this "magic hack" in order to get something cool.

error

Discover sent me a letter saying, basically: "we see you haven't used your card in a long time. To encourage you to start using your card again, for the next year, we're giving you a promotional 0% APR on all purchases and balance transfers!"

I think, "awesome, I'll start by transferring all my balances to this no-interest card!"

Except, the limit on this card is less than a tenth of the limit for all my other cards. Oh, well, when I opened this card, my credit rating was substantially worse, so they gave me a low limit. Now that my credit score is almost 800, it should be easy to get a credit line limit increase...

Denied. We'll snail mail you a letter with the reasons for your denial.

** two weeks lateur **

Your credit limit increase was denied for the following reason(s):
low/insufficient account usage

Yes, that's the reason you gave me this promotion and the reason I asked for a limit increase. So I could use the damn account more.

Watson

@levicki said in WTF Bites:

TIL you can do that in the console.

Mason_Wheeler

As part of an application process, I just got directed to take an online coding evaluation. No problem. Those are pretty simple. Except...

When I go to take the test, they spring a surprise requirement on me: this is a video-proctored test. Please turn on your webcam. Sorry, but I don't have one.
I explain this to the recruiter. He asks if I could "find one." (AKA go out and buy something with my own money so as to be able to participate in a job application.) I point out that this is completely unreasonable.
Recruiter finds a non-video-proctored version and sends me a link. I go into the test, and immediately things start looking just a little bit off. For example, asking questions about LINQ to SQL and other technologies that have been obsolete for 10 years.
The test appears to have been written by someone with a minimal grasp of the English language. Almost every multiple-choice question had a spelling or grammar error somewhere, exceptionally ambiguous language, multiple answers that might be correct depending on how you interpret them, no answer that was completely correct, or more than one of the above.
In addition to the multiple-choice part of the test, there was a "coding challenge" section that basically ran as a HackerRank clone. (Here's the issue description, here's where you enter your code, it gets compiled and run against a bunch of test cases and you get back input on how it went.) All of the "problems" were trivially easy, but on one of them, I passed all of the test cases except for one, labeled "Basic case 1", the one that was supposed to be the simplest of all. (I did fine with the ones marked with names like "edge case" and "range check" that were apparently deliberately designed to break your code.) But all of the test cases were hidden, with no way to determine what was going wrong. So... sorry, but no. In real-world usage, if you tell me there's some vague "something" wrong with my code, but refuse to provide any details whatsoever, that's not happening. If I can't reproduce the bug, it does not exist.

At the bottom of the test window, there were two support phone numbers, marked with little flag icons: and . Once I realized what that second flag is, I thought "well that explains a lot." So, we'll see how I did, and I'm starting to seriously wonder if I even want to work at this place at all...

error

@Mason_Wheeler The trick is, if both conditions are true, you need to print "fizzbuzz"

HardwareGeek

@Mason_Wheeler said in WTF Bites:

What a surprise. I guessed that way back at

The test appears to have been written by someone with a minimal grasp of the English language.

Carnage

@Mason_Wheeler said in WTF Bites:

So... sorry, but no. In real-world usage, if you tell me there's some vague "something" wrong with my code, but refuse to provide any details whatsoever, that's not happening.

I've had that happen on a professional setting. Several times in a specific project. A vast majority of the times, I had to prove that the other guys system was the source of the problem.

Polygeekery

Client had a RAID array shit itself today. They are running an older PowerEdge T420 with a H310 PERC card. RAID card appears to be the root cause.

No biggie. We should be able to find tons of them on eBay for dirt cheap, right?

Ha! Most of them seem to have been flashed to IT mode for use with ZFS. I dug enough and eventually found one, but I bet 95% of them have been flashed with LSI firmware so they can be used with ZFS.

Tsaukpaetra

@Polygeekery said in WTF Bites:

Ha! Most of them seem to have been flashed to IT mode for use with ZFS. I dug enough and eventually found one, but I bet 95% of them have been flashed with LSI firmware so they can be used with ZFS.

https://www.youtube.com/watch?v=R6LzEbuI7e8

While you're searching/waiting.

Tsaukpaetra

Lol whoops? Thank goodness that wasn't me that coded that portion.

LaoC

@Mason_Wheeler said in WTF Bites:

The test appears to have been written by someone with a minimal grasp of the English language. Almost every multiple-choice question had a spelling or grammar error somewhere, exceptionally ambiguous language, multiple answers that might be correct depending on how you interpret them, no answer that was completely correct, or more than one of the above.

They're just checking whether you can deal with customer requirements.

Jaloopa

@Mason_Wheeler said in WTF Bites:

. sorry, but no. In real-world usage, if you tell me there's some vague "something" wrong with my code, but refuse to provide any details whatsoever, that's not happening. If I can't reproduce the bug, it does not exist

I see you've never done third line support

THING IS BROKEN

could you please tell me what version of thing you are using, what you did and what happened?

I DON'T UNDERSTAND YOUR TECHY TALK JUST FIX THE THING

Carnage

@Jaloopa said in WTF Bites:

@Mason_Wheeler said in WTF Bites:

. sorry, but no. In real-world usage, if you tell me there's some vague "something" wrong with my code, but refuse to provide any details whatsoever, that's not happening. If I can't reproduce the bug, it does not exist

I see you've never done third line support

THING IS BROKEN

could you please tell me what version of thing you are using, what you did and what happened?

I DON'T UNDERSTAND YOUR TECHY TALK JUST FIX THE THING

Extra bonus when someone from the upper echelons of manglement has gotten involved and is constantly repeating "WE'RE LOSING ALL THE MONIES! FIX NAO!!!"

Zerosquare

@Polygeekery said in WTF Bites:

Ha! Most of them seem to have been flashed to IT mode for use with ZFS. I dug enough and eventually found one, but I bet 95% of them have been flashed with LSI firmware so they can be used with ZFS.

I assume flashing the original firmware back wasn't possible?

topspin

An internal tool that converts some user input from 1/min to 1/s before processing further things came up with this number: 16.666666666668

kind of rounding rounds that digit up to 8?!

INB4 :pendant: of course the answer is floating point inaccuracy, dummy.

BernieTheBernie

@topspin said in WTF Bites:

An internal tool that converts some user input from 1/min to 1/s before processing further things came up with this number: 16.666666666668

kind of rounding rounds that digit up to 8?!

INB4 :pendant: of course the answer is floating point inaccuracy, dummy.

Why did that user enter a value of 1000.0000000000015?

topspin

@BernieTheBernie Unlikely, no.

(I'd remember that. )

PleegWat

@BernieTheBernie said in WTF Bites:

@topspin said in WTF Bites:

An internal tool that converts some user input from 1/min to 1/s before processing further things came up with this number: 16.666666666668

kind of rounding rounds that digit up to 8?!

INB4 :pendant: of course the answer is floating point inaccuracy, dummy.

Why did that user enter a value of 1000.0000000000015?

We had a guy like that in QA. It's a shame he got moved to a different product; a tester like that keeps a man honest.

pie_flavor

@levicki said in WTF Bites:

Oh man

This is ridiculous...

Especially if the reason I am looking at the console is to figure out why your live chat is not working and why your live chat CDN is on uMatrix blacklist.

That's not ridiculous at all, that's perfectly sane. Someone can post a 'guide' on how to 'fix' something, which involves using the console, in order to get access to your account. You know, that thing it's warning you about. Lots of sites do this because it's a relatively common exploit.

Discord, for reference:
https://i.imgur.com/q84ZDbw.png

Bulb

This post is deleted!

Bulb

@topspin said in WTF Bites:

An internal tool that converts some user input from 1/min to 1/s before processing further things came up with this number: 16.666666666668

>>> 1000/60
16.666666666666668 ← Python's result

BUT that is long-double and has four more digits than yours. And in double precision it does not happen:

> print(1000.0/60.0)
16.666666666667 ← Lua's result

Also

>>> math.log10(2**53)
15.954589770191003

so a double should have enough precision for almost 16 significant digits. So there should only be gaps in the representable sequence at 16 digits, but your value only has 14 significant digits¹ (and so does Lua's).

So the calculation must have done something more complex than merely dividing by 60.0. confirmed.

¹ Python's value even has 17, which is again weird, because it is too many for double and too few for x86's long double, which should have 19-and-a-bit significant digits with it's 64-bit mantissa. And Python actually does print out floating point values as shortest decimal representation that parses back to the same value, so using 17 implies that the 17th place still makes a difference, but the 18th does not.

⇒ Floating point is weird.

Update: Python can actually work with hexadecimal representation of floats. So

>>> 16.666666666666668.hex()
'0x1.0aaaaaaaaaaabp+4'
>>> float.fromhex('0x1.0aaaaaaaaaaaap+4')
16.666666666666664

hm, the actual distances don't really match the logarithms very well. But that hex representation has 13 hex digits behind hexadecimal point, which corresponds to the 52 bits + implicit 1 for plain old double.

So there is two more digits of precision when rounding to 14 digits. still confirmed (and me left a bit less , but still nonetheless).

PleegWat

@Bulb How are you counting the significant bits from the 16 before the decimal point? That takes up exactly 5 bits in the binary mantissa; you should do your logarithm maths with the remaining 48 bits.

Bulb

@PleegWat said in WTF Bites:

@Bulb How are you counting the significant bits from the 16 before the decimal point? That takes up exactly 5 bits in the binary mantissa; you should do your logarithm maths with the remaining 48 bits.

I am not separating the number to integeral and decimal part, because it is not represented that way. The position of the decimal point is taken care of by the exponent. So I am taking 17 digits in total and that's correct.

What makes the difference though is that there is a difference in what the leading digit is. A 1 is worth less than a 9 is, so if the leading digit is 1, the 17th digit matters, but when the leading digit is 9, it does not:

>>> 96.666666666666668
96.66666666666667

See, one fewer digits.

loopback0

@pie_flavor




                                            
 .d8888b.  888                       888    
d88P  Y88b 888                       888    
Y88b.      888                       888    This is a browser feature intended for 
 "Y888b.   888888  .d88b.  88888b.   888    developers. If someone told you to copy 
    "Y88b. 888    d88""88b 888 "88b  888    and paste something here to enable a 
      "888 888    888  888 888  888  Y8P    Facebook feature or "hack" someone's 
Y88b  d88P Y88b.  Y88..88P 888 d88P         account, it is a scam and will give them 
 "Y8888P"   "Y888  "Y88P"  88888P"   888    access to your Facebook account.
                           888              
                           888              
                           888              

See https://www.facebook.com/selfxss for more information.

PleegWat

@Bulb said in WTF Bites:

@PleegWat said in WTF Bites:

@Bulb How are you counting the significant bits from the 16 before the decimal point? That takes up exactly 5 bits in the binary mantissa; you should do your logarithm maths with the remaining 48 bits.

I am not separating the number to integeral and decimal part, because it is not represented that way. The position of the decimal point is taken care of by the exponent. So I am taking 17 digits in total and that's correct.

What makes the difference though is that there is a difference in what the leading digit is. A 1 is worth less than a 9 is, so if the leading digit is 1, the 17th digit matters, but when the leading digit is 9, it does not:
>>> 96.666666666666668
96.66666666666667
See, one fewer digits.

The integral and fractional parts aren't stored separately, it's binary exponential notation. However it is valid see how many bits are taken up representing the integral part, and deriving from that how many bits are left for the fractional part. The decimal point will be in the same place regardless of whether you're representing the number binary or decimally.

Zecc

@levicki said in WTF Bites:

I need to put that warning on my site.

It will read:

"STOP! If you are looking here because you are a developer then you are a moron."

Bonus points if your website does not have any other JavaScript.

Polygeekery

@Tsaukpaetra said in WTF Bites:

Lol whoops? Thank goodness that wasn't me that coded that portion.

955b7123caa9cd6b9acd8c164957749f8f313655082446f5aa1f72fa1431a259.jpg

Polygeekery

@Zerosquare said in WTF Bites:

@Polygeekery said in WTF Bites:

Ha! Most of them seem to have been flashed to IT mode for use with ZFS. I dug enough and eventually found one, but I bet 95% of them have been flashed with LSI firmware so they can be used with ZFS.

I assume flashing the original firmware back wasn't possible?

Maybe, but not something I want to dick with or take a chance on considering I don't know for certain the competency level of the person that did it or what the source of the firmware is.

Also, I have heard that Dell>LSI firmware flashing is a one-way operation. Dell utilities check for Dell signatures before they will flash. PERC cards are made by LSI. I assume they don't want you buying the LSI equivalent (at half the price, or less) and flashing a Dell firmware and popping it in your server and not paying them their upcharge.

It probably could be done but I'm not fucking with it.

Polygeekery

@levicki said in WTF Bites:

I need to put that warning on my site.

It will read:

"STOP! If you are looking here because you are a developer then you are a moron."

Mine would read:

Stop. Don't look further. You will just judge me and I did a lot of this while my BAC was beyond the legal limit of drink-drive laws. My computer really should have a breathalyzer on it.

topspin

@Bulb said in WTF Bites:

So the calculation must have done something more complex than merely dividing by 60.0.

Maybe some other unit conversion I ignored and things also got multiplied/divided by another 1000 somewhere? Whatever, it was just a curiosity seeing that expanding to 8, I don't care to look deeper into it.

loopback0

HardwareGeek

@loopback0 said in WTF Bites:

it is a scam and will give them
access to your Facebook account.

Nothing of value lost.

HardwareGeek

@Polygeekery said in WTF Bites:

@Tsaukpaetra said in WTF Bites:

He's on staff at a MLP convention. No, he doesn't.

Polygeekery

@HardwareGeek said in WTF Bites:

@Polygeekery said in WTF Bites:

@Tsaukpaetra said in WTF Bites:

He's on staff at a MLP convention. No, he doesn't.

cvi

@Polygeekery Maybe if you ask him nicely, he can hook you up with a discount?

Polygeekery

@Polygeekery said in WTF Bites:

Client had a RAID array shit itself today. They are running an older PowerEdge T420 with a H310 PERC card. RAID card appears to be the root cause.

No biggie. We should be able to find tons of them on eBay for dirt cheap, right?

Ha! Most of them seem to have been flashed to IT mode for use with ZFS. I dug enough and eventually found one, but I bet 95% of them have been flashed with LSI firmware so they can be used with ZFS.

Got the array imaged yesterday. Coalesced it down to a single disk to get them back up temporarily. Hand off the drive and badge to one of my guys to install this morning. 9am this morning he calls me to tell me they are back up and running and says that some guy stopped by while he was there and said he was a board member who was there to sign checks.

"Did you let him in?"
"No, but I told him that if he could have someone call from the NFP to verify he is who he says he is and that he should be there I would be happy to let him in but that I couldn't let anyone in without being authorized to do so."
"Good deal, that's how I would have handled it."

40 minutes later....

"Hello?"
"Hey, one of our board members stopped by the office to sign and pick up checks and your guy wouldn't let him in and now he has to come back later so we have payroll for tomorrow."

My first thought is:

"I will file this under NMFP....Not My Fucking Problem"

What I said was:

"Yeah, I understand yours and his frustration, but look at this from our perspective. We are in the office, working alone early in the morning and someone we do not know stops by and says they are someone that we have no way to verify and have no way to know if they are who they say they are and that they are supposed to be there. He could have called you, you could have called me, I would have called him and he could have come in. Hell, if it is who I think it is he could have called me direct. He has my number.

But do you really want us letting random people we don't in your office just because they say they are members of the board of trustees?"

A bit of silence.

"I had not really thought of that. Good point and that was poor communication and planning on our part."

giphy (5).gif

Zerosquare

@Polygeekery said in WTF Bites:

"I had not really thought of that. Good point and that was poor communication and planning on our part."

A client admitting they goofed up and you're right? Come on, if you want us to believe your stories, you gotta make them believable.

JBert

@Zerosquare This is outside of The Lounge, so maybe he just embellishes stories a bit to uphold the positive image of his customers.

hungrier

@JBert "And that's why I had no motivation to burn down [client]'s house, Your Honor"

Polygeekery

@Zerosquare said in WTF Bites:

A client admitting they goofed up and you're right? Come on, if you want us to believe your stories, you gotta make them believable.

It all depends on how you handle it. If you're right, and you know you're right you need to explain it to them in a way that gets your point across incontrovertibly without being a dick about it and making them defensive.

That last part is where I usually struggle.

TimeBandit

@Polygeekery said in WTF Bites:

you need to explain it to them in a way that gets your point across incontrovertibly