WTF Bites
-
@sebastian-galczynski Yes, but that requires putting the conditions on the type parameters right so that you can actually safely consider the subtype rule to be satisfied. Which means you've now got to do a non-trivial set of type constraints to determine safety (and things will only get worse as you push that further in). It's really easy to end up with the cure being worse than the disease; you don't want the types in your program to be more complicated than the program they describe constraints on.
-
you don't want the types in your program to be more complicated than the program they describe constraints on.
Some people do
-
@sebastian-galczynski Yes, but that requires putting the conditions on the type parameters right so that you can actually safely consider the subtype rule to be satisfied. Which means you've now got to do a non-trivial set of type constraints to determine safety (and things will only get worse as you push that further in). It's really easy to end up with the cure being worse than the disease; you don't want the types in your program to be more complicated than the program they describe constraints on.
Last year I hit a case where invariant collections led exactly to the "types being longer than the program" problem, because despite the collections in question being of a similar thing, I couldn't put a common interface on them. Which required multiple template arguments to deal with collections inside collections, and a big union type on top. Not exactly "easy".
-
@sebastian-galczynski Ultimately, the intent of types is to provide a way to reason about part of the correctness of a program without having to tackle the total correctness of the program, because that latter problem is generally intractable (it's where you hit the halting problem and all sorts of unsolved conjectures in math, as well as some really rotten problems with state space explosion with complexity classes like EXP2SPACE). But the more we ask types to prove, the worse the tractability for those types becomes. It's inescapable; if we have the actual correctness of the underlying program being often insanely difficult to work with, the more of the correctness features we push into the type system, the harder it will be to reason about the types of the program.
Runtime failures are a way out of the trap, but force you to deal with operations that can fail, and there really are a great many of those.
-
Status:
How did I find myself in the word salad thread?
-
@Tsaukpaetra said in WTF Bites:
Status:
How did I find myself in the word salad thread?That's easy - you browsed TDWTF.
-
Not particularly portable
The purpose is being able to serialize a data structure and restore it exactly, which can be done inside one language only, so it does not need to be particularly portable.
For sending data between different applications, possibly written in different languages, you need to know what data is expected. That is, you should have some kind of schema. And you can define which members should be deserialized as which type there.
You always have one of three schemas in YAML that differ in portability. Even a Perl hash dumped according to the "failsafe" schema isn't always exactly restorable say to a Java
Map<String, Object>, probably not even to a Python dictionary, but if you do cross-platform YAML at all, that's what you have to live with. They could have added other slightly less portable standard schemas, but they didn't.
-
@LaoC That's not what I meant. I even forgot YAML had that term used this way.
By schema I mean you need some kind of definition not just that you get JSON (or YAML or XML), but what it will actually contain. You never expect ‘any JSON’.
{"this": "is a json", "but": ["useless"], "it": {"probably: "is"}}.You expect something specific encoded as JSON. Say, a list of files, where file is an object with name: string, mtime: datetime, owner: user and uri: uri. And those subtypes have their own definition. In Java or C# or Go or Rust you'll just declare the corresponding class/structure type and tell the deserializer to populate it. But even in untyped language you'll need something to covert that iso-8601 string to a Date and validate that uri.
For XML there is XSD, for JSON there is JSONSchema. For YAML there is … also JSONSchema and it's usually written out as JSON though it would technically be possible to write it as YAML and it would probably be a bit more readable that way.
-
@Bulb There certainly is something, in that I've seen that the YAML language server I use in Eclipse pick up a way of guiding what some of the files I write accept and provide content assist. (Specifically, I've noticed it help with writing Github Actions workflow files. I also get assistance with Spring Boot application configuration files, but I think that's separate as it involves app-specific metadata.)
-
@Bulb There certainly is something, in that I've seen that the YAML language server I use in Eclipse pick up a way of guiding what some of the files I write accept and provide content assist. (Specifically, I've noticed it help with writing Github Actions workflow files. I also get assistance with Spring Boot application configuration files, but I think that's separate as it involves app-specific metadata.)
Yes, the JSON schema @Bulb mentioned is gaining traction...
-
@Kamil-Podlesak said in WTF Bites:
@Bulb There certainly is something, in that I've seen that the YAML language server I use in Eclipse pick up a way of guiding what some of the files I write accept and provide content assist. (Specifically, I've noticed it help with writing Github Actions workflow files. I also get assistance with Spring Boot application configuration files, but I think that's separate as it involves app-specific metadata.)
Yes, the JSON schema @Bulb mentioned is gaining traction...
Pfff... I remember when compressed binary was all the rage.
-
The results I get for "apple iphone".
11 billion results. 3 pages
Click page 3... now there's 8 pages.
But only 74 results suddenly.
Aha!
Including those omitted results...
9 billion - which is obviously more than 74 but still less than 11 billion
-
@loopback0 said in WTF Bites:
The results I get for "apple iphone".
11 billion results. 3 pages
Click page 3... now there's 8 pages.
But only 74 results suddenly.
Aha!
Including those omitted results...
9 billion - which is obviously more than 74 but still less than 11 billion
It's "about", as in "up to", as in Math.random()
-
@loopback0
did you forget to put the hyper in your loglog?
-
-
: "Your computer needs to restart because retards in IT pushed too many useless updates today"
: Fuck. This shit again. Ok, restart.
: Hai. I'm back 
: No, you aren't, you lying turd. [wait 5 minutes for all retarded IT startup shit to run]
: Now I'm not only back, but clicking things works again!
: Astonishing. Run 
: Trying to connect. Trying really hard...
: ...
: Aaaand... failed.
: What now, ffs.
: While you browse old emails, could you take a look at updated list of disabled plugins? One of them was slow, so I disabled it. Now I'm super fast.
: You never are even tolerably slow, not to mention approaching fast. What plugin did you disable?
: Exchange.
: I see. Yes, very helpful. Enable it and don't disable it ever again.
: Best I can do is 30 days. After that I'll stop it again.
: I love my work 
-
This post is deleted!
-
This post is deleted!
-
⸘What The Mighty Flying Fuck is happening with Azure‽
From my company notebook, connected to the company network, I can't reach either portal, or a company application installed in Azure.
But from the customer notebook, sitting right next to it and connected to a wifi that is separate from the company network, but ultimately routed through the same cable, I can reach both fine.
And a colleague wrote to me that he can't reach them while working from home, so he's at a completely different network.
… imps seem to be attacking the infernet today.
-
-
-
From my company notebook, connected to the company network, I can't reach either portal, or a company application installed in Azure.
But from the customer notebook, sitting right next to it and connected to a wifi that is separate from the company network, but ultimately routed through the same cable, I can reach both fine.
And a colleague wrote to me that he can't reach them while working from home, so he's at a completely different network.
Sounds like some sort of bad DNS info was pushed, but hasn't reached all networks yet. (Or maybe a BGP fuck-up; I don't know how the caching for them works.)
-
@Vault_Dweller That would then be Microsoft 364, though
-
Guess this explains the slightly obscure message about some outage that I received through Teams earlier today.
-
@dkf DNS no; the IP address resolved was the same on the computer where it worked and the one where it didn't, and they are stable anyway. BGP might be.
… coincidentally we had a big BGP breakdown between the company and the provider on Monday that took something like four or five hours to fix.
-
@Vault_Dweller said in WTF Bites:
This shit's so amazing.
The day started with being told Teams doesn't work ("finally I can get some work done, yay!"), to messages that basically all of Office 365 is down, and emails don't work either. Thankfully, I don't do anything Azure, which I guess is also fucked.Isn't it great that we went all in with Teams/O365/all that other crap just because "everybody is using it" and "nobody ever got fired for buying
IBMMicrosoft"? Fucking central IT one day during the pandemic just said "stop using everything else, we're doing Teams now" and that's that.
Email used to just work.Who wants to bet that even the offline desktop apps of Office don't work anymore because for
you need to log in to use them?
-
Email used to just work.
In every company it blew up now and then. The difference is that then everybody was using their own mail server while now everybody uses the same one, so when it blows up, it affects a lot of people.
-
Thankfully, I don't do anything Azure, which I guess is also fucked.
Not anymore, but it was earlier this morning.
For "some" customers in Western Europe, between roughly 3 and 5 UTC.Guess who got rudely awaken by an effing automated call because several databases were down?
-
@nerd4sale said in WTF Bites:
Not anymore
The outage is now marked as resolved on isdown, but portal still isn't loading on my company notebook (but it has been loading on the customer one throughout the outage).
-
-
@Bulb With a self-hosted email server, internal email would (typically) keep on working even when the external internet was out. Overall, there are noticeably more points of failure now: email server, network of the service host, various network connections, and your network. Before it was the email server and your network, at least for internal email. (A lot of my email is internal.)
-
@cvi Only as long as it's physically internal. Once it's vnet across geographically distant locations, and some people work remotely, you've got all those network failure points anyway. And things have always been distant locations for me.
-
@Bulb You also hit that when you're doing a lot of collaboration with external people. For us, if we're off the net then it's disaster time.
-
Google security WTF while logging into my email in a private window on my work laptop.
It shows the "confirm on your phone to login" message like normal, so I pick up my phone, waiting to get the security popup. And waiting. And waiting. I check internet connectivity on my phone, it's working; my computer is fine too.
I hit resend on the computer, and still nothing. So I go to "Try another way" in which one of the options is to get a security code through settings -> google -> security -> whatever. I enter that code and get into my email.
About a minute later, the security prompt comes up on my phone
-
-
@JBert The problem is clearly with the cabling. If the cables nearly hanging out the bottom had been going in all directions and been attached to nearby walls and the roof, they could have probably held up the whole rack.
-
@cvi As seen in this documentary:
-
@hungrier Looks more like he's highly constipated.
-
@hungrier Looks more like he's highly constipated.
But making progress. Let's just give him a minute.
-
@hungrier Looks more like he's highly constipated.
You'd look like that too if you were shitting out a train
-
@loopback0 reminds me of
-
One of our clients uses a thing. It has an on-site server component. That server software recently had a critical security vulnerability discovered. Time to update said software at all of their sites.
So their support site requires a login. I have the password saved in my password manager. That password does not work. As soon as I click on their password reset link I remember that it does not work. You have to email support. It does not tell you that, it just silently fails. But it does tell you if you entered an invalid email. Brillant!
But it has been a year or so since I have had to access their support site. Perhaps they fixed it? It does not appear so as no reset email ever comes through.
So I email support requesting a password reset and in the same email I ask a couple of questions that spin off into email chains with three separate people. One of these people is a generic support account, the other two are proper named accounts.
Every single time I email the support@initrode.com address my goddamn password gets reset. Not "I get password reset email that I can ignore". Every reset email invalidates my previous password when it sends the email. Brillant!
I tell support this. They say that it is best to either reset my password on each email or if I do not need to access the support site just ignore them for now. I suggest the option of starting a new email chain. No. Need to keep everything on one chain and procedures require merging tickets if they reference the same issue. How about merging later after we are done? Nope, can't do that. SOPs have to be followed. Can I switch to someone who can assist on a named account outside of this thread? Nope, can't do that, SOPs. So what about the other two fucking people? Oh, those are different types of issues.
I'm getting deja vu. I think this is how my password got fucked up and not stored in my password manager last time. Reminder to myself that next time I have to contact their support to send two emails. One for password reset, another for the actual issue. This is what happens when businesses roll their own ticketing and support systems.
-
@Polygeekery said in WTF Bites:
a thing
A fully automated fire detecting and extinguisher device? No?
One can hope ...
-
@Polygeekery said in WTF Bites:
a thing
A fully automated fire detecting and extinguisher device? No?
One can hope ...Why stop there? Couldn't it also start them?
-
@Gribnit
They would get along like a house on fire. A roofless house on fire.
-
@loopback0 reminds me of
And that's not even the lame 7.5G the rhino can pull, just about 5. While real fighter jets can pull 9.
-
And that's not even the lame 7.5G the rhino can pull, just about 5. While real fighter jets can pull 9.
Psht. Any decent centrifuge can put those numbers to shame.
-
@Gribnit Most centrifuges aren't armed with either missiles or machine guns, and the number with bombs is probably classified.
-
I was disappointed to find out that it's not his real name.
-
@Gribnit Most centrifuges aren't armed with either missiles or machine guns, and the number with bombs is probably classified.
However, nearly all allow for fully remote or even automated operation, a handy feature past a handful of G's.
The number with bombs is (also) all of them, if you drop a screwdriver in.
