“Typed language? Yeah, I’m using a keyboard.”
I am currently teaching python and JavaScript devs Typescript. Everytime they hit a problem they switch to any
Sigh
Must be the same people who just comment out failing unit tests.
“Your crappy tests are failing again on my branch. I’ve commented them out until you fix them.”
Sadly that sort of thing got so common where I work that I’ll run the tests three times before considering looking into the error message to see if it is something I broke.
From time to time we take some days just to fix tests with inconsistent results, but there’s always more popping up.
Serious answer: You can’t write tests for untestable code. Your code needs to be pure if you want reliable tests: https://en.m.wikipedia.org/wiki/Pure_function
For integration tests, they should handle retries themselves
Yeah, we have a team whose job is to make sure all our tests run well and fixing them if they don’t
…or skip em
That’s why I kinda don’t like Python and JavaScript anymore. Every time I want types for a library it’s gonna take me time to get it working. For every serious project I do, I use a strongly typed language.
the beatings will continue until typing improves
Eslint is your friend :)
Just create a al Inter rule that rejects Any types and a pre-commit hook that refuses the commit if the linter fails. Sometimes the brute force approach is the best way to teach
You told them not to?
I am happy there is no obvious “any” type in Rust.
There is, it’s just not easy to use
i like when my strongly typed language can type itself, why should i have to type extra words because the compiler is stupid?
So that next time your coworker uses the wrong type, the compiler can scream at him: “NO I WONT COMPILE THIS YOU DUMBASS, LOOK JOHN SAID ON LINE 863 THAT IT SHOULD BE A DOUBLE, NOT A FLOAT FOR FUCK SAKE”
Tell me you are a Java dev without telling me you are a a Java dev 😂
As a JS dev, I can only wish we had those types 🥲
you can still have that without having to declare the type manually. check out Swift or OCaml for example
deleted by creator
Type error unless there’s an implementation of
+
that specifies adding together and integer and a string.💯% accurate. funny how the typescript developer thinks this is some kind of “gotcha!”… like maybe just try a language besides typescript and find out for yourself 😆
deleted by creator
“brought” 😏
my complaint is that typescript is stupid, yes. so why wouldn’t i compare to what other languages do that is less stupid?
on the plus side, at least now i know that the ad-hominem minded devs came here too, and brought their righteousness with them.
deleted by creator
Exactly. Most languages I know of that allow this at all will coerce the “1” to an integer and give x = 2. They get away with this because they define the “+” operator as taking numbers only as arguments, so if you hand them
x = x + "cheese"
they’ll error out.OCaml 😍
I’m not sure if you’re being rhetorical or not, but “string|number” is definitely correct here. A computer could definitely figure this out, but typing is for the benefit of the coders more than the code itself. It’s basically functional documentation
deleted by creator
Yeah that’s what I’m saying, I hate it when coworkers will assign everything as “any” just to avoid the scary red squigglies. Oh well I guess that’s what code reviews are for 🙃
In the world of C and pointer arithmetic this makes perfect sense /s
Yeah, at some point my new team switched off null safety, because some consultants told them to.
Ride into the Danger Zone …
Indeed, and just as my old team fell for consultants, my new team also went ahead and let them add some overcomplex garbage into their codebases. And crap still keeps piling up. It’s just like it’s impossible for them to understand that from an average consultants perspective the only way to go forward is to keep adding complexity, wether they are aware of it or not.
Oh, the consultants know, but they get paid, don’t complain about “risks” and “code debt”, and management only sees their delivery on time without increasing operation costs
This. However, in our specific scenario dynamics were even slightly worse. In a first meeting said consultants apparently met some resistance but management decided to go through with it anyway. So in a later meeting, if I was the consultant, would I go and claim “Alright, I fucked up, got paid and got you gaslighted, but now we have to refactor to clean up our codebase with no immediate tangible benefit for your bosses” in front of everyone? Honestly, I don’t think so.
Does it compile???
… Compile???
And if you have linter rules preventing
any
as a boundary type you just useRecord<String, any>
.I’m in this post and I don’t like it.
That being said I try to have specific types in my typescript but coming from working without typescript, there’s so much more words involved using typescript and for what I use it for I don’t really see the use case. Sure it helps you realize what part of the script needs what data types but it adds so much more complexity in the code that I’m not really sure it’s worth in the first place.
Typescript saves ridiculous amounts of time in bugfixes and is IMO a lot more readable than JS.
I don’t know how many times TS has complained about some type mismatch in my code that made me scratch my head for 2 seconds to only then realize I was doing something stupid. With plain JS that would’ve been no issue, until I have some obscure bug 30 minutes later and have to figure out it’s source.
Also, whatever piece of code you are working on, to do anything you have to have the types of your variables/functions in mind. If you have to keep track of all of them in your head, you will definitely mess it up at some point or have to look through a bunch of different methods/files to track down the source of some piece of data to be certain what’s contained in it.
So yeah, TS might take slightly longer to type out, but it saves you a lot of dev time.
I mean I guess that could be helpful, I’ve never really had that issue so I have yet to see the benefit of it. I just find it useless work that you’re typing out for something that the engine itself isn’t going to be able to see anyway, which means you’re going to have to have unit tests coded in regardless. And I wouldn’t say just a little more coding, typescript when implemented into my project doubled the amount of code provided, I’m trying to use it because I do understand it’s a standard, but I really don’t understand why it’s a universal standard, considering that everything it does is completely syntax sugar/coder side and it doesn’t actually interact with the underlying engine. I feel the same way about coffee script honestly.
Just wait until you have to work as part of a team on a big project. The lack of types will murder the team’s productivity
OK, so, a lot here.
First of all, Typescript isn’t a standard, much less a universal standard - it’s a fairly patchy implementation of strong typing over JavaScript, created by a private corporation. If you’re going to write strongly typed JavaScript, it’s probably the best option, though not the only option. If you want a strongly typed, functional language, and are not wedded to JavaScript, it’s definitely not the best option.
Strong typing is an incredibly powerful tool,and will make your life easier, once you wrap your head around it, but it isn’t the only way to go, for all projects. There are definitely some (typically small) projects that benefit from a more dynamic typing system.
I am not really sure what you are saying about unit tests. It sounds like you are saying you think you need unit tests to assure types on the JavaScript level, even when working in typescript. That’s not the case. You can trust the compile-time type system, if you type your code. You still should have unit tests, of course, but you really don’t need unit tests that check that the types are what your typescript typing says they are. In fact, the more skilled you get at building good types, the fewer tests you’ll need. They’re always a compliment, the goal isn’t to eliminate tests at all, but you will get better safety with less testing.
Typescript probably shouldn’t be doubling the amount of code. This sounds like you might be adding a lot of redundant type aliases, or specifying types that should be inferred. I’m not sure, I’d have to see the code.
You also might want to look at your toolchain. Are you doing a bunch of coding, then running tsc and looking at the results? Or do you have an editor set up that can show you typescript errors and suggested completion in real time? It makes a huge difference, having a toolchain that really supports rapid, iterative, strongly typed development.
So the biggest issue is the project relies extremely heavily on a third party API service, and since the data is received over said service, typescript is unable to infer what the objects the API is sending is because it sends during runtime, to get around this I have to define everything that I expect that the library is going to have to handle that would be Recieved, since any object that the API is going to return is just going to have a type of any if it’s not defined, this on top of the fact that the API has stated that the data being sent should not be relied on for being accurate and types may change randomly(usually it does not but it has happend, it sucks but out of my control) means that I generally also have to have a function level test the data when it’s received to make sure that the value is being supplied are the correct type and are formatted in a way that the library can still understand it. Which means that it’s able to catch any inconsistency of typing before it would be processed anyway, and would either warn or throw depending on how important the function is to actual operation.
The reason why I would call it standard is because it seems like basically anywhere you look if you are using node, you’re using typescript they go hand in hand it seems as of the last two or three years, but honestly I’ve never really understood the benefit of, I’ve always thought it was a fairly standard to have at the beginning of a function the documentation of what each perimeter should be unless it is easily verified by looking at it.
As for my setup, it’s not very advanced it’s just Sublime Text with linter hooked to it, which does tell me on save if there’s a typescript error or if I formatted something wrong, but again even if one did happen to slip through that it would fail during the testing phase due to the fact that it would throw at the function level.
My opinion of my experience with typescript has been that it’s great if everything is operated in house, but the second you start having to deal with stuff that comes from an external source any advantage of the check just seems not worth the extra effort to make sure typescript works right.
TypeScript is essentially the “measure twice, cut once” approach to JavaScript.
Yeah, anything can be anything in JS and the type declarations don’t make it into the compiled JS, but allowing anything to be anything starts to become fairly dangerous when the size of your projects starts to grow and especially when you’re working with a team.
Rather than writing functions and just hoping they always get called with a parameter that has the properties you expect to use, TypeScript helps you make sure that you always are calling that function with the right object in the arguments. You don’t need to debug some runtime error up and down 8 frames in the call stack because this week you named a property “maxValue” but last week you used “maxVal” or you forgot to parseInt some string because you thought it would be coerced - you just need to make sure your types match and eliminate that type of debugging altogether.
All in all, TS really just enforces a bit of sanity to the foot gun that is vanilla JS.
Yeah I fully agree typescript does help in terms of knowing what type of types you should be supplying to functions, and for the most part I do use it for non-library purpose/anything that doesn’t rely on a third party, I just feel like typescript isn’t worth it when you have data that’s returned at run time that’s controlled by a third party service. You end up coding more in class definition files then you would just using normal tests
typescript is not a strongly typed language
I mean that is the first step. ¯_(ツ)_/¯ The next step is to start defining the types more strictly than any.
Hmmm a more reasonable first step would be to just not even type anything until you’re ready. But TS makes it hard to iteratively type parts of your codebase over time. One could type using JSDoc syntax for these cases, though.
Yeah, true
Well, you can always just add the type definitions later on.
I did port some C code to D, by just pasting it in a D file, then fixing the differences (changing type names, rewriting precompiler macros with D metaprogramming and inline functions, etc.).