A huge thanks to you guys who are sharing more useful habits and pointing out some of the benefits that I left out in the comment section. You help both me and many other developers to understand more about this beautiful programming world! I'm super appreciative when you guys share cool information :)
It’s good practice (like he said), a lot of beginners/beginner orientated videos don’t use it though, because either they’re not writing code big enough for it to matter (most beginners use a single file), or for the beginner orientated videos they just pass over it because it’s not really important for the scope of the tutorial, however I believe its function should be taught a bit more than it is…
@@dt3688 Sorta. So, the purpose of the if statement is to explicitly check if your file is run as main (When ran directly the interpreter sets the variable __name__ to “__main__”, but when imported it’ll set __name__ to whatever your filename is.) In doing so, whatever you have inside that if __name__ == __main__ statement will be ran ONLY if you directly run that file/script. Otherwise, if you import the file, whatever you have inside the if statement will not be run. This is useful (mostly) for test code, or if your file is meant to run as both an import and a script on its own. This is to keep whatever imports your script from accidentally running the functions in the file. Ex: def PrintSomething(): print(“This will run”) if __name__ == __main__: PrintSomething() In this code, it will only print the string to your console if ran as a script directly. If you import it, PrintSomething() will not run. (Formatting may have broken for youtube with the underscores for name and main, but yk what I meant)
Another important thing with main() is the scope. Code run in "if name main" uses global scope and can/will pollute it. Running code in main has the function's scope :)
I would argue that this is _the_ most important reason to use a main() function. It's not just a way to make your code look better organised, it has fundamentally different semantics from just putting code in a top-level if-name-is-main
A main(with parameters) is even more useful when testing. Allow the if __name__ block to handle command line arguments and then pass sanitized versions of those to your main. Then your tests can just call main directly and test various scenarios.
I'm just learning python, but if you were to do this, how would you suggest you test your input parsing and sanitation itself? I'm thinking I'd have the '__name__' block call a 'parse_and_sanitise_inputs' function to do that (and pass its results to main), which could then be tested separately.
@@mnxs you would have separate files that use those functions to run some tests that test they do their job. You don't need a dedicated "parse_and_sanitise_inputs" function, for example if you want to check that you provided a correct number within a certain range, such as an age, then you would do a "parse_age" or similar function, that would check if age > 0 and is a number and return that number. I would use something like arg_parser in python
A way to avoid them becoming unreadable without resorting to loops is to split the transformations into multiple separate list comprehensions on new lines, and assigning the result of each line to a variable with a descriptive name to show what the intermediate state represents. Ideally you would use generator comprehensions as well for the intermediary steps and only convert to list at the end (if it's even needed in list form at all).
One major example I've seen, unless you've got a very descriptive comment attached or it's doing a super simple operation, don't put two for loops in one list comp
It depends how you use them. Variable Naming has a huge impact in readability as well. For example: taking the person names > 7 case, if we were to rewrite with more descriptive names it can convey a sense of reality and make the code more relatable to the readers. listLongNames = [ pp for pp in people if len(pp)>7 ] See, makes so much difference.
Actually, the best practice is to throw a not implemented error so you don't forget about it.. Honestly though, it's too much effort. I always thought python should have a keyword for this, something nice and short that throws an exception.
Another bonus is containing the variables. You won't accidentally use a globally-scoped name in another function, if that name is local to the main instead
Slight tweak to this main() thing. I think its better to have main return an int and then have sys.exit(main()) in the if name == main bit. That way the program behaves like a normal commandline program with an exit status. You can add in error handling and return a non-zero exit status easily this way.
I find that raise SystemExit(main()) is even more flexible for ease in writing end-to-end texts of the exception class exit int value that is still in python as compared to trying to write tests checking actual exit return code where you’d have to check the system after the in-test python has fully exited.
At 10:45 it is better to use Iterable from the typing module to annotate "elements". This way you could pass not only lists but also tuples or even generators.
@@DrDeuteron yes but the point of type hinting is to reduce the issues you have with duck typing. Duck typing is both a strength and a weakness of Python. And better type hints are more meaningful so it is encouraged to use type hints imported from the typing module.
@@kycklingmannen931 i agree, especially containers vs iterables, but what do you do for a function that takes floats, int, all the numpy versions, or arrays, matrix etc…is there super that covers all that?
In the case of making [p for p in people if len(p) > 7] more understandable inherently, I would opt for naming, such as 'n', 'name,' 'name_str' .. [name for name in people if len(name) > 7]. Great vid!
I think the best reason to make a main function is the fact you can push utility functions to the bottom of the file and leave the *main* code at the top, not needing to scroll down to see the what the actual main code is.
for list comprehensions: a lot of common patterns can be covered by `filter` and `map`. For example, instead of doing [p for p in people if len(p) > 7] you can do filter(lambda p: len(p) > 7, people) which I find quite a lot more readable because it's explicitly telling you that the only thing happening in this statement is we're getting rid of some values based on a predicate. Although, as a counterpoint, python's lambdas and higher order functions like `filter` and `map` can be somewhat verbose so the choice is not so straightforward.
list(itertools.compress(people, map((7.).__lt__, map(len, people))) or functools.partial(operator.lt_, 7). which I like because it tells people to stay tf out of your code.
Given Comprehensions can both "filter" and "map" in one statement, that is a real advantage over filter and map. Also map and filter produce generators which can be a Gotcha when newer programmers are expecting a list. Meanwhile comprehensions give you a choice of list, set, dict and also generator, using the same syntax with just a different bracket.
@@apmcd47 the thing about providing generators is a really annoying point imo that stops me from using filter and map so much. When filtering and mapping at the same time I agree it's probably better to use a comprehension
For the first two tips i would encourage to write real tests instead of "tests inside a module". Yes, it will take extra time, but will serve much better.
Some of these are definitely bad habits. Definitely much better to do it this way, by writing a unit test. Funny to mention "testing" but not use any built-in tests idiomatic to the language... It won't actually take more time since the execution environment will be more predictable and you avoid all the issues while having better design of code.
4:11 thanks for this. Structuring my code to have a main function that runs the others and an if statement that runs main is a great idea to clearly show what the program does.
i have to agree with most but id say sometimes doing one big function is the way to go. sometimes stuff is so intertwined and so very unlikely to get used anywhere else, that its just better to have it all in one place rather than split between a bunch of smaller functions. essentially write it the same though, by splitting it into smaller parts and denoting when some "step" has been achieved. if you ever do find a piece needs to be reused it isnt usually too hard to just pull that piece out into its own function later on
I was absolutely mind blown when you spoke about your type annotations. I've been watching MANY other RUclipsrs and not once seen this at all! This is something that I've been practicing since! I feel this should just absolutely be normal coding 'etiquette' for the lack of a better term. Thank you, and amazing job at explaining!
Just starting to learn python. That (-> None/ ->int) is amazing. I know it might not be amazing to others, but I am just learning the basics for now. I will try to watch all your videos.
A huge thanks to you guys who are sharing more useful habits and pointing out some of the benefits that I left out in the comment section. You help both me and many other developers to understand more about this beautiful programming world!
I've always loved list and generator comprehensions. I often each part on it's own line because I feel it becomes more readable. For example: [ p for p in people if len(p) > 7 ]
Came into the comments to add this. Comprehensions, whether list, dict, set, etc., are much more readable if on multiple lines and you can even add more conditions, though I would advise for moderation.
@@ValkoinenShiro I tend to be guilty of overdoing it; almost like I'm trying to win an award for fewest statements. There's a point at which a generator function is cleaner and more readable.
One additional tip for type hinting: You can still use them even if you need to maintain compatibility with older versions of Python. Instead of putting the type hints inline, you can add them to a "stub" file. It's a bit more work to build and maintain, but it gets the job done.
Cool video! These are great tips, but I want to point out that IMO there’s some nuance to be had with your “big function” tip, it’s important to recognize there’s costs that come with extracting code. Oftentimes you are never going to use that functionality again, extracting that code can actually make it more difficult to find certain logic just by increasing the sheer number of functions and lines of code, and it will inevitably take additional time and effort to do this extraction which can often be better used elsewhere. Reusable small functions are great, especially in large collaborative projects, but there are plenty of times where the overhead just isn’t worth it. Generally I find it isn’t worth it to abstract functionality until I actually need it somewhere else or if the function is unreadably long, and in that case you often also want to extract it into its own module
Definitely agree. I also think the example function was not "big," at only seven lines and two if statements. Breaking up big functions is often good, but I probably would have left this example function alone.
@@TheFwip But as a tutorial, it's good to plant the seed of the idea of abstraction into a beginner's mind so they're cognisant of it when it comes to coding larger functions. They may overuse it to begin with with trivial code, but with practice they will perfect that skill before the time they really need to use it.
A thing to mention about using the main function in your code is that anything you write directly in the if __name__ == "__main__": check will be part of the global namespace, while with main it will only be part of main's namespace, so it does have a reason beyond just a personal preference.
Comprehensions come in various flavours. There are also set { x for ...} and dict { kev: value for ...} comprehensions, as well as generator expressions ( x for ...) . The latter is interesting if performance matters and you know that you traverse the result at most once. Exceptionally good for loading tuple data into polars or pandas for example.
Tip: Instead of creating a specific is_bob function, try making an is_banned function, which receives a name and a blacklist, and returns if such name is in the blacklist
On the point of "Big Functions". I think that while the core idea of breaking up large functions like this into smaller reusable functions the statement certainly is correct "Make your code as reusable as possible" often has led me down a path of overengineering. I think its more important to think about things in the domain we are programming. Do we need to know if someone is bob in multiple places? Then maybe its a good idea to write a separate function for that and use it as you demonstrated
I've been following you for few months now, you provide amazing content, and I really appreciate your videos. Thank you so much. You are helping so many students😊
For simple scripts intended for command line use I do this: import sys def main(arg1, arg2): ... if __name__ == '__main__': main(*sys.argv[1:]) # The first argument is the script's name, skip it.
Thanks so much for this video. You're such a life saver. Coming from an environment of using c++, I always dreaded the time I would run into an error when I run my python program simply because my editor couldn't catch the errors when I wrote them. Now with type annotations and mypy, I can begin to truly appreciate the power and simplicity of python.
This is a good video and you absolutely should use type annotation. Type annotation will not help the user as stated in the video. The user most likely will be interacting with an app of some sort. Type annotation help developers who will use your code and help you, if you are using the function somewhere else in the code or reusing the function. Documentation will definitely help developers and in some instances can be used to create help for the end user.
I know this is just a quick example, but I'd argue a function like is_blacklisted or similar would be better than is_bob. You wouldn't want to add another function call for every person who is banned from the club.
I think the most important python habit I have is to write as little code as possible in python, because python is a programming language by mathematicians and nowadays data scientist for themselves, and not actual software developers.
imo, large functions are not bad. make functions as big as necessary without repeating yourself... if you notice patterns/checks that are used more than once throughout the code, only then should you extract them in separate function...
@@yuhongwang9052 how so? what fundamentally makes shorter functions easier to test? you still have to hit same number of branch paths/combinations if you were to split larger function into smaller ones.
Yes, but it's a lot more obvious when a function is smaller. And more importantly, when a smaller function fails, you know exactly what's wrong. This may not be the case if there are 2^13 (13 if statements) instead of 2^3
@@saltybaguette7683 how is it more obvious when function is smaller? you should name your test cases meaningful names and target only that case in that particular unit. you can still know exactly where your function is wrong because you see which test names failed and which passed for that function. it has nothing to do with function size.
Well using main is not just prettier / more java like, it allows you to call the module externally from elsewhere which you can't do with the if statement. So there's a practical reason to do it, assuming you do intend to allow it to be called.
@@philippk5446 Often there is a point, especially with scripts. Last week for example I wrote a simple script which sends a request to update a product. I called it update product and gave it a main. It's Django so the usual entrance is run() in terminal but I also call it from admin which runs main. Then I have another update products which loops through the products and updates each one. Point is, the same script is being called from numerous places - sometimes main, sometimes not - so, yes, there is very much a point.
Putting anything in ___main___ instead of main() is dangerous because variables in ___main___ will have a script-wide scope instead of function-wide, which can lead to surprizes. It's best to use ___main___ just to call main() unless global variables are intended.
I've never used type annotations because it never had any functional effect on the program, I've always worked in a team of one and I always just used vim, so had no syntax analysis. Also, I started with Python 2.6 and don't tend to catch up with new features. This is definitely something I need to work on, along with commenting my code more (or even at all).
These are not habits. They are called language idioms. When you learn a language and want to become proficient at programming within that language, learn the idioms of the language that make your code function within the specifications of the language.
I clicked on this out of curiosity (I don't work with Python), but holy moly, what font is that?! I don't know if it's the size, rendering or what, but it looks sooooo clean. I need that.
Love this video. For the past 2 years I don't know how many times I've been helping my friends, who learn Python, to find a bug. And at the end the bug was simply using str int instead of int str. Also list comprehensions are super cool. My friends for some reason love writing list(filter(lambda ..., map(lambda ..., data ))) when you could just [int(n, 2) for n in data if len(n) == 8]. Also splitting big functions into smaller can sometimes be replaced with list comprehensions too. For example when you need to change data type with some specific data editing. While function is reusable, sometimes its functionality is very specific and unless you are good at naming things, small functions will be littering the code instead of helping
The filter/map approach is nice to write when coming from other functional programming languages. You're right that they're not always as nice to read as list comprehensions, but in certain contexts they can make a lot of sense since they allow for greater re-useability than comprehensions. Say for example you already have a function that verifies someone is an adult like so: def is_adult(person): return person.age >= 18 And that you have a function that returns the full name of a person like so: def full_name(person): return f"{person.first_name} {person.last_name}" Then at some point you want to take a list of people, and get the full names of everyone who's an adult. With list comprehension, if you wanted to re-use your existing functionality you would write the following: adult_full_names = [full_name(person) for person in people if is_adult(person)] Whereas with map/filter since they already expect functions, you dont need to invoke them, just pass them as is: adult_full_names = map(full_name, filter(is_adult, people)) Obviously I left off the list() conversion in that, but in most cases with functional-style code you wouldn't be converting to lists very often, since leaving the data in generator form is much more efficient as each element can be streamed through multiple transformations without creating a bunch of intermediate lists. It allows you to write your code as if python was a lazy language like haskell, where values are only calculated when they're actually used/required. So basically I would use map/filter if I already have functions for the transformation I'll be doing, but if I would be having to write a lambda inside the map/filter, then I would instead use a list comprehension (or more likely a generator comprehension, again for performance reasons).
I was thinking about filter while I was writing the list comprehension, but the instant benefit of using the list comprehension for me was that I could write it out in a split second, while with filter I always need to hover over the function to see the little pop-up with the documentation to make sure I'm passing in things in the correct order.
@@LittleLily_ That is a very good point. In my example I pointed on using map() and filter() with lambda-functions. Usually those lambda-functions make code hard to read. I thing using existing functions is very clean Also I was talking about people, who is new at Python. So usually I can see my friends making map -> filter -> ... -> filter combinations. In such cases even using existing functions does not save readability. My point was that beginners sometimes make overcomplicated map/filter combinations that can be replaced with a single much more readable list comprehension
@@Indently that's fair enough. the order of the arguments for higher order functions in any language generally always has the function first and data second, as that way you can more easily partially apply the function, for example with python you might write this: from functools import partial def full_name(person): return f"{person.first_name} {person.last_name}" people_to_full_names = partial(map, full_name) And now you have a reusable function that can operate on any list of people to return the full name of all of them like so: full_names_a = people_to_full_names(people_a) full_names_b = people_to_full_names(people_b)
Type annotations saved Python for me. I apparently really rely on type annotations in more complicated code. I was doing type definitions in comments until realized it was an option in the language.
Thanks so much for the great video on good Python habits. I know you put a lot of time and effort into it, and it shows. The video is well done and easy to understand.
One of the great things about Python, is you can write code without a lot of boilerplate. I'm certainly not going to add boilerplate for someone else's idea of "clean code".
When you are making programs that are 1000+ lines of code, you'll realize how much simpler debugging would have been had you added type hinting / doc strings and breaking up functions into smaller ones. Granted his example of breaking up the code into separate functions was rather extreme, but he was just using it as an example. i don't necessarily agree with him on using type hinting when it's not part of the function parameters though.
@@resresres1 Definitely. You write a function which expected certain functions from an object, the type hinting gives you a hint that the duck typing will work (or not if you violate it) as intended. It doesn't matter whether the code is interpreted or compiled, it would achieve similar protections as you'd see in a language with a type system and type safety.
Question: how can i do a type annotation for a function passed as a parameter of annother funcion? Ex: def do_this(arg: float): return blablabla with arg def do_that(arg: float): return blobloblo with arg def what_to_do(arg: float, func: >Type that i wanna know
As someone who learned C before python, I immediately looked up if I could create an "int main()" equivalent, thus I always used the if __name__ == '__main__' out of habit and preference
You don't need to annotate variables that are declared unless you have good reason (either it is mutable by an end user, or the type cannot otherwise be inferred). It's a waste of time to annotate everything like that. Even mypy with --strict doesn't care.
Regarding Big Functions. It sounds really similar to the argument that you should write all your small functions directly in cython. That way even if performance is not a concern right now those functions can be reused in potential other future parts of the code where performance might matter making your code more reusable. To that I would say: stop it, get some help. Just write the code and after it works if you see you have a performance bottleneck or a good reason to split off functions (code duplication or wanting to test only a part of a function) than do the work instead of always doing it prematurely. Stop optimizing for potential future scenarios at the cost of the present.
My only recommendation is that you plan your code before you write your code. Living in the moment is only advised if your program is as big as printing "hello, world".
16:50 - true. You could use Ruby instead to have nicer code: long_names = people.select{|p| p.length > 7} and if you need them in upper case too: long_names = people.select{|p| p.length > 7}.collect{|p| p.upcase}
similar to how you'd write it in JS/TS: longNames = people.filter(p => p.length > 7) longNames = people.filter(p => p.length > 7).map(p => p.toUpperCase()) the lambda syntax in python isn't quite as nice, so I think that's why most people don't like using stuff like map/filter over list comprehensions in python.
The problem with list comprehensions is not the syntax but how you layout the comprehension. I prefer to put everything on a new line as it makes it more readable: long_names = [ p.upper() # list element for p in people # iteration if len(p) > 7 # condition ] and it is more readable then using filter and map.
12:20 "someone will say it's obvious. But in programming obvious is never obvious enough." It's not about "obvious". It is about _explicit._ It must be explicit in the code, so that other programmers can easily understand what the code does, what's its intention. But also it must be explicit so that your tools can work with it, like PyCharm here. Otherwise you need ChatGPT to look at your code and guess what's meant by "upper" in the function name. Is it a verb "convert to upper" or is it a test for upper? Good code explicitly says what it does.
Obvious is a common synonym in the English language, you're more than welcome to be pedantic and picky about words, but it's a lot of typing gone in the wind :)
@@Indently Well, explaining something like this to a non-technical person might make you sound... full of yourself, if you say obvious often. It's not obvious to everybody!
You can overdo the whole "small functions" thing and I think this is actually a perfect example of when not to do that. Unless you reuse the functionality elsewhere, don't extract the logic into it's own function bc it's much easier to read code that hasn't been spread out all over the codebase for no good reason. If you must have a predicate function with a descriptive name, just make a lambda. "You never know" is a good way to pollute the codebase with functions that are called in like one place. Plus, naming is hard and the more functions you create, the more likely it is that you have semantic inaccuracies in the names.
I agree. I once came in behind someone who wrote about 5 functions for a config file parser, including a function to return each line of the file and a function to find '#' at the beginning of a line. It took about 60 lines of code. (Admittedly, some of those were nonsense like i = 0 for i in range(len(line)) and so on.) I was able to replace it with about 12 lines all in one place.
Nice video! Can you make a video about pypi? It has become difficult and confusing to publish a python package, it would be appreciated if you make a tuto about it! Thank you!!
For number 1. if __name__ == __main__ is ment to be used in modules that should be run not for testing porpouses. You should test your code with unittest in separate module
Yes, in an ideal world where you also always have all the time you need to make perfect fully documented code. Here in the real world we need to prioritize stuff. So for code that is most likely only being used by myself, and maybe a few collegues, for a few tasks. I always put testing in the 'if main' section. The test cases in there also function as a minimum documentation on the intended use of that module. I do this because here in the real world, insisting on full fledged unit test and other "good practices", only means that test and doc is skipped entirely for such low use code. Sometimes such code do end up being more useful, and more used. And needs to be upgraded to production ready code (unit test etc) Ask your self "would I prefer to upgrade a module with test cases in the 'if main', or a module where test was done on command line on some unknown pc?"
@@martin_hansen I disagree wih you. if __name__ == "__main__" statement indicates that this module is ment to be used. And about "real world". I have only 3+ years of proffessional experience with python but in each project (but he last one) we had tests. lots of test. In fact there was much more code for test than for real functionality. It may look like unnecessary work but it is very helpfull and much needed in large projects. On the other hand, I'm working with quite small project without any tests and it is nightmare. PS. I understand that with your own project you do not unittest (I dont myself) because delivering functionality is more fun :)
I have no argument against people who extensively test their code, you guys are the true heroes who really go out of your way to make your apps reliable, and I respect it. For some of us though, when we are building modules, we tend to test functionality as we write it. The complete opposite of TDD. Forgetting those function calls in a module you created can be a headache if you don't use the "if name is equal to main" check.
If _name__ == "__main__" foo() Does NOT indicate module is ment to be run. It is the exact opposite. The whole reason of adding this condition is to shield the call to foo() from being called when it is imported by another module. If the module is intended to just be run and not imported, you should just put foo() So the 'if _name__ == "__main__"' statement tells me that this module was intended to be imported but can also be run.
I understand what you're thinking Martin, but if you add an "if name is equal to main check" anywhere in your code, it will tell anyone who sees the script that something was meant to be run directly. Whether it's just for personal testing purposes, or to prevent it from being executed when being imported. It tells whoever is reading it, that there is functionality that can be executed directly in that script. I would never run "foo()" out in the open like that. You will have to track that call down later if you forget about it when you are importing the module that contains it.
Some good tips there, many thanks! I always wondered about the if __name__ == ‘__main__’: line, what was the real advantage in doing that, but you explained it really well.
@@daveys no. The import statement executes all the code, which is why you do have functions at all, because it executes all the def-statements too. This all happens before you do call one of those functions. If the file is meant to be imported, it makes no sense at all to have code in the main body of file which you don't want to run. So with if __name__ ... you prevent code from being run which should not be there at all.
I've yet to see a useful example of writing a variable type annotation. It is useful to write input and output type annotations for functions, but for variable it can be inferred and you don't want it to fail if you change the value - subsequent function calls will fail if needed, so everything is good
Everything was solid up to the p for p in people I don't use any single character variables anymore it doesn't help with readability and I'm not trying to minify everything if the goal is to make it readable. I would just use a list comprehension when absolutely necessary.
The data types is one reason why I dislike Python. But now that you have shown me type annotations, I will probably start trusting my code and what python does with it. It sounds silly, but it makes me more comfortable
I first learnt Python in version 2. I didn’t know half of these things were available now because they weren’t in Python 2. I did try to learn the differences when v 3 came out, but I think a lot of these were added after the first version of 3? Anyway thanks for broadening my horizons.
I really like the "big functions" part but I have a few questions: 1. how do I tell the difference between more "top-layer" functions and these small mini functions that just make things more abstract? i mean, in terms of naming, often you would want to be bale to have a good hierarchy of this. 2. I'm not sure if this example is the best because one also needs to understand at which point should this separation happen. In practise the example probably should stay as it is, so that the code doesn't become more complex and hard to read.
Def main() is a must. Makes the code more readable and easier to use. Can help you seperate your functions in a nice and neat order. Great video overall.
Also makes it easier to call functions and kinda follow the order in which makes sense. Helps you walk through and see where variables are going and how they are affected.
"So, here I'm going to type in long_names, and that's going to shadow the name above, but... Who cares for now? And, actually, apparently, MyPy cares a lot ^^`" Literally the best part of the video
A huge thanks to you guys who are sharing more useful habits and pointing out some of the benefits that I left out in the comment section. You help both me and many other developers to understand more about this beautiful programming world!
I'm super appreciative when you guys share cool information :)
after hour and hours and trainings, someone explains in 3 seconds the role of if __name__ = ''__main__' ,omg, thx a lot !
100000%
if I get a script w/o it, I freak out.
It’s good practice (like he said), a lot of beginners/beginner orientated videos don’t use it though, because either they’re not writing code big enough for it to matter (most beginners use a single file), or for the beginner orientated videos they just pass over it because it’s not really important for the scope of the tutorial, however I believe its function should be taught a bit more than it is…
Still don’t really understand so it’s like if the module is directly imported from the main file then run?
@@dt3688 Sorta. So, the purpose of the if statement is to explicitly check if your file is run as main (When ran directly the interpreter sets the variable __name__ to “__main__”, but when imported it’ll set __name__ to whatever your filename is.)
In doing so, whatever you have inside that if __name__ == __main__ statement will be ran ONLY if you directly run that file/script. Otherwise, if you import the file, whatever you have inside the if statement will not be run.
This is useful (mostly) for test code, or if your file is meant to run as both an import and a script on its own. This is to keep whatever imports your script from accidentally running the functions in the file. Ex:
def PrintSomething():
print(“This will run”)
if __name__ == __main__:
PrintSomething()
In this code, it will only print the string to your console if ran as a script directly. If you import it, PrintSomething() will not run.
(Formatting may have broken for youtube with the underscores for name and main, but yk what I meant)
Another important thing with main() is the scope. Code run in "if name main" uses global scope and can/will pollute it. Running code in main has the function's scope :)
I would argue that this is _the_ most important reason to use a main() function. It's not just a way to make your code look better organised, it has fundamentally different semantics from just putting code in a top-level if-name-is-main
True, but you should never ever run a main function AND THEN run extra code in the global scope later, so namespace pollution doesn’t rlly matter.
I didn't think about that, the arguments in the video didn't convince me at all, but this is a really good reason to use a main funciton.
What do you mean by "pollution"?
@@AmodeusR I think he means having a bunch of variables exposed to the global scope is polluting it, and can be problematic when importing a module.
Hats off to my main man Bob, he did nothing wrong.
:3
A main(with parameters) is even more useful when testing. Allow the if __name__ block to handle command line arguments and then pass sanitized versions of those to your main. Then your tests can just call main directly and test various scenarios.
I'm just learning python, but if you were to do this, how would you suggest you test your input parsing and sanitation itself? I'm thinking I'd have the '__name__' block call a 'parse_and_sanitise_inputs' function to do that (and pass its results to main), which could then be tested separately.
@@mnxs you would have separate files that use those functions to run some tests that test they do their job. You don't need a dedicated "parse_and_sanitise_inputs" function, for example if you want to check that you provided a correct number within a certain range, such as an age, then you would do a "parse_age" or similar function, that would check if age > 0 and is a number and return that number.
I would use something like arg_parser in python
Just one thing about list comprehensions: they can very easily become very hard to read, so I'd say "don't overuse them"
A way to avoid them becoming unreadable without resorting to loops is to split the transformations into multiple separate list comprehensions on new lines, and assigning the result of each line to a variable with a descriptive name to show what the intermediate state represents. Ideally you would use generator comprehensions as well for the intermediary steps and only convert to list at the end (if it's even needed in list form at all).
One major example I've seen, unless you've got a very descriptive comment attached or it's doing a super simple operation, don't put two for loops in one list comp
Exactly!
It depends how you use them. Variable Naming has a huge impact in readability as well.
For example: taking the person names > 7 case, if we were to rewrite with more descriptive names it can convey a sense of reality and make the code more relatable to the readers.
listLongNames = [ pp for pp in people if len(pp)>7 ]
See, makes so much difference.
["["".join(letter) for letter in word if letter.isalpha()] for word in "why would you say that?".split()
]
don't use "..." for a placeholder, use "pass".
Whats the advantage of pass vs …?
@@NathanSMS26 I think pass is more visually obvious than the ellipse, but if you prefer the ellipse, then enjoy :)
Actually, the best practice is to throw a not implemented error so you don't forget about it.. Honestly though, it's too much effort. I always thought python should have a keyword for this, something nice and short that throws an exception.
@@davidmurphy563 "assert false"
@@Zaniahiononzenbei Oh, that's a good one! I'm going to use that!
For number 2, another bonus is if you declare a main function you can expose it in an init or import it. One example is importing when testing.
Another bonus is containing the variables. You won't accidentally use a globally-scoped name in another function, if that name is local to the main instead
Thanks for sharing the benefits guys! I often forget to mention simple things like these when creating the videos as a one man team.
It's similarly useful if you're using asyncio.
Slight tweak to this main() thing. I think its better to have main return an int and then have sys.exit(main()) in the if name == main bit. That way the program behaves like a normal commandline program with an exit status. You can add in error handling and return a non-zero exit status easily this way.
That's a useful tip if you're building a more complex script, but probably a bit outside what I'd put in a boilerplate.
I find that raise SystemExit(main()) is even more flexible for ease in writing end-to-end texts of the exception class exit int value that is still in python as compared to trying to write tests checking actual exit return code where you’d have to check the system after the in-test python has fully exited.
At 10:45 it is better to use Iterable from the typing module to annotate "elements". This way you could pass not only lists but also tuples or even generators.
it's called duck typing.
@@DrDeuteron yes but the point of type hinting is to reduce the issues you have with duck typing. Duck typing is both a strength and a weakness of Python. And better type hints are more meaningful so it is encouraged to use type hints imported from the typing module.
@@kycklingmannen931 i agree, especially containers vs iterables, but what do you do for a function that takes floats, int, all the numpy versions, or arrays, matrix etc…is there super that covers all that?
In the case of making [p for p in people if len(p) > 7] more understandable inherently, I would opt for naming, such as 'n', 'name,' 'name_str' .. [name for name in people if len(name) > 7]. Great vid!
I think the best reason to make a main function is the fact you can push utility functions to the bottom of the file and leave the *main* code at the top, not needing to scroll down to see the what the actual main code is.
for list comprehensions: a lot of common patterns can be covered by `filter` and `map`. For example, instead of doing [p for p in people if len(p) > 7] you can do filter(lambda p: len(p) > 7, people) which I find quite a lot more readable because it's explicitly telling you that the only thing happening in this statement is we're getting rid of some values based on a predicate.
Although, as a counterpoint, python's lambdas and higher order functions like `filter` and `map` can be somewhat verbose so the choice is not so straightforward.
list(itertools.compress(people, map((7.).__lt__, map(len, people)))
or
functools.partial(operator.lt_, 7).
which I like because it tells people to stay tf out of your code.
@@DrDeuteron ahahahahhaha
@@DrDeuteronlol
Given Comprehensions can both "filter" and "map" in one statement, that is a real advantage over filter and map. Also map and filter produce generators which can be a Gotcha when newer programmers are expecting a list. Meanwhile comprehensions give you a choice of list, set, dict and also generator, using the same syntax with just a different bracket.
@@apmcd47 the thing about providing generators is a really annoying point imo that stops me from using filter and map so much. When filtering and mapping at the same time I agree it's probably better to use a comprehension
For the first two tips i would encourage to write real tests instead of "tests inside a module". Yes, it will take extra time, but will serve much better.
how about a video on unittest or the current most bestest argparse?
Some of these are definitely bad habits.
Definitely much better to do it this way, by writing a unit test. Funny to mention "testing" but not use any built-in tests idiomatic to the language... It won't actually take more time since the execution environment will be more predictable and you avoid all the issues while having better design of code.
4:11 thanks for this. Structuring my code to have a main function that runs the others and an if statement that runs main is a great idea to clearly show what the program does.
i have to agree with most but id say sometimes doing one big function is the way to go. sometimes stuff is so intertwined and so very unlikely to get used anywhere else, that its just better to have it all in one place rather than split between a bunch of smaller functions. essentially write it the same though, by splitting it into smaller parts and denoting when some "step" has been achieved. if you ever do find a piece needs to be reused it isnt usually too hard to just pull that piece out into its own function later on
Super appreciate this. Need to include type hinting into my coding. Your explanations make sense.
This came at PRECISELY the right time for the type annotations stuff 🎉 big thanks
I was absolutely mind blown when you spoke about your type annotations. I've been watching MANY other RUclipsrs and not once seen this at all! This is something that I've been practicing since! I feel this should just absolutely be normal coding 'etiquette' for the lack of a better term. Thank you, and amazing job at explaining!
OMG!!! He said Java!! I leave!!!
Just starting to learn python. That (-> None/ ->int) is amazing. I know it might not be amazing to others, but I am just learning the basics for now. I will try to watch all your videos.
A huge thanks to you guys who are sharing more useful habits and pointing out some of the benefits that I left out in the comment section. You help both me and many other developers to understand more about this beautiful programming world!
I've always loved list and generator comprehensions. I often each part on it's own line because I feel it becomes more readable. For example:
[
p
for p in people
if len(p) > 7
]
Came into the comments to add this. Comprehensions, whether list, dict, set, etc., are much more readable if on multiple lines and you can even add more conditions, though I would advise for moderation.
@@ValkoinenShiro I tend to be guilty of overdoing it; almost like I'm trying to win an award for fewest statements. There's a point at which a generator function is cleaner and more readable.
@@TheJaguar1983 "clever" is not a compliment in python
One additional tip for type hinting: You can still use them even if you need to maintain compatibility with older versions of Python. Instead of putting the type hints inline, you can add them to a "stub" file. It's a bit more work to build and maintain, but it gets the job done.
Cool video! These are great tips, but I want to point out that IMO there’s some nuance to be had with your “big function” tip, it’s important to recognize there’s costs that come with extracting code. Oftentimes you are never going to use that functionality again, extracting that code can actually make it more difficult to find certain logic just by increasing the sheer number of functions and lines of code, and it will inevitably take additional time and effort to do this extraction which can often be better used elsewhere.
Reusable small functions are great, especially in large collaborative projects, but there are plenty of times where the overhead just isn’t worth it.
Generally I find it isn’t worth it to abstract functionality until I actually need it somewhere else or if the function is unreadably long, and in that case you often also want to extract it into its own module
Definitely agree. I also think the example function was not "big," at only seven lines and two if statements.
Breaking up big functions is often good, but I probably would have left this example function alone.
As someone who has had to comb through dozens of 3-5 line functions to understand what a script is doing, I agree.
@@TheFwip But as a tutorial, it's good to plant the seed of the idea of abstraction into a beginner's mind so they're cognisant of it when it comes to coding larger functions. They may overuse it to begin with with trivial code, but with practice they will perfect that skill before the time they really need to use it.
A thing to mention about using the main function in your code is that anything you write directly in the if __name__ == "__main__": check will be part of the global namespace, while with main it will only be part of main's namespace, so it does have a reason beyond just a personal preference.
The best __name__ = 'main' explanation i've ever seen and im writing this comment just 2 minutes in the video.
Today I encountered the first problem while I was learning Python.
Surprised to see this video popped up to my scroll and I’m glad I watched this.
As a beginner I found these habits very helpful thanks for providing
Comprehensions come in various flavours. There are also set { x for ...} and dict { kev: value for ...} comprehensions, as well as generator expressions ( x for ...) .
The latter is interesting if performance matters and you know that you traverse the result at most once. Exceptionally good for loading tuple data into polars or pandas for example.
Actually incredible top tier tips, 10/10, would watch again.
Tip: Instead of creating a specific is_bob function, try making an is_banned function, which receives a name and a blacklist, and returns if such name is in the blacklist
On the point of "Big Functions". I think that while the core idea of breaking up large functions like this into smaller reusable functions the statement certainly is correct "Make your code as reusable as possible" often has led me down a path of overengineering. I think its more important to think about things in the domain we are programming. Do we need to know if someone is bob in multiple places? Then maybe its a good idea to write a separate function for that and use it as you demonstrated
The point of 2 isn't a "feeling" or anything like that. It's: a) testable, and b) importable from another package.
Another great video with great examples. Thank you. Mypy, who knew?
9:50 before you use a function, you should know, what it does. And you can insert every type which provides an upper-function and not only strings.
I've been following you for few months now, you provide amazing content, and I really appreciate your videos. Thank you so much. You are helping so many students😊
For simple scripts intended for command line use I do this:
import sys
def main(arg1, arg2):
...
if __name__ == '__main__':
main(*sys.argv[1:]) # The first argument is the script's name, skip it.
Thanks so much for this video. You're such a life saver. Coming from an environment of using c++, I always dreaded the time I would run into an error when I run my python program simply because my editor couldn't catch the errors when I wrote them. Now with type annotations and mypy, I can begin to truly appreciate the power and simplicity of python.
I'm incredibly happy to hear you enjoy the type annotations :)
This is a good video and you absolutely should use type annotation. Type annotation will not help the user as stated in the video. The user most likely will be interacting with an app of some sort. Type annotation help developers who will use your code and help you, if you are using the function somewhere else in the code or reusing the function. Documentation will definitely help developers and in some instances can be used to create help for the end user.
You should use `List` from `typing`, insted of `list` to make linters work well.
I know this is just a quick example, but I'd argue a function like is_blacklisted or similar would be better than is_bob. You wouldn't want to add another function call for every person who is banned from the club.
I think the most important python habit I have is to write as little code as possible in python, because python is a programming language by mathematicians and nowadays data scientist for themselves, and not actual software developers.
imo, large functions are not bad. make functions as big as necessary without repeating yourself... if you notice patterns/checks that are used more than once throughout the code, only then should you extract them in separate function...
IMO, large functions are hard to unit test.
@@yuhongwang9052 how so? what fundamentally makes shorter functions easier to test? you still have to hit same number of branch paths/combinations if you were to split larger function into smaller ones.
Yes, but it's a lot more obvious when a function is smaller.
And more importantly, when a smaller function fails, you know exactly what's wrong. This may not be the case if there are 2^13 (13 if statements) instead of 2^3
@@saltybaguette7683 how is it more obvious when function is smaller? you should name your test cases meaningful names and target only that case in that particular unit. you can still know exactly where your function is wrong because you see which test names failed and which passed for that function. it has nothing to do with function size.
I really like how you explain everything, it’s very clear. Thanks a lot
Well using main is not just prettier / more java like, it allows you to call the module externally from elsewhere which you can't do with the if statement. So there's a practical reason to do it, assuming you do intend to allow it to be called.
yeah but there is no point in calling a main from somewhere else! If you need to do that, the function you wrote is not a main.
@@philippk5446 Often there is a point, especially with scripts. Last week for example I wrote a simple script which sends a request to update a product. I called it update product and gave it a main. It's Django so the usual entrance is run() in terminal but I also call it from admin which runs main. Then I have another update products which loops through the products and updates each one.
Point is, the same script is being called from numerous places - sometimes main, sometimes not - so, yes, there is very much a point.
Putting anything in ___main___ instead of main() is dangerous because variables in ___main___ will have a script-wide scope instead of function-wide, which can lead to surprizes. It's best to use ___main___ just to call main() unless global variables are intended.
I've never used type annotations because it never had any functional effect on the program, I've always worked in a team of one and I always just used vim, so had no syntax analysis. Also, I started with Python 2.6 and don't tend to catch up with new features. This is definitely something I need to work on, along with commenting my code more (or even at all).
you're really helpful man, I'll recommend your videos to my friends
These are not habits. They are called language idioms. When you learn a language and want to become proficient at programming within that language, learn the idioms of the language that make your code function within the specifications of the language.
I clicked on this out of curiosity (I don't work with Python), but holy moly, what font is that?! I don't know if it's the size, rendering or what, but it looks sooooo clean. I need that.
Love this video. For the past 2 years I don't know how many times I've been helping my friends, who learn Python, to find a bug. And at the end the bug was simply using str int instead of int str. Also list comprehensions are super cool. My friends for some reason love writing list(filter(lambda ..., map(lambda ..., data ))) when you could just [int(n, 2) for n in data if len(n) == 8]. Also splitting big functions into smaller can sometimes be replaced with list comprehensions too. For example when you need to change data type with some specific data editing. While function is reusable, sometimes its functionality is very specific and unless you are good at naming things, small functions will be littering the code instead of helping
The filter/map approach is nice to write when coming from other functional programming languages. You're right that they're not always as nice to read as list comprehensions, but in certain contexts they can make a lot of sense since they allow for greater re-useability than comprehensions.
Say for example you already have a function that verifies someone is an adult like so:
def is_adult(person):
return person.age >= 18
And that you have a function that returns the full name of a person like so:
def full_name(person):
return f"{person.first_name} {person.last_name}"
Then at some point you want to take a list of people, and get the full names of everyone who's an adult. With list comprehension, if you wanted to re-use your existing functionality you would write the following:
adult_full_names = [full_name(person) for person in people if is_adult(person)]
Whereas with map/filter since they already expect functions, you dont need to invoke them, just pass them as is:
adult_full_names = map(full_name, filter(is_adult, people))
Obviously I left off the list() conversion in that, but in most cases with functional-style code you wouldn't be converting to lists very often, since leaving the data in generator form is much more efficient as each element can be streamed through multiple transformations without creating a bunch of intermediate lists. It allows you to write your code as if python was a lazy language like haskell, where values are only calculated when they're actually used/required.
So basically I would use map/filter if I already have functions for the transformation I'll be doing, but if I would be having to write a lambda inside the map/filter, then I would instead use a list comprehension (or more likely a generator comprehension, again for performance reasons).
I was thinking about filter while I was writing the list comprehension, but the instant benefit of using the list comprehension for me was that I could write it out in a split second, while with filter I always need to hover over the function to see the little pop-up with the documentation to make sure I'm passing in things in the correct order.
@@LittleLily_ That is a very good point. In my example I pointed on using map() and filter() with lambda-functions. Usually those lambda-functions make code hard to read. I thing using existing functions is very clean
Also I was talking about people, who is new at Python. So usually I can see my friends making map -> filter -> ... -> filter combinations. In such cases even using existing functions does not save readability. My point was that beginners sometimes make overcomplicated map/filter combinations that can be replaced with a single much more readable list comprehension
@@Indently that's fair enough. the order of the arguments for higher order functions in any language generally always has the function first and data second, as that way you can more easily partially apply the function, for example with python you might write this:
from functools import partial
def full_name(person):
return f"{person.first_name} {person.last_name}"
people_to_full_names = partial(map, full_name)
And now you have a reusable function that can operate on any list of people to return the full name of all of them like so:
full_names_a = people_to_full_names(people_a)
full_names_b = people_to_full_names(people_b)
@@cosy_sweater "pointed" --> "punted" maybe?
I’m glad! Your videos become my scripts raise to another level
Very nicely put together, thanks!
Type annotations saved Python for me. I apparently really rely on type annotations in more complicated code. I was doing type definitions in comments until realized it was an option in the language.
3:08 I love programming in Java man it only made me wanna watch more!
having a psvm alike functionality in python is nice, cool tip :)
Thanks so much for the great video on good Python habits. I know you put a lot of time and effort into it, and it shows. The video is well done and easy to understand.
One of the great things about Python, is you can write code without a lot of boilerplate. I'm certainly not going to add boilerplate for someone else's idea of "clean code".
When you are making programs that are 1000+ lines of code, you'll realize how much simpler debugging would have been had you added type hinting / doc strings and breaking up functions into smaller ones. Granted his example of breaking up the code into separate functions was rather extreme, but he was just using it as an example. i don't necessarily agree with him on using type hinting when it's not part of the function parameters though.
@@resresres1 Definitely. You write a function which expected certain functions from an object, the type hinting gives you a hint that the duck typing will work (or not if you violate it) as intended. It doesn't matter whether the code is interpreted or compiled, it would achieve similar protections as you'd see in a language with a type system and type safety.
Question: how can i do a type annotation for a function passed as a parameter of annother funcion? Ex:
def do_this(arg: float): return blablabla with arg
def do_that(arg: float): return blobloblo with arg
def what_to_do(arg: float, func: >Type that i wanna know
You'll need to import a special type from the typing module.
from typing import Callable
def what_to_do(arg: float, func: Callable):
I do appreciate list comprehension
As someone who learned C before python, I immediately looked up if I could create an "int main()" equivalent, thus I always used the if __name__ == '__main__' out of habit and preference
Coming from a C++ world learning Python, your main() and and type definitions just makes me feel so much more secure.
You don't need to annotate variables that are declared unless you have good reason (either it is mutable by an end user, or the type cannot otherwise be inferred). It's a waste of time to annotate everything like that. Even mypy with --strict doesn't care.
Regarding Big Functions. It sounds really similar to the argument that you should write all your small functions directly in cython. That way even if performance is not a concern right now those functions can be reused in potential other future parts of the code where performance might matter making your code more reusable.
To that I would say: stop it, get some help.
Just write the code and after it works if you see you have a performance bottleneck or a good reason to split off functions (code duplication or wanting to test only a part of a function) than do the work instead of always doing it prematurely.
Stop optimizing for potential future scenarios at the cost of the present.
My only recommendation is that you plan your code before you write your code. Living in the moment is only advised if your program is as big as printing "hello, world".
Very good tips, thank you. It would be useful to copy/paste some of the code demonstrated, as I'm slow/bad at typing. :-)
the 4th tip is pretty useful, thank you
16:50 - true. You could use Ruby instead to have nicer code:
long_names = people.select{|p| p.length > 7}
and if you need them in upper case too:
long_names = people.select{|p| p.length > 7}.collect{|p| p.upcase}
you could also use filter an map in py, iirc
similar to how you'd write it in JS/TS:
longNames = people.filter(p => p.length > 7)
longNames = people.filter(p => p.length > 7).map(p => p.toUpperCase())
the lambda syntax in python isn't quite as nice, so I think that's why most people don't like using stuff like map/filter over list comprehensions in python.
The problem with list comprehensions is not the syntax but how you layout the comprehension. I prefer to put everything on a new line as it makes it more readable:
long_names = [
p.upper() # list element
for p in people # iteration
if len(p) > 7 # condition
]
and it is more readable then using filter and map.
@@LittleLily_ why does "p =>" appear in every call signature? what else is there? seems like redundancy...Guido don't like.
@@DrDeuteron Its no different to the |p| in the ruby example, it's just how you define an anonymous function with an input argument named p
Awesome explanation!
12:20 "someone will say it's obvious. But in programming obvious is never obvious enough." It's not about "obvious". It is about _explicit._ It must be explicit in the code, so that other programmers can easily understand what the code does, what's its intention. But also it must be explicit so that your tools can work with it, like PyCharm here. Otherwise you need ChatGPT to look at your code and guess what's meant by "upper" in the function name. Is it a verb "convert to upper" or is it a test for upper? Good code explicitly says what it does.
Obvious is a common synonym in the English language, you're more than welcome to be pedantic and picky about words, but it's a lot of typing gone in the wind :)
@@Indently Well, explaining something like this to a non-technical person might make you sound... full of yourself, if you say obvious often. It's not obvious to everybody!
In the past 4 years of teaching, i have seen the most ridiculous code in the universe
This line is sool crazy😂😂
Valuable work! thanks.
You can overdo the whole "small functions" thing and I think this is actually a perfect example of when not to do that. Unless you reuse the functionality elsewhere, don't extract the logic into it's own function bc it's much easier to read code that hasn't been spread out all over the codebase for no good reason. If you must have a predicate function with a descriptive name, just make a lambda. "You never know" is a good way to pollute the codebase with functions that are called in like one place. Plus, naming is hard and the more functions you create, the more likely it is that you have semantic inaccuracies in the names.
I agree. I once came in behind someone who wrote about 5 functions for a config file parser, including a function to return each line of the file and a function to find '#' at the beginning of a line. It took about 60 lines of code. (Admittedly, some of those were nonsense like
i = 0
for i in range(len(line))
and so on.)
I was able to replace it with about 12 lines all in one place.
What is the purpose of -> None?
Indicates that a function does not return a value. Similar to creating a void function in other languages
Excellent video @Indently. Keep going
We can use filter function for the last one too
long_names : list[string] = filter ( lambda name : len(name) > 7 , names)
great habits! Thanks!
Love your videos :)
Nice video! Can you make a video about pypi? It has become difficult and confusing to publish a python package, it would be appreciated if you make a tuto about it! Thank you!!
For number 1. if __name__ == __main__ is ment to be used in modules that should be run not for testing porpouses. You should test your code with unittest in separate module
Yes, in an ideal world where you also always have all the time you need to make perfect fully documented code.
Here in the real world we need to prioritize stuff. So for code that is most likely only being used by myself, and maybe a few collegues, for a few tasks.
I always put testing in the 'if main' section. The test cases in there also function as a minimum documentation on the intended use of that module.
I do this because here in the real world, insisting on full fledged unit test and other "good practices", only means that test and doc is skipped entirely for such low use code.
Sometimes such code do end up being more useful, and more used. And needs to be upgraded to production ready code (unit test etc) Ask your self "would I prefer to upgrade a module with test cases in the 'if main', or a module where test was done on command line on some unknown pc?"
@@martin_hansen I disagree wih you. if __name__ == "__main__" statement indicates that this module is ment to be used.
And about "real world". I have only 3+ years of proffessional experience with python but in each project (but he last one) we had tests. lots of test. In fact there was much more code for test than for real functionality. It may look like unnecessary work but it is very helpfull and much needed in large projects. On the other hand, I'm working with quite small project without any tests and it is nightmare.
PS. I understand that with your own project you do not unittest (I dont myself) because delivering functionality is more fun :)
I have no argument against people who extensively test their code, you guys are the true heroes who really go out of your way to make your apps reliable, and I respect it.
For some of us though, when we are building modules, we tend to test functionality as we write it. The complete opposite of TDD. Forgetting those function calls in a module you created can be a headache if you don't use the "if name is equal to main" check.
If _name__ == "__main__"
foo()
Does NOT indicate module is ment to be run. It is the exact opposite.
The whole reason of adding this condition is to shield the call to foo() from being called when it is imported by another module.
If the module is intended to just be run and not imported, you should just put
foo()
So the 'if _name__ == "__main__"' statement tells me that this module was intended to be imported but can also be run.
I understand what you're thinking Martin, but if you add an "if name is equal to main check" anywhere in your code, it will tell anyone who sees the script that something was meant to be run directly. Whether it's just for personal testing purposes, or to prevent it from being executed when being imported. It tells whoever is reading it, that there is functionality that can be executed directly in that script.
I would never run "foo()" out in the open like that. You will have to track that call down later if you forget about it when you are importing the module that contains it.
Holy cow, didnt know about type annotations!
Also if you have the exact use case with the long_names, just use filter()
I was about to leave the video when you mentioned Java, but then you asked, 'Please don't leave the video because I said Java,' so I stayed 😆
Some good tips there, many thanks!
I always wondered about the if __name__ == ‘__main__’: line, what was the real advantage in doing that, but you explained it really well.
The advantage is to support bad habits.
@@ThomasVWorm - as far as I now understand it, if you call the function from another program, it’ll not run anything in the __main__ section.
@@daveys no. The import statement executes all the code, which is why you do have functions at all, because it executes all the def-statements too.
This all happens before you do call one of those functions.
If the file is meant to be imported, it makes no sense at all to have code in the main body of file which you don't want to run.
So with if __name__ ... you prevent code from being run which should not be there at all.
@@ThomasVWorm - Ahh, I get it…hence your comment about bad practice! Thanks!!
@@ThomasVWorm you're just no using it right.
Documentation is redundant in 10% cases, while it's missing in 90 % cases
I've yet to see a useful example of writing a variable type annotation. It is useful to write input and output type annotations for functions, but for variable it can be inferred and you don't want it to fail if you change the value - subsequent function calls will fail if needed, so everything is good
Everything was solid up to the p for p in people I don't use any single character variables anymore it doesn't help with readability and I'm not trying to minify everything if the goal is to make it readable. I would just use a list comprehension when absolutely necessary.
I agree, but sometimes there's too much redundancy, and the expression is simple enough that's it's obvious at 1st glance what the single-char var is
great content!
subscribed.
The data types is one reason why I dislike Python. But now that you have shown me type annotations, I will probably start trusting my code and what python does with it. It sounds silly, but it makes me more comfortable
Thank you for a great tutorial...
Very good video. Thanks
You can also use a Strong Type language like Swift which does all this work by the compiler instead of the Text editor and it is much more safer
Thanks a lot for the advices!!
I first learnt Python in version 2. I didn’t know half of these things were available now because they weren’t in Python 2. I did try to learn the differences when v 3 came out, but I think a lot of these were added after the first version of 3? Anyway thanks for broadening my horizons.
It took me only hours to code, but a lifetime to program.
even better, by replacing [ ] with ( ) around list expression you will get a generator! and able to call values one by one using next()
Besides comprehensions I prefer to use map or filter. I am not sure, but I think it's faster
That is a awesome kind of tips 👌
I really like the "big functions" part but I have a few questions:
1. how do I tell the difference between more "top-layer" functions and these small mini functions that just make things more abstract?
i mean, in terms of naming, often you would want to be bale to have a good hierarchy of this.
2. I'm not sure if this example is the best because one also needs to understand at which point should this separation happen. In practise the example probably should stay as it is, so that the code doesn't become more complex and hard to read.
Def main() is a must. Makes the code more readable and easier to use. Can help you seperate your functions in a nice and neat order. Great video overall.
Also makes it easier to call functions and kinda follow the order in which makes sense. Helps you walk through and see where variables are going and how they are affected.
"So, here I'm going to type in long_names, and that's going to shadow the name above, but... Who cares for now? And, actually, apparently, MyPy cares a lot ^^`"
Literally the best part of the video