This project was interesting. Recently, I’ve been digging into the functional programming paradigm. This is one of my first tries at it. I’ve been doing OOP until recently, but this project really was impressive to me. The whole implementation took about nine hours. Functional programming is much less convoluted. I spent more time programming than deciding on a good name for a certain identifier.
Not only is functional programming more efficient, but I can also see that it’s much less tedious to write automated tests for. I only have to take care of the local scope of the function I am writing a test for; there is no need to deal with the parent’s inherited mess or even any parent’s mere state. I just have to write the test for the function.
I have scraped the Cambridge Dictionary to collect the data.
The project is licensed under MIT at:
If your OO experience is from Java, the problem is with Java culture, not OO in general.
Putting functions and properties into classes is good and helpful. Whyever the fuck Java people want to implement the factory pattern three times in every application they make is convoluted and asinine.
My OOP experience isn’t from Java, but I get your point. I don’t really have a dislike for OO; it sure does have its applications. I once met a dude who was trying to use an object oriented library in a functional way; the result of that was a mess full of complications. I feel a good balance is necessary.
Object oriented programming rose to popularity a few decades ago, along with a great deal of indoctrination and marketing for both the paradigm and related products. It was pushed so hard and so widely that more than a few folks grew into their developer roles assuming it was the best approach to every problem. Of course, it isn’t.
(It definitely is useful at times, though, so please don’t condemn it just because it’s sometimes a square peg in a round hole.)
I’m glad to see you have discovered an alternative, and I hope you will continue to expand your toolbox and aim for simplicity.
When it’s easier to think about a system in terms of objects. The classic example is video game Entities/Game Objects. GUI stuff has also been very pleasant with OOP as buttons and the like are all easily conceptualized and worked with when seen as objects.
One of the best use cases is implementing abstract data types and hiding the memory management and other potentially unsafe optimization tricks behind a clean and high level abstraction.
Also since it’s a logical/mathematical construct and not attempting to model the real world (like business logic), it’s one case where inheritance hierarchies will remain stable.
Wasn’t there a video a while back of a presentation that OOP was created as a joke or something, and the person was surprised people were taking it seriously. Might have been advertised as the creator or OOP.
I took the video as a joke anyways, not a serious thing, but who knows. I don’t even remember if it was OOP, or some other paradigm, or language, or who knows what else.
If anyone could find/link to it, I’d love to watch it again, but I’m having no luck; so my memory may be faulty.
Looks beautiful with that gruvbox theme
As someone else mentioned, you should rewrite this in Haskell (or Purescript or Scala or even Python’s ‘Coconut’) because using vanilla Python for functional programming is like driving Formula 1 with a Toyota Camry.
I will try porting this project to Haskell and Coconut later. I am currently doing a rewrite of this in Nim.
Nice. I don’t know much about Nim.
I should also mention: Once Purescript switches over to a Chez Scheme compiled back end, it would be (IMO) the ideal lightweight Haskell, suited to quick terminal apps that you’re looking for.
Thanks for note. Do they currently have that backend?
That aside, you might want to try Nim. It’s pretty cool. It can compile to C and C++, and JS. There have been browser extensions made with it. Heck, it even has an LLVM backend. And the C code it generates it pretty fast on benchmarks. It’s filled with tons of metaprogramming stuff and AST-level macros. And it has this cool thing where it can ignore name casing of identifiers like variables and functions; so isSome == is_some.
I saw “terminal” and “dictionary” and thought this was going to be using the DICT protocol to retrieve definitions. If that’s not supported, could that be added?
Not for any reason other than I think it’s cool to look up stuff remotely using nothing more than curl. But this is definitely easier to read.
You mean, like, support for the dict protocol for this program’s interface? I am also scraping a dictionary’s data, so I am a little confused.
What I mean is, this program draws from a scraped database, right? Is there any reason it couldn’t also search an online database using DICT, to potentially get more definitions that aren’t in the Cambridge Dictionary?
Excellent work, btw
Ah that. That shouldn’t be a lot of work as all the visual stuff are done by separate functions. I can do it. I will look into it.
Hi, I spent some time trying out the dictd package. I also read this protocol’s specification. As things are right now, each host-name would require its own parser, because I couldn’t notice a very similar pattern between them. Webster, Jargon, wn, all these have their own standardization for including synonyms and examples.
The specification doesn’t enforce any pattern on the definitions either. I don’t think it’s going to be very useful even if I do implement it because the parsers are going to be quite complicated.
