Practice makes perfect?
-
If I find myself repeating the same thing over and over it occurs to me that I need to optimize. Doubly so when working with a computer - after all, the computer is there to do work so I don't have to. There's something to be said for repetition of course. Practice helps, and there's something to be said about tried and true methods, especially in production code, but again, repetition raises the question of optimization. This is why design patterns make me low key uncomfortable. I feel like they should be baked into the language, rather than having to repeat the same boilerplate over and over again. With C# they spend all this time adding fluff to the language, when they could be baking in design patterns. Seems a missed opportunity. I'm a fan of the idea of Domain Specific Languages for this reason, even though I've never used one aside from Synthmaker/Flowstone which I'm not sure counts. What I'd really like is a language that allows one to augment the grammar, sort of like C# source generator tech except it extends the keywords and syntax of the language and then uses the new grammar to generate code.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
honey the codewitch wrote:
What I'd really like is a language that allows one to augment the grammar,
Extensible languages, bell-bottoms and polka dots were a thing in the 60-es. They didn't age too well :) For your reading pleasure check this SPG language.
Mircea
-
honey the codewitch wrote:
What I'd really like is a language that allows one to augment the grammar,
They call it Forth. :-D
"In testa che avete, Signor di Ceprano?" -- Rigoletto
I would argue that Forth doesn't have any grammar to speak of. It's just a string of words strewn together. Would stream-of-consciousness language* be an appropriate name? :laugh: *) Name calling made with love: in another life I wrote a Forth interpreter for a PDP-11
Mircea
-
If I find myself repeating the same thing over and over it occurs to me that I need to optimize. Doubly so when working with a computer - after all, the computer is there to do work so I don't have to. There's something to be said for repetition of course. Practice helps, and there's something to be said about tried and true methods, especially in production code, but again, repetition raises the question of optimization. This is why design patterns make me low key uncomfortable. I feel like they should be baked into the language, rather than having to repeat the same boilerplate over and over again. With C# they spend all this time adding fluff to the language, when they could be baking in design patterns. Seems a missed opportunity. I'm a fan of the idea of Domain Specific Languages for this reason, even though I've never used one aside from Synthmaker/Flowstone which I'm not sure counts. What I'd really like is a language that allows one to augment the grammar, sort of like C# source generator tech except it extends the keywords and syntax of the language and then uses the new grammar to generate code.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
I say, stop bloating the languages. Do the basics well, and use libraries for everything else. If you keep doing the same thing repeatedly, libraries are well suited for that purpose - do it once, wrap it up, and re-use it. Otherwise you're just extending the language to suit your own needs, and everybody's needs are different - and that's what leads to bloat. Not only that, but you end up spending more time tinkering with the language extensions than the actual code and business logic. And not everybody can work at that level.
-
I say, stop bloating the languages. Do the basics well, and use libraries for everything else. If you keep doing the same thing repeatedly, libraries are well suited for that purpose - do it once, wrap it up, and re-use it. Otherwise you're just extending the language to suit your own needs, and everybody's needs are different - and that's what leads to bloat. Not only that, but you end up spending more time tinkering with the language extensions than the actual code and business logic. And not everybody can work at that level.
Seems like C# got bloated without really doing any of this. is it really such a bad idea for example to do public component Foo { // implements IComponent, IPropertyNotify, IDisposable etc public Bar { get; set; } // generates OnBarChanged/BarChanged, etc }; and save all of that code? You can't really make that into a library. At best you're using a C# Source Generator and some sort of input document. or even public entity Contact { // implements all of the data bound stuff }; And if that's too much, what about simply the ability to extend the grammar via Source Generators so you could create or install these (as nuget packages) to extend the language as the project requires?
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
-
Quote:
It's just that if libraries could fill the gap by themselves, I feel like they would have already?
Stop inventing new problems and processes. Then the current libraries will be enough. As long as there is a perceived need for something new, someone will write a library to fill that need. Take your pixel library for instance. I am sure there is already libraries out there that would do what yours does, but not the way you want it done. So you wrote another library for it. Just more people reinventing the same wheel over and over again.
Within you lies the power for good - Use it!
Forgive me but I think you're missing my point. You can't write a library to do this:
public component Foo { // implements IPropertyNotify, IComponent, IDisposable, etc
public Bar { get; set; } // implements OnBarChanged, BarChanged, all the metadata attributes like Browsable etc
}The issue there is at some point you need to insert those actual methods into your code, or inherit from something that already implements Bar which you can't do if Bar is your project specific property. As such, you're left with one other option for automating this. Source Generators, which require some kind of input spec. Metadata attributes don't quite cut it here, because they aren't descriptive enough. Being able to extend the language via Source Generators solves this. That was the last mile integration I was talking about.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
-
honey the codewitch wrote:
What I'd really like is a language that allows one to augment the grammar,
Extensible languages, bell-bottoms and polka dots were a thing in the 60-es. They didn't age too well :) For your reading pleasure check this SPG language.
Mircea
There are all kinds of ways to skin that cat though, and the 1960s you had some severe limitations in terms of compilers due to the limited RAM and speed of computers in that era.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
-
If I find myself repeating the same thing over and over it occurs to me that I need to optimize. Doubly so when working with a computer - after all, the computer is there to do work so I don't have to. There's something to be said for repetition of course. Practice helps, and there's something to be said about tried and true methods, especially in production code, but again, repetition raises the question of optimization. This is why design patterns make me low key uncomfortable. I feel like they should be baked into the language, rather than having to repeat the same boilerplate over and over again. With C# they spend all this time adding fluff to the language, when they could be baking in design patterns. Seems a missed opportunity. I'm a fan of the idea of Domain Specific Languages for this reason, even though I've never used one aside from Synthmaker/Flowstone which I'm not sure counts. What I'd really like is a language that allows one to augment the grammar, sort of like C# source generator tech except it extends the keywords and syntax of the language and then uses the new grammar to generate code.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
-
I would argue that Forth doesn't have any grammar to speak of. It's just a string of words strewn together. Would stream-of-consciousness language* be an appropriate name? :laugh: *) Name calling made with love: in another life I wrote a Forth interpreter for a PDP-11
Mircea
-
What you describe is exactly why I think MSFT is on point with the "targetted models" approach to AI-assist in code.
Now that's interesting. It occurred to me that AI might be able to insert itself here with some benefit to the developer but I wouldn't know enough about it - having not even used CoPilot for example - to really say so this is news to me.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
-
If I find myself repeating the same thing over and over it occurs to me that I need to optimize. Doubly so when working with a computer - after all, the computer is there to do work so I don't have to. There's something to be said for repetition of course. Practice helps, and there's something to be said about tried and true methods, especially in production code, but again, repetition raises the question of optimization. This is why design patterns make me low key uncomfortable. I feel like they should be baked into the language, rather than having to repeat the same boilerplate over and over again. With C# they spend all this time adding fluff to the language, when they could be baking in design patterns. Seems a missed opportunity. I'm a fan of the idea of Domain Specific Languages for this reason, even though I've never used one aside from Synthmaker/Flowstone which I'm not sure counts. What I'd really like is a language that allows one to augment the grammar, sort of like C# source generator tech except it extends the keywords and syntax of the language and then uses the new grammar to generate code.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
That is what IDEs should help with. They will write all of that boiler plate for you. The new AI generators might help, too. I have not used them directly, but I have watched team mates flounder trying to have AI generate simple commands that I was able to find in the man page.
-
That is what IDEs should help with. They will write all of that boiler plate for you. The new AI generators might help, too. I have not used them directly, but I have watched team mates flounder trying to have AI generate simple commands that I was able to find in the man page.
I mean, I think it would be better if the boilerplate didn't have to exist, if possible, simply because even if generated, it still creates a maintenance point. If some of the boilerplate were baked in as language constructs, or if the language was extensible, you could eliminate that boilerplate at the last mile .i commented to some others here giving an example that a library couldn't solve in C#, but the IDE approach, while it would solve it and does, even with like Winforms tech and InitializeComponent(), now you have InitializeComponent() to maintain. Not the greatest example because it round trips from the designer but most of the time with most tools that's not the case.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
-
There are all kinds of ways to skin that cat though, and the 1960s you had some severe limitations in terms of compilers due to the limited RAM and speed of computers in that era.
Check out my IoT graphics library here: https://honeythecodewitch.com/gfx And my IoT UI/User Experience library here: https://honeythecodewitch.com/uix
And yet even though there are probably 100 new languages a year that actually make it into some sort of larger visibility there are none like that. Which suggests to me at least that they just do not work. In the same way that I have seen, several different times, people attempt to 'improve' C/C++ by adding clever macros that serve as nothing but attempts to do custom enhancements to the language. For example adding 'begin' and 'end' as substitutions for open/closing brackets.
