Rich Felker
It's so weird to me that ppl use and like IDEs.
I'm not writing any code that can't be written just as well with pen & paper on top of a mountain far away from any computer.
If your language, idioms, or frameworks aren't compatible with that, they're a non starter for me.
ck
@dalias why use a computer ever then obviously a pen and paper is better at everything 😎
Rich Felker
@cursedsql Because it can compute things that actually have to be repeated at scale to be meaningful, like pixels in an image or cryptography. Not to add gratuitous work.
ck
@dalias if you can write code to do work, and coding involves work, does it not follow we could use code to improve how to do work? Even if you hate all IDEs today its like saying there's no way code could be used to improve coding in a way that creates an IDE, that's a weird take!
Rich Felker
Useful exercise when programming: imagine your copy&paste doesn't work so anything repetitive/boilerplate actually has to be written out that many times by hand.
So you think about if it's really necessary, and either generalize so that it's not, or instead write out the rules by which each instance can be generated at build time rather than repeating yourself and calling that "source".
Wanja
@dalias I have written a ton of code in nano, and I could write the code I write today the same way. I do use an IDE these days however, not because of necessity but simply for comfort.
It hardly helps with writing code - I have autocomplete and such turned off - but I find it makes it easier for me to navigate code. "Go to definition" is probably 90% of the reason I bother with an IDE today.
Rich Felker
@olsner Just everything that makes you dependent on somebody's (pretty much always some capitalist's) software suite. The boilerplate generation needs, the integration of entire process, the encouragement to write code that's unmaintainable without their tooling, ...
Simon Brenner
@dalias which part of the IDE? Sounds like your main gripe is the editor (annoying itself though with popups and ”helpful” completions and other clutter), but the part I dislike the most is the integratedness. I don’t want my debug/test/build/edit/navigate stuff all tangled up in the same program and blocking each other. One vim and a big bowl of terminals is what I want.
Rich Felker
@alexsaezm The whole point of Java is to justify the price tag of bids for government contracts & similar where the only metric of productivity you can sell is MLoC.
Álex Sáez :fedora: :redhat:
@dalias this reminds me of a teacher I used to have at university, in the very first programming course. The exams were with pen and paper, and it was a mess. But it taught us one thing: no human can write a hello world in Java without an IDE, let alone compile it.
Rich Felker
@alwayscurious Symbol renaming is actually what I consider one of the biggest IDE misfeatures. It makes churn that disrupts understanding someone of an old version if the code could otherwise apply to newer versions, gratuitously breaks patches/branches folks might be maintaining out of tree, and when applied to public interface surface, gratuitously creates version mismatches in the interface contracts.
Naming that looks suboptimal in hindsight is *okay*.
Demi Marie Obenour
@dalias I use an editor with LSP support (Kate) because autocompletion, symbol renaming, and viewing errors in the editor are helpful when writing Rust. In C they might be less useful.
Rich Felker
@alwayscurious I'm all in favor of committing expensive-to-regenerate derived files to your repo in cases like this, but they should still be produced by declarative rules documented inside your repo in a way that can be done reproducibly with known versions of the tool that does the derivation.
It shouldn't be "the IDE generates the file via user interaction, and the user interaction (that was not recorded) is the actual source code".
Demi Marie Obenour
@dalias What about code that is autogenerated but is computationally very expensive to generate?
Consider a dependently typed language reliant on SMT solvers. To me, it makes much more sense to include the SMT solver output in the repository, so that you don't need to re-run it every time.
Rich Felker
@alwayscurious Even if the user interaction were recorded, the "integrated" part of IDE often means that the GUI and code generation logic are a package deal. It's unlikely you're going to be able to easily run a version from 10 or 20 years ago that was used to produce the code, because the UI framework it ran on top of is no longer a thing except perhaps by retrocomputing enthusiasts.
Rich Felker
@alwayscurious I think largely so, yes. Provided that command line tool is portable and future-proof and available independent of the IDE that will bitrot.
Demi Marie Obenour
@dalias Would having the IDE also generate a transcript of user interactions that could be replayed by a command-line tool solve the problem?
Rich Felker
@alwayscurious @ity @navi And very intentionally so - obvious from the old logo which was like a Rorschach test for which kind of "cock" you see in it.
Demi Marie Obenour
@ity @navi @dalias It's because the English pronounciation of the previous name was the same as that of an English NSFW slang term.
Demi Marie Obenour
@navi @dalias So dependently typed languages (such as Rocq) are actually more reliant on IDEs, due to the huge amount of information hidden in type signatures.
![ity [unit X-69] - VIOLENT FUCK](https://gnusocial.jp/avatar/308499-48-20241222022318.webp){kind=small}
ity [unit X-69] - VIOLENT FUCK
@alwayscurious @navi @dalias still strange to me to see the lang referred to as Rocq
anna
GNU social JP is a social network, courtesy of GNU social JP管理人. It runs on GNU social, version 2.0.2-dev, available under the GNU Affero General Public License.
All GNU social JP content and data are available under the Creative Commons Attribution 3.0 license.