4 ms·

Today, you could paste that into an LLM and it would immediately identify the issue. The next generation of developers have so much power at their fingertips.

deathanatos 2y ago
I was young at the time, but when I TA'd, the lesson I'd try to impart here is to take the time to learn the lexicon of the compiler error, and understand what it is trying to tell you. Then it's not just "code broke yo" but "is here is what is wrong." Much better to learn to have a conversation with a deterministic compiler than an LLM that'll just cook up some random junk when it gets backed into a corner.
- zahlman 2y ago
  I feel like after all these decades of research and implementation, compilers (and runtimes) ought to be better at directly explaining what's wrong. Python has made some significant strides in this area recently, but in my mind it's nowhere near enough.
  - IIsi50MHz 2y ago
    Same. My first development environment was FutureBASIC (not counting booting an Apple ][ into the ROM's AppleSoft). The compiler error messages usually showed the right line number, the content of the line, and the where in that line something was wrong. Sometimes the real error was one line away. Rarely was the error further away. It didn't fight me about whether a new line was actually a new line. It never told me mostly useless stuff like C, which I was exposed to a few years later. Error messages in C-ish languages seemed like they could be summarised as "Syntax error on line 247. Additionally, you have errors on 3123 other lines.". Often, the real error was on some earlier line, pages away. The following may seem shocking to people who got started with typical programming languages, but most IDEs after my experience with FutureBASIC felt more like "a loose arrangement of parts that kinda work together if you configure them just right", rather than an _integrated_ development _environment_. (-: …Just trying to get a build environment set up or reconfigured seemed error-prone and had its own cryptic error messages. With FB, I just copied the application/executable file wherever I wanted, and copied a small folder/directory of support files to the same location. I didn't have to worry about separate "linking" and "object files", or how to tell the editing environment where the compiler was before the environment would know what to do when I told it to compile
  - LandR 2y ago
    > directly explaining what's wrong. My limited experience with the Rust compiler is that its very good at this, at least for a total newbie to Rust.

deathanatos 2y ago

I was young at the time, but when I TA'd, the lesson I'd try to impart here is to take the time to learn the lexicon of the compiler error, and understand what it is trying to tell you. Then it's not just "code broke yo" but "is here is what is wrong." Much better to learn to have a conversation with a deterministic compiler than an LLM that'll just cook up some random junk when it gets backed into a corner.

zahlman 2y ago
I feel like after all these decades of research and implementation, compilers (and runtimes) ought to be better at directly explaining what's wrong. Python has made some significant strides in this area recently, but in my mind it's nowhere near enough.
- IIsi50MHz 2y ago
  Same. My first development environment was FutureBASIC (not counting booting an Apple ][ into the ROM's AppleSoft). The compiler error messages usually showed the right line number, the content of the line, and the where in that line something was wrong. Sometimes the real error was one line away. Rarely was the error further away. It didn't fight me about whether a new line was actually a new line. It never told me mostly useless stuff like C, which I was exposed to a few years later. Error messages in C-ish languages seemed like they could be summarised as "Syntax error on line 247. Additionally, you have errors on 3123 other lines.". Often, the real error was on some earlier line, pages away. The following may seem shocking to people who got started with typical programming languages, but most IDEs after my experience with FutureBASIC felt more like "a loose arrangement of parts that kinda work together if you configure them just right", rather than an _integrated_ development _environment_. (-: …Just trying to get a build environment set up or reconfigured seemed error-prone and had its own cryptic error messages. With FB, I just copied the application/executable file wherever I wanted, and copied a small folder/directory of support files to the same location. I didn't have to worry about separate "linking" and "object files", or how to tell the editing environment where the compiler was before the environment would know what to do when I told it to compile
- LandR 2y ago
  > directly explaining what's wrong. My limited experience with the Rust compiler is that its very good at this, at least for a total newbie to Rust.