1. 8
    1. 4

      Man, there’s so many people that don’t read anything.

      I currently fill the role of IT and it’s absolutely bonkers the stuff people get stuck on. It’s so basic, and the screen is telling you what’s wrong, but people aren’t even parsing it.

      1. 4

        The irony here is that the article claims that people actually do spend a significant amount of their time reading compiler errors.

        1. 3

          The bonus irony is that it’s discussing a paper that, guess what, no one will read! Because it links to a paywalled version of it instead of the author’s copy available freely on their web site: https://people.engr.ncsu.edu/ermurph3/papers/icse17.pdf

    2. 4

      A lot of developers might not, but personally something I’ve found recently is that you really, really miss clear error messages when you do not have them. I’ve been trying to get something to work in Expo + React Native and it just will not work, and the error messages are the most obtuse I’ve ever seen – and my other side project is hacking together a bootloader lisp in 8086 assembler.

    3. 2

      SFINAE errors (C++) must be a category of its own. The nature of those is that there are any number of irrelevant errors (that’s why C++ devs need terminals with 10k lines of scrollback buffer), and the first thing to determine is whether the relevant error is among them or not – if you haven’t included the right header, it won’t be.

      1. 2

        When I’m getting template-related errors in C++, I’m reading it more or less: “you have some error somewhere in this file, somewhere in your templates. Please fix.”

    4. 2

      […] nothing in software engineering makes sense except in light of human psychology; the more attention researchers and developers pay to that, the faster our profession will make real progress.

      GCC and Clang have gotten a lot better at providing useful error messages in the last few years. They seem to have taken this (good) advice to heart.