4 ms·

> I didn't do XDG because this route seemed simpler Simpler how? This requires custom config, instead of following what I set system-wide. > and XDG isn't som

> I didn't do XDG because this route seemed simpler Simpler how? This requires custom config, instead of following what I set system-wide. > and XDG isn't something that is used everywhere. Yeah, that‘s why it defines defaults to fall back on.

burntsushi 3y ago
It's far simpler to implement. No, you don't understand. I'm not saying The XDG variables might not be defined. Give me a little credit here lol. I have more than a passing familiarity with XDG. I've implemented it before. I'm saying the XDG convention itself may not apply. For example, Windows. And its controversial whether to use them on macOS when I last looked into it. I don't see any significant problem with defining an environment variable. You likely already have dozens defined. I know I do. I'm not trying to convince you of anything. Someone asked why. This is why for ripgrep at least.
- ijustlovemath 3y ago
  Could ripgrep not simply add a check for the XDG environment variables and use those, if no rg environment variable is given? Of course if both are not available you would use the default.
  - burntsushi 3y ago
    Of course. But now you've complicated how config files are found and it doesn't seem like an improvement big enough to justify it. Bottom line is that while ripgrep doesn't follow XDG, it also doesn't force you to litter your HOME directory. That's what most people care about in my experience. I would encourage you to search the ripgrep issue tracker for XDG. This has all been discussed.
  - smaudet 3y ago
    The issue is complexity - we could create some sort or 'standard tool' library that 'just works' on all platforms, but now building the tool and runtime bootstrapping the tool become more complex, and hence more likely to _break_. Really most people want it in their path and it just to work in as many scenarios as possible. Config almost shouldn't be the responsibility of the tool at all... (Options passed to the tool via env variables, perhaps)...

burntsushi 3y ago

It's far simpler to implement. No, you don't understand. I'm not saying The XDG variables might not be defined. Give me a little credit here lol. I have more than a passing familiarity with XDG. I've implemented it before. I'm saying the XDG convention itself may not apply. For example, Windows. And its controversial whether to use them on macOS when I last looked into it. I don't see any significant problem with defining an environment variable. You likely already have dozens defined. I know I do. I'm not trying to convince you of anything. Someone asked why. This is why for ripgrep at least.

ijustlovemath 3y ago
Could ripgrep not simply add a check for the XDG environment variables and use those, if no rg environment variable is given? Of course if both are not available you would use the default.
- burntsushi 3y ago
  Of course. But now you've complicated how config files are found and it doesn't seem like an improvement big enough to justify it. Bottom line is that while ripgrep doesn't follow XDG, it also doesn't force you to litter your HOME directory. That's what most people care about in my experience. I would encourage you to search the ripgrep issue tracker for XDG. This has all been discussed.
- smaudet 3y ago
  The issue is complexity - we could create some sort or 'standard tool' library that 'just works' on all platforms, but now building the tool and runtime bootstrapping the tool become more complex, and hence more likely to _break_. Really most people want it in their path and it just to work in as many scenarios as possible. Config almost shouldn't be the responsibility of the tool at all... (Options passed to the tool via env variables, perhaps)...