1. 31
  1.  

  2. 12

    Nice; I hope this will speed Guix up significantly.

    I’ve written a fair bit of Guile on Guix, and the experience has been pretty uncomfortable:

    • It’s slow (it seems to combine the downsides of “being interpreted”-slow with “having a slow compiler”-slow)
    • It spams useless output (compare the “recompiling” messages in the linked thread). Those happen all the time; with Guix’ long list of package modules that means frequent screen-fulls of such output. Python seems to have solved this better – I don’t remember ever running into issues with the *.pyc files.
    • Like Python, Guile code likes to throw stack traces at you. Unlike Python, they’re frequently unintelligible. It doesn’t help that lines are truncated to 80 characters.

    If even just the first point is addressed, it would be a very welcome change. I’m unsure how soon Guix plans to switch over though.

    1. 2

      Nice; I hope this will speed Guix up significantly.

      As a fellow Guix user, me too.

      If even just the first point is addressed, it would be a very welcome change.

      Out of curiosity, have you opened bug reports for the latter two points?

      1. 5

        Out of curiosity, have you opened bug reports for the latter two points?

        Yes, I made some efforts.

        1. 2

          Great! And thanks for the links, will check them out shortly. And I agree; it is pretty frustrating…

    2. 6

      I haven’t used Guile in a long time, but it’s always been what I want to be my favorite Scheme. In particular, Guile always feels like S-expression Dylan, or a cleaned-up Common Lisp, in the best possible way: a huge chunk of batteries and nods to practicality are included, even if they might not be the most academically pure option, but the overall environment still feels far more consistent and coherent than a lot of real-world languages. It basically has always looked like a have-my-cake-and-eat-it-too environment. The two main things that have held me back in the past have been its speed (it’s generally been slower than Chicken Scheme by enough I care) and concern over the project health. But both have picked up a lot over the past couple years, from what I’ve seen. I’ll definitely look forward to finding an excuse to do more here in the future.

      1. 4

        I feel like Guix has probably revived some interest in Scheme.

        1. 3

          That, and Guile in Emacs increasingly looking like a thing that might actually happen, have been two big parts, IMVHO.

          1. 3

            Has the Guile Emacs project had activity lately? My impression (mostly from https://www.emacswiki.org/emacs/GuileEmacs and the linked repos) was that it had effectively died off a few years ago. I’d be excited to learn that’s not the case.

          2. 1

            (To clarify, I meant Guile specifically rather than Scheme)

          3. 2

            The two main things that have held me back in the past have been its speed (it’s generally been slower than Chicken Scheme by enough I care)

            If what is being said is true, the these pre-preleases to 3.0, and of course the 3.0 release itself should deliver considerable improvements when it comes to performance. I know that the publishing of Chez scheme’s source pushed Racket to adopt some of their performance tricks, but it might have also been the case with Guile too.

            1. 4

              If I’m not mistaken the publishing of Chez Scheme’s source encouraged racket to host their language on top of it. So now Racket code compiles down to Chez.