26 replies

1. Stuart Langridge 2 min later

   @zachleat go full old car grille and normalise typing +=+ which is basically OK always

   1. Zach Leatherman 6 min later

      @sil done. I also added you as the primary support point of contact for all of my repositories moving forward

      1. Stuart Langridge 13 min later

         @zachleat I am perfectly fine with this. I believe that maintainers of rival static site generators are so fine with that that they’re throwing a party

         1. Zach Leatherman 16 min later

            @sil *immediate regrets*

            1. Stuart Langridge 21 min later

               @zachleat in breaking news (for me alone), += works on strings, it turns out. Does anybody do that? Nobody does that, right?

               1. Bill Mill 24 min later

                  @sil @zachleat I do it all the time, why not?

                  1. Stuart Langridge 44 min later

                     @llimllib @zachleat After a certain amount of investigation of my own psyche, basically I don't because I have internalised the idea "don't build up strings bit by bit by appending to an existing string" so hard that I don't do it even if there are only two bits to build up.

                     1. Bill Mill 53 min later

                        @sil @zachleat it used to be such a huge python antipattern! In JS, the join still does a bit better but it will never matter except in a very hot loop.

                        It's 29% faster in this microbenchmark

                        1. Bill Mill 1 hr later

                           @sil @zachleat And, to satisfy my own curiousity, I tested a template string with "x = `${x} ${parts[i%3]}`", and that was quite a bit slower

                     2. Private reply 9 hr later
               2. Zach Leatherman 33 min later

                  @sil yeah I used to see that pattern a lot (especially before template literals) but I see it less now

               3. paulfreeman 48 min later

                  @sil @zachleat all the time Aq!

   2. Private reply 9 hr later
2. Zach Leatherman 3 min later

   oh! eslint has `space-unary-ops` which would have alerted me—wonder if I’m missing that in my config

   https://eslint.org/docs/latest/rules/space-unary-ops

   (and `space-infix-ops` too! https://eslint.org/docs/latest/rules/space-infix-ops)

   1. Eric Eggert 6 min later

      @zachleat should really be space-unary-oops

      1. Zach Leatherman 11 min later

         @yatil I literally can’t disagree

         1. Eric Eggert 1 hr later

            @zachleat It’s really hard to.

   2. Marco Rogers 6 hr later

      @zachleat fascinating.

3. Andrew 5 min later

   @zachleat oh yeah, they seem to have a rule to forbid a++ but not one to forbid +a (as far as I can see) when unary plus is exclusively used for terrible cursed type munging and should be binned off along with the comma operator

   that wouldn't help with -= but i mean it's something

4. Bill Mill 6 min later

   @zachleat From an AST perspective, it's code that humans might write; `x = +string_or_number` will coerce `string_or_number` to a number. The d3 source code uses this idiom a lot, for example.

   (Obviously you might still want an eslint rule that disallows this idiom, but typescript definitely shouldn't/can't flag it imo)

   1. Zach Leatherman 9 min later

      @llimllib yeah, for sure. Unit tests caught the issue but it would have been nice to see an error in my editor—I don’t use that pattern in my code 👀

      1. Bill Mill 21 min later

         @zachleat Agreed, and the formatting is a dead giveaway that you didn't intend the coersion idiom.

         Just tested in my editor/eslint setup, and eslint doesn't notice/flag it but prettier moves the + to the proper spot, which makes it clearer what's happening, as long as you happen to notice it.

         1. Zach Leatherman 36 min later

            @llimllib prettier moving the + so that the eslint rule I added doesn’t catch it would be a gnarly combo https://fediverse.zachleat.com/@zachleat/112722798334691839

5. webbureaucrat 4 hr later

   @zachleat

   tf does =+ do?

   1. Zach Leatherman 5 hr later

      @webbureaucrat like = but also coerce to a number

      1. webbureaucrat 5 hr later

         @zachleat JavaScript was a mistake.