1.0.0 roadmap

[SuperchupuDev]

still likely not happening anytime soon, just leaving future plans here to gather thoughts:

plan is to make 1.0.0 as full of major changes as possible to ideally never have to release a major version again

• removal of deprecated patterns option
• make expandDirectories default to false (considering, unsure, would involve MOST users changing their code to maintain the same behavior, would make it less compatible with globby by default but more compatible with fast-glob)
• switch to esm (would half package size, users on old node versions could still use 0.x which id try to still maintain)
  • would this involve upping engines range? not sure. ideally would be >=20.19.0 or something like that for require-esm
  • to avoid people migrating back to fast-glob this could wait until fast-glob goes esm-only. i've seen it's planned for 5.0.0. it's not even at 4.0.0 yet so it might take a while
• switch to zeptomatch. would need switching to esm
  • this would be great since it's faster than picomatch, smaller and actively maintained. picomatch also has some longstanding bugs
  • downside: it's two dependencies instead of none. considering node-glob just got to 6 dependencies i really don't want to increase the number of dependencies to remain competitive
• should ignore stop taking strings? every other library only takes string[]s

---

super future potential plans (2.0.0 or something like that, might never happen)

• go zero dependency and use fs.glob and path.matchesGlob
  • they both currently use minimatch under the hood which is slow. would need node to make it more performant. itd be cool if they moved the implementation to c++
  • would need a huge engines increase (id be most comfortable with >=24). if this ever happens enough time would have passed for it to be a common engine range
  • the point of tinyglobby would then be for those who can't migrate to fs.glob due to needing the same api and behavior as globby/fast-glob