Skip to content

CHANGELOG.md

Latest commit

 

History

History
61 lines (44 loc) · 3.85 KB

CHANGELOG.md

File metadata and controls

61 lines (44 loc) · 3.85 KB

@kemdict/gettext Changelog

v0.0.3 (unreleased)

Removals

  • Remove .getComment, which doesn't really make sense in a gettext runtime.
  • Remove the event system (gettext.on, gettext.off, gettext.emit). It has only ever been used for errors, and adds complexity for something that I don't see any use of.
  • Remove domains. Domains in upstream gettext is for loading multiple catalogs in the same app (same process), with there only being one “gettext instance”. In a class-based approach where you could just initialize a separate instance if you really need multiple message catalogs, domains are not necessary. As an example, Python's builtin gettext implementation also doesn't implement domains.

v0.0.2 (2026-02-04)

Changes

  • Include a Rollup plugin, rollup-plugin-gettext, for importing and bundling PO and MO files. This plugin will be published separately.
  • Set up TypeScript declarations
  • Add another GNU Hello clone which directly reads GNU Hello's installed message catalogs on the system (if any)
  • Add a loaders module which provides functions for loading PO files from directories
  • Rename guessLocale to guessEnvLocale, and allow it to take an object instead of directly reading from process.env, making it easier to test.
  • Instead of setLocale and setTextDomain, use the new bindLocale function to get functions that translate to the given locale(s) (and domain). These functions can be used standalone. This removes state from the Gettext instance itself and makes races against setLocale impossible.
  • Use a Map to store catalogs in the class. This should remove any concerns of prototype pollution, even though the translations being registered really are supposed to be trusted input. The types are clearer this way anyways. (Another approach to fix this is seen in postalsys/gettext.)

Removals

  • Remove .setLocale and .setTextDomain and their C-style aliases in favor of the new .bindLocale

Project changes

  • Set up some limited performance testing
  • Rename main test file to look better when there are multiple test files
  • Set up local coverage

v0.0.1 (2025-12-30)

First forked version.

This version is tagged, but the release pipeline hasn't actually been set up. There's even a chance that this ends up published under my personal npm scope and not Kemdict's. But the changelog is getting annoyingly large, so I felt I should just mark it now.

Changes

  • Remove getLanguageCode since it's too simplistic
  • Try to use the plural forms defined in the PO file. Fallback to using the locale to guess the plural forms only if it isn't found. When using locale to guess, also allow matching on regions, and add a default for Brazilian Portuguese.
  • Add guessLocale() to guess locale based on environment variables like GNU Gettext does.
  • Add .setFirstAvailableLocale to allow passing in a list of preferred locales and using the first that is available in the catalogs
  • Remove .addTranslations in favor of passing translations in in the constructor, because adding translations dynamically at runtime sounds like trouble
  • Add .getLocales to allow hacking on selecting locale based on what is available
  • Remove functions that node-gettext already deprecated
  • Add more types
  • Use optional chaining instead of lodash.get
  • Be an ES module

Project changes

  • test with node:test and node:assert instead of mocha and chai; remove grunt
    • TODO: move from sinon to node:test's mocking facilities
  • set up typescript in-editor checking (ie. turn on checkJs)
  • stop using jshint
  • set up prettier
  • remove unused dependency
  • use pnpm