Hi everyone!

I just published my first Hex package, erli18n — a GNU gettext–compatible internationalization library for Erlang/OTP, written in pure Erlang (and callable from Elixir too, since it’s a normal Hex dep: :erli18n.gettext(...)).

I want to be upfront about one thing: the main goal of this project is learning. I built it to dig deep into Erlang/OTP — gen_server + supervision, ETS ownership/heir patterns, the .po format and CLDR plural rules, property-based testing with PropEr, telemetry, native EEP-59 docs, and the whole Hex release pipeline. So please read it as a 0.1.0 learning project rather than battle-tested production infra — though I worked hard to make it correct and thoroughly tested.

What it does — the full GNU gettext C-macro family as plain Erlang functions:

application:ensure_all_started(erli18n).

{ok, _} = erli18n_server:ensure_loaded(my_domain, <<"pt_BR">>,

<<"priv/locale/pt_BR/LC_MESSAGES/my_domain.po">>).

<<"Olá, mundo">> = erli18n:gettext(my_domain, <<"Hello, world">>, <<"pt_BR">>).

%% ngettext returns the correct plural FORM for N (you format the number yourself)

<<"arquivo">> = erli18n:ngettext(my_domain, <<"file">>, <<"files">>, 1, <<"pt_BR">>).

<<"arquivos">> = erli18n:ngettext(my_domain, <<"file">>, <<"files">>, 42, <<"pt_BR">>).

%% pgettext for context; npgettext for context + plural

<<"Maio">> = erli18n:pgettext(my_domain, <<"month">>, <<"May">>, <<"pt_BR">>).

A few things I focused on:

• Drop-in .po / .pot — loads the files translators already produce in Poedit, Crowdin, Weblate, or xgettext.

• Real CLDR pluralization — an actual Plural-Forms evaluator, CLDR rules inlined for 49 locales.

• Lock-free lookups — reads run straight from ETS in the calling process; only writes go through a gen_server, so there’s no bottleneck on the hot path.

• Optional telemetry — 7 events (catalog spans, lookup misses, plural divergence, memory warnings); telemetry is an optional dependency.

• Heavily tested — Common Test + PropEr + fuzzing, plus a parity suite that checks output byte-for-byte against GNU msgfmt as a ground-truth oracle.

(Pure Erlang, OTP 27+, Apache-2.0.)

Why I’m posting here:

1. I’d genuinely love feedback — on the API design, the OTP patterns, anything that makes a seasoned BEAM dev wince. Since I’m here to learn, blunt and critical opinions are exactly what I’m after.

2. If you have an Erlang project that wants gettext-style i18n without routing through Elixir’s build, please try it and tell me where it breaks.

Links:

• Hex: https://hex.pm/packages/erli18n

• Docs: https://hexdocs.pm/erli18n

• GitHub:

If you find it useful or even just interesting to read, a on GitHub would mean a lot and helps me gauge whether it’s worth continuing. Thanks for reading — any feedback, however harsh, is hugely appreciated!