Reading the Polish Usenet archive with galaxy-web

 TITLE: READING THE POLISH USENET ARCHIVE WITH GALAXY-WEB
 CREATED: 7 OCTOBER 2026
 UPDATED: 7 OCTOBER 2026

A LinkedIn thread about pl.pregierz, the Polish newsgroup where people went to publicly shame and condemn things, set off a wave of nostalgia. I went looking for its archives, and since I collect things and have a soft spot for old computer days, a quick look ended with me downloading all of it.

A BIT OF USENET

Usenet is a distributed discussion system that predates the web: newsgroups organised in hierarchies (pl.*, alt.*), threaded messages, read with a newsreader over NNTP. The Polish side was big. By the Kluska article on Usenet's Polish beginnings, the pl.* hierarchy saw nearly 20,000 posts a day around the turn of the millennium, with strict moderation, "lurk before you post" etiquette and a culture that slowly slid into trolling.

pl.pregierz was the group for shaming and condemning. Its FAQ is a 2005 page, version 2.5, and looks exactly as old as it is.

THE ARCHIVE

usenet.nereid.pl hosts archives of Polish Usenet: 333 groups (274 pl.*, 58 alt.* and soc.culture.polish) as .usenet.xz files, plus a galaxy.7z of about 1.1 GiB. The usenetarchive README lists it as the Polish archive, and the author's address in the license is on the same domain. I downloaded all of it, including pl.pregierz, which alone is 2.5 GB unpacked, and the whole collection takes 66 GB on disk.

To read them I found usenetarchive by wolfpld, a toolkit that turns raw newsgroup archives into indexed ones you can browse and search locally. wolfpld is also an old PLD Linux Distribution developer, and PLD has a special place in my heart.

WHY UAT ARCHIVES ARE NICE

The README starts from a blunt premise: Usenet is dead, old discussions are rotting in Google Groups, and there is no easy way to get the data, browse it or search it. The archive format has a few nice properties:

• messages are compressed individually for instant access, with a dictionary shared across the archive, so the total is smaller than the raw messages
• search uses a word lexicon with hit tables, a design the README compares to Google's original paper
• duplicates, stray messages from other groups and (optionally) spam are filtered out, and everything is transcoded to UTF-8, including messages with broken headers or bad encodings
• thread connectivity is precalculated, and missing links can be rebuilt by looking for quoted text in other messages
• archives are memory-mapped, so tbrowser needs about 10 bytes per message, around 25 MB for a group with 2.5 million messages

THE MISSING BROWSER

usenetarchive comes with tbrowser, a curses TUI, and a small web component that opens a message in the browser if you already know its message ID. Fine for a lookup, painful for browsing: I wanted to pick a group, scroll through threads and search.

So I added galaxy-web, a browser UI for the archives, as a small contribution to the project (PR #3). It has:

• a list of newsgroups with message and thread counts
• paginated thread lists, 50 per page
• nested thread views with permalinks for single messages
• full-text search across all groups or inside one, using each archive's built-in lexicon, so no extra index is needed
• message-ID routing, so links in old posts can resolve

The review included testing against a real galaxy of 333 newsgroups, plus fixes for buffer overflow protection and installation.

SCREENSHOTS

The group listing, a group's thread list and a threaded message view. Click to enlarge.

SETUP

Build the tools first. The CMakeLists.txt wants CMake 3.29 or newer, a C++17 compiler and these libraries via pkg-config: libcurl, OpenSSL, ICU, ncursesw, GMime 3 and lz4. Tracy is pulled in by CMake itself. On macOS I already had all of them from Homebrew (cmake, pkgconf, curl, openssl@3, icu4c, ncurses, gmime, lz4), so this was enough:

git clone https://github.com/wolfpld/usenetarchive.git
cd usenetarchive
cmake -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build -j

That produces tbrowser, web, galaxy-util and galaxy-web in build/.

The simplest way to read one archive is tbrowser. It opens a .usenet file and remembers it between sessions:

build/tbrowser ~/Usenet/archives/pl.pregierz.usenet

To browse everything together you need a galaxy: a directory with cross-reference data for a set of archives, which is what lets the tools show crossposts and followups across groups. Create a directory, put an archives file in it with one archive path per line, and run galaxy-util on it:

mkdir ~/Usenet/galaxy
ls ~/Usenet/archives/*.usenet > ~/Usenet/galaxy/archives
build/galaxy-util ~/Usenet/galaxy

Use absolute paths in the archives file (the shell expands ~ for you), because the tools open them as written, relative to wherever you run them. For my 333 archives, galaxy-util filled the directory with about 2.6 GB of index files (msgid, midhash, midgr and friends). Indexing 66 GB of archives takes a while. All archives must be present while it runs, but afterwards they are only needed for reading.

galaxy-web reads a small INI file:

[server]
bind = 127.0.0.1
port = 8119

[galaxy]
path = /home/me/Usenet/galaxy

Start it with:

build/galaxy-web ~/Usenet/galaxy/config.ini

With this config the archive is at http://127.0.0.1:8119/. Both bind and port are optional and default to 127.0.0.1 and 8120. The server runs in the foreground, so use tmux or run it in the background if you want it to keep going.