Open-source project
xiaoyifang/goldendict-ng avatar
xiaoyifang/goldendict-ng

goldendict-ng: the default branch is called staged, and two of its ten listed features are maintenance claims

The Next Generation GoldenDict

3,162 stars178 forksC++NOASSERTION

At a glance

What is it?
A dictionary lookup program for three desktop platforms that continues an older well known project, licensed under the GNU General Public License and shipped as daily builds plus a package manager export. The comparison list it publishes against its predecessor is the most interesting document in the file, partly because two of its ten entries are not features at all and partly because the performance numbers arrive with no conditions attached.
Who is it for?
goldendict-ng is a sensible choice if you want a desktop dictionary with large-dictionary and high-resolution support and you are willing to read the documentation site rather than the readme. Three things to check first.
Can I use it commercially?
Check first. The repository uses a licence we do not classify automatically, so read its LICENSE file before any commercial use.
Is it still maintained?
Yes. The repository last received commits 1 day ago.
What is it written in?
Mainly C++, according to GitHub's language statistics.

Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.

From daily snapshots

Star history

  • xiaoyifang/goldendict-ng3,162 ★ · +128
Star history
Repository2026-09-242026-11-15
xiaoyifang/goldendict-ng3k3.2k

128 Stars in the last 52 days, measured from the daily GitHub snapshot. Compare with other projects

Editorial analysis

The default branch is called staged and the tags carry commit hashes

Two naming decisions make this repository unusual to navigate. The branch you get by default is not called main or master but staged, which tells you something about how releases are cut from it but tells a newcomer nothing about which branch to read. And the release tags are not semantic versions. The two most recent daily builds are named with a date and year in the major position, an alpha qualifier, and a short commit hash, and the third is not a build at all but a package manager export archive whose tag name is a full forty-character commit hash. So there is no way to ask for version 1.2.3 here, and no way to compare two releases by their names. A plain version file at the repository root appears to be the actual source of truth for the version, which is a fourth answer to the same question.

Two of the ten comparison entries are not features

The section listing what this project adds over its predecessor runs to ten bullets. Eight of them are things you can use: a web engine with current markup and styling support, dictionaries larger than four gigabytes, high-resolution display support, a full text engine built on a specific search library, the current version of the desktop toolkit, integration with a spaced-repetition flashcard program, a dark theme, and performance work. The other two are not features. One says the project has a lot of bug fixes and improvements. The other says the code was generally improved to make development easier. Both describe the passage of time rather than the software. Counting them, the file presents a ten-point improvement list in which two points are that the project has continued to exist, and the daily release cadence is listed as a feature alongside them.

The performance numbers arrive without hardware or conditions

Two quantities appear in the list. One is that dictionaries above four gigabytes are supported, which is a capability rather than a speed claim and is the kind of limit users hit for real. The other is headword counts, given in parentheses as an example of what performance optimisation achieved, at more than ten million. Neither figure comes with a hardware description, a dictionary format, a memory figure, or a time. The file does link a code coverage badge from a hosted static analysis service, so the project is instrumented enough to have measured something, and it publishes a badge but no result. A reader deciding between this and another dictionary has to take both numbers on trust.

The licence is stated in the readme and missing from the metadata

The licensing section says the project is under version three or later of the GNU General Public License and that a copy is in the licence file at the root of the repository, which it is. The licence recorded for the repository itself is empty, marked as no assertion rather than as a named licence. So the terms are unambiguous for a human reader and invisible to tooling, which is the wrong way round for a program distributed as binaries through a package manager. It matters more here than it would for a small library, because the package export release described above is precisely the path by which someone installs this without reading a readme, and a machine resolving that package will find no licence metadata at all.

Three languages, three linters, and a generated changelog

The root of the repository carries quality tooling for three different kinds of file. There is a C++ formatter and a separate C++ static analyser configuration, a Markdown linter configuration for the documentation, and a JavaScript formatter configuration with an ignore file, which is a little surprising in a desktop C++ program until you notice the documentation site and the translation tooling live in the same tree. On top of that there is a static analysis properties file matching the hosted analysis badge in the readme, an editor configuration, and a configuration file for a changelog generator that builds release notes from commit messages. That last one pairs with the daily build cadence: the releases are frequent enough that writing release notes by hand would not be viable.

Binaries travel through a package manager manifest, not a download page

Two files at the root are a manifest and a configuration for a binary package manager for the language toolchain this project is written in, and the most recent release is an export archive produced by that tool. So the distribution story is not a page of installers but a manifest of dependencies plus a cache of prebuilt binaries, which is a more reproducible arrangement and a less obvious one. Alongside that sit a vendored third-party directory and a submodules file, so part of the dependency graph is committed and part is fetched, and a Qt resource file listing the assets compiled into the binary. That last detail is the one to be aware of when comparing formats: anything not in the resource file has to be found on disk at runtime.

Translations run through a hosted platform and the project continues an older one

The readme credits a hosted translation platform with a badge and tells contributors to use it to translate the interface, and the tree carries a translation configuration file and a locale directory to match. It also says the original project was developed at its own website and in its own repository, which makes this a continuation rather than a fork of convenience, and the file carries a credits file alongside the licence. The final line thanks a commercial tool vendor's open source support programme, with a link to it, which is a sponsorship acknowledgement rather than a technical dependency. Taken together, the three facts a newcomer needs are that this continues an older project, that its interface is translated by volunteers through a third-party site, and that its releases come out of a daily pipeline rather than a curated one.

Editorial conclusion

goldendict-ng is a sensible choice if you want a desktop dictionary with large-dictionary and high-resolution support and you are willing to read the documentation site rather than the readme. Three things to check first. The readme does not tell you which dictionary formats are supported, only that there are many and where to look, so confirm your formats before installing. The performance figures are stated without hardware, format or test conditions, so treat them as a direction rather than a measurement. And the version you get depends on which branch and which tag you pick, because the default branch has an unusual name and the release tags carry a date and a commit hash rather than a semantic version.

Frequently asked questions

what is goldendict ng

An advanced dictionary lookup program for Linux, Windows and macOS, licensed under version three or later of the GNU General Public License. It continues an earlier project of the same name, supports dictionaries larger than four gigabytes and high-resolution displays, and is built on a full text search engine and the current version of the desktop toolkit.

What format does GoldenDict use for dictionaries?

The readme does not enumerate the formats. It says the program supports many of them and links to a formats page on the documentation site, which is where the list actually lives. Both this project and the one it continues point readers there rather than restating the set.

How do I add a dictionary to goldendict-ng?

The readme does not give instructions for adding one. It links to the documentation site, a formats page, an issue tracker for bug reports and a discussions area, and it points developers to a separate developer guide for working on the code itself.

How is goldendict-ng different from the original GoldenDict?

The readme lists ten points. Eight are capabilities, including large-dictionary support, high-resolution support, a current desktop toolkit, a full text search engine, flashcard integration and a dark theme. Two are not features at all but state that the project has accumulated bug fixes and general code improvements.

How are goldendict-ng releases named?

Not by semantic version. The daily builds use a date-based number with an alpha qualifier and a short commit hash, and one release is a package manager export archive whose tag is a full forty-character commit hash. A separate version file at the repository root appears to be the real source of the version, and the default branch is named staged.

What license is goldendict-ng under?

The readme states version three or later of the GNU General Public License and points to the licence file at the repository root, which exists. The licence recorded for the repository itself is empty, so tooling sees no licence at all.

Official sources

  1. Issues
  2. Project website
  3. README
  4. Releases
  5. xiaoyifang/goldendict-ng on GitHub
Add this badge to your README

If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.

Add this badge to your README

markdown
[![Hysen Labs](https://hysenlabs.com/badge/xiaoyifang-goldendict-ng.svg)](https://hysenlabs.com/projects/xiaoyifang-goldendict-ng)

Star history chart for your README

markdown
[![Star history](https://hysenlabs.com/chart/xiaoyifang-goldendict-ng.svg)](https://hysenlabs.com/projects/xiaoyifang-goldendict-ng)

Was this useful?