Icon packs are read in TeamSpeak's own format — a zip (or unpacked folder)
of SVG/PNG art plus a settings.ini mapping icon keys to files — so the packs
of an installed client and the ones from its add-on site work unchanged.
Legacy packs without that mapping (default.zip) are resolved by their file
naming convention instead, picking the resolution closest to the drawn size.
The pack draws the tree, toolbar, menus, context menus, file browser and the
default group icons; a pack's FALLBACK option decides whether what it lacks
comes from default.zip, and anything still missing falls back to the icons
the client draws itself. Options → Design picks the pack and shows a viewer
of everything in it.
Vector art is rasterised with JSVG, and the pack search roots are shared
with the sound packs.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Three things bloated it. Nothing tree-shook the dependencies, so all of
bcprov shipped even though only its lightweight API (AES-EAX, ECDSA over
prime256v1, ASN.1, EC point math) is ever touched. Its multi-release
copies under META-INF/versions accounted for another 2.3 MB that could
never even load, since the shaded jar is not itself multi-release.
And ts3j dragged in commons-lang and dnsjava for one call each.
Enable minimizeJar, keeping our own modules and ts3j whole (they are the
entry points, and ts3j instantiates packet bodies reflectively), and drop
the versioned entries. The shade plugin has to move to 3.6.2 for this:
3.5.1 bundles an ASM that cannot read class file major version 70.
The remaining large entry is the bundled Windows Opus DLL, at 612 KB.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Add ServicesResourceTransformer so overlapping service-provider files
(e.g. java.security.Provider) are concatenated rather than clobbered,
future-proofing the single executable JAR build.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Swing desktop client (core/desktop/swing Maven modules) built on the
ts3j protocol library, included as a submodule. Native Opus voice with
voice-activation detection, push-to-talk, and audio pre-processing.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
