CyberEther v1.6.0 focuses on extensibility and ML workflows. Highlights include an experimental Python runtime and new Python block for scripting inside flowgraphs, a new inference block with ONNX support in container builds, a modernized plugin v2 interface, and the new retained Sakura component system. Distribution improves with CUDA containers and Linux AppImages, alongside a documentation overhaul and numerous UI fixes. Welcome to first-time contributors @radonnachie, @F5OEO, and @paoloo!
What's Changed
- refactor: improve documentation coverage by @luigifcruz in #176
- feat: add experimental python runtime by @luigifcruz in #178
- feat: show tensor metadata in output pin tooltips by @luigifcruz in #182
- fix: use bundled catch2 and avoid installing bundled dependencies by @luigifcruz in #183
- fix: improve ui by adjusting node badge and content width by @luigifcruz in #184
- feat: add frame block by @luigifcruz in #186
- fix: node rendering issues with z-order and surface interception by @luigifcruz in #188
- feat: add doubles support in file_reader, ones_tensor and multiply_constant blocks by @radonnachie in #189
- feat: add CUDA containers and Linux AppImages binaries by @luigifcruz in #190
- refactor: parse flowgraph metadata at registration without pyyaml by @luigifcruz in #192
- refactor: new plugin v2 interface by @luigifcruz in #193
- docs: add Quick Start guide by @luigifcruz in #194
- fix: websocket when using local address and less strict on certificate by @F5OEO in #191
- fix: cyberether dependency in example plugin by @radonnachie in #197
- fix: flowgraph node config dropdown padding alignment by @luigifcruz in #198
- feat: add renaming blocks interface by @luigifcruz in #199
- fix: node resizing when incomplete by @luigifcruz in #200
- feat: add inference block by @paoloo in #175
- feat: add default node size configuration for blocks by @luigifcruz in #201
- docs: add architecture, blocks, platforms, and tensors by @luigifcruz in #203
- feat: add python block by @luigifcruz in #202
- feat: introduce retained sakura component system by @luigifcruz in #195
- feat: replace imgui text editor with retained jetstream render by @luigifcruz in #155
- feat: add throttled compute taint by @luigifcruz in #204
- chore: remove docs content package by @luigifcruz in #206
- docs: add plugin installation page by @luigifcruz in #207
- docs: add tensor interoperability direction matrix by @luigifcruz in #208
- feat: improve python runtime resiliency by @luigifcruz in #209
- feat: enable ONNX inference in container builds by @luigifcruz in #205
- feat: share source-piece expansion with Python by @luigifcruz in #210
- chore(version): bump version to 1.6.0 by @luigifcruz in #211
Known Issues
- The new Markdown rendering on the Vulkan backend is not yet optimized for performance. This will be addressed in the next version.
New Contributors
- @radonnachie made their first contribution in #189
- @F5OEO made their first contribution in #191
- @paoloo made their first contribution in #175
Full Changelog: v1.5.0...v1.6.0