Hello, I have some good news on the UIMACPP front.
Got a multi-stage docker build working for the project: https://github.com/apache/uima-uimacpp/blob/main/Dockerfile This compiles the whole project, including working pythonnator and perltator, then produces a smaller image just with the compiled libraries and their runtime dependencies. This image will hopefully be available on Docker Hub as "apache/uimacpp". A docker image in the examples then can build on top of that to install additional compiler dependencies and produce the annotator shared library ("dll") out of the container: https://github.com/apache/uima-uimacpp/blob/main/examples/Dockerfile The beauty of the approach is that the annotators can be run against the apache/uimacpp by mounting some folders containing the annotator shared library and its descriptors. And in the case of pythonnator and perltator, all that is needed are the python or perl scripts. (See the docker run incantations in https://github.com/apache/uima-uimacpp/tree/main/examples/scriptators). What is missing for a release: * Some debugging with XMI handling in runAECpp (the tests all pass but it fails to load the XML correctly). * Rewrite the documentation to explain the above and remove stale docs. * Get the docker hub approval https://infra.apache.org/docker-hub-policy.html <- I think I need the chair help with this Other: * Add issues for planned improvements: * python install system-wide the pythonnator * perl install system-wide the perltorator * Handle aggregate analysis engines * add the UIMA C++ back to the website (it currently points to https://uima.apache.org/doc-uimacpp-huh.html ;-) If anybody want to take it for a spin, clone the repo and on its root do sudo docker build . -t apache:uimacpp Any feedback is most welcomed! P
