Hi everyone, I was looking at removing the Jackson dependencies in the OpenNLP core and noticed that there is "opennlp-brat-annotator" which was moved from the sandbox to the core in OPENNLP-867.
Looking at the name finder resource, it's just a web service that doesn't actually use the model parameter and doesn’t bring much value to the core project. I even doubt, that anybody is using it.There are no tests. Therefore, I would like to propose that this component be moved back to the sandbox, as the module does not provide any value (IMHO) as it just bloats our binary distribution. If anyone actually useds it (from the core project), there is still the possibility to build from sandbox or just copy the related code to their projects. From my POV, these two classes do not offer much value. If we want to provide such a web service layer (in the future), we should adhere to Jakarta EE standard and just bundle a WAR application, which can be deployed in any EE application server to serve OpenNLP capabilities via REST. This could be started via a Maven plugin or in a standalone server context. However, this has some other limitations in terms of resource loading, but if there is interest, we could do that in a separate feature request. Gruß Richard
