On November 16, 2018 12:00:30 AM UTC, "Rémy Maucherat" <r...@apache.org> wrote: >On Fri, Nov 16, 2018 at 12:35 AM <ma...@apache.org> wrote: > >> 2. Remove org.apache.tomcat.jni and replace with the minimum >necessary to >> interface with OpenSSL and clones. >> + We might want to park this one until we see what is available >direct >> from >> + the JRE with project Panama. >> > >Ok, I guess the OpenSSL code is here and works already, it could simply >be >a bit nicer. So no big loss. But I don't see any reason why that >approach >wouldn't be an order of magnitude more complex than the simple JNI >wrapper >we would have. Besides building it I guess (but the JDK req could be a >problem).
I'm still experimenting but if Panama does what I think It will, it would remove the need for the native wrapper altogether. I think that is worth pursuing. I agree the JDK requirement is an issue. As is the fact that Panama isn't delivered yet. The debate I was hoping to start is exactly the question you asked. Do we want to simplify native for Tomcat 10 and then make another change in what is probably the Tomcat 11 timeframe? Or do we want to leave native as is for Tomcat 10 and have one change in Tomcat 11? Given the current uncertainties I think we should switch to a simpler wrapper in Tomcat 10 but i think it is worth discussing as others may view this differently. Mark --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
