Hi all, Seems that the JK2 development has been stalled for a while, cause either developers loose the interest, or it's so damn good :)
I would like to propose few things that IMO could make the JK2 a better. 1. Get rid of JNI from core and make a new 'server/jni/(iis|apace2|xxx)'. This new would require java jre on server side. Also focus only on TC5 embedded mode. It would require a lot of work like rewriting aprImpl and TomcatStarter, but it should finally work something like Jrun or ServletExec. This would be better explained as jk2jni then jk2/jni, but since they will be using the same code base... 1.1 Make APR_JNI project. Something like APR_UTIL or APR_ICONV that will give a portable JNI interface, using APR. 2. workers2.properties -> workers2.xml using apr_utils xml support. Get rid of 'assumed' properties like figuring out the context from url. Get rid of copying mappings from 'default' to virtual hosts. Of course, it would require few extra 'boring' statements in the config for each mapping. 3. Get rid of Jk2* from Apache module and use only 'Jk2On' That will register the JK2 in the same way as filter on IIS (for each virtual server). All this imply that the workers2.xml is the main config point, meaning that the same workers2.xml is operable either on IIS or Apache or any other web server. Also there are no other 'per-server' directives rather then 'on or off'. 4. Get rid of all platform specific configuration stuff like WIN32 registry, event log, apache log, and use only JK2 log. Something like I proposed two years ago, but since people tend to change its opinions, and since there was no major proposals for JK2 enhancements for two years (not counting myself), do I have a 'card banshee', or do I need to make a 'fork'? Waiting for response from someone that can 'provide' something different, or just file a veto :) MT.
smime.p7s
Description: S/MIME cryptographic signature
