Costin Manolache wrote: > jean-frederic clere wrote: > > >>>As I said, for 'chuid' functionality I prefer using a direct >>>call - I have most of it implemented using jk2, I'll >>>finish this well before 5.0 is released. >> >>I prefer to have a C wrapper that start the JVM and call methods than >>having the reverse. > > > That's perfectly fine as long as you accept that others may have > different preferences :-) > > I agree that a C wrapper to start the VM may be better than the > current .sh - but again I disagree on using JNI invocation instead > of a simple exec to start the VM ( for the simple reason that Kaffe > and GCJ don't support invocation - at least last time I checked ).
Having both in the same executable may not be necessary but a lot of native code could be shared. The C code (mostly written by Pier) is quite easy to follow and well structured. The environment to set before JNI_CreateJavaVM() or exec(java) is very similar. About invocation, I do not know about Kaffe and I will check for gcj. > > > >>I have rethinked my position about the need of the daemon interfaces >>specialy about the controler part and I am ready to +1 for moving the >>interfaces and replace it a description of methods and classes that would >>be called/instancied from a native program. But I need time (about 2 >>weeks). I will try to provide a description of the features I need. > > > If the daemon is not forcing tomcat to implement any interface - > you have my -0 on the release. > Ok but I need time. -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
