[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
Re: [jetty-dev] Jetty Plugin Functionality / Getting rid of Jetty Hightide release at codehaus
|
All,
I've been mulling this over for a few days and I still think that we should stick with our existing disk layout and not add a new plugins structure (at least not for libs and configs).
I see no need for creating $JETTY_HOME/plugins/foobar/lib when we already well support $JETTY_HOME/lib/foobar. The current distro already has lib modules for
annotations
jndi
jsp
jta
monitor
policy
The hightide distro add the extra lib modules
jdbc,
setuid,
spring
jta/atomikos
This modular system is well supported by start.jar. For example if you mkdir libs/wibble, then wibble is listed as an available OPTION when you do "java -jar start.jar --help"
Similarly, --help lists the available configurations in etc and then lists the defaults for both OPTIONS and configs listed in start.ini
While this mechanism is perhaps not perfect, i do not think we are able to discard it as it has been established for a while. So any module system is going to either have to extend on this existing mechanism or run in parallel to it. I think having parallel module systems will just be totally confusing, especially if one of the module systems ends up being a module inside the other module system.
My strong preference is to improve the capabilities of the existing module system - specifically add the ability to download new modules.
For the module download feature, I think the maven repository is the obvious place to download modules from, which gives us the benefit of being able to use the existing foobar-VERSION-config.jar artefacts to download and apply configuration. Ie we already have modules, we don't need to invent new module packaging, we just need a way to download and unpack the modules that we already have.
I do understand the elegance of having all the files related to a module under a plugin/foobar directory, but I just do not see that as a good enough reason to throw away the existing modularisation and to duplicate that effort.
cheers
cheers