Hi all, an idea for future Pivot releases, based on what I see/use on other modern frameworks like Grails, Griffon, Play, Rails, to give to us (and our users) a more Installable Internet Applications (IIA) feel, even for Pivot itself ... and for example to run Tutorials and Demos as Standalone applications without having to deploy pivot wars in a local Tomcat. So for example the run application command could propose a list of candidate classes (maybe with the name without the default Pivot package for them, if unique). Note that for this we should even deploy pivot-demos and pivot-tutorials jars, and not only inside wars.
Proposal: add a pivot-shell jar and a bootstrap shell script (one for Windows and another for Unix/Linux) that set the environment to Pivot jars (and maybe others if present in the right folder) that could give us (and our users) to run Pivot tasks simply, for example some command: run-appl, run-applet, run-webstart, alias (cache a simple name for a full class name), install (copy the given jar under a .pivot-shell/lib folder under user profile), and maybe even create-app from some templates, etc ... Note that single commands usually are searched in dedicated folders (and/or jars) for better modularity. Ok, I know this is a lot of stuff, and for example most frameworks providing this feature are using Groovy to simplify (a lot) things ... but in a mid/long term vision could be useful. Play for example goes further and embed a Python distribution inside it, while I've seen other frameworks distribute only shell scripts (for example in Ruby) without the interpreter. But of course I don't want to bloat our binary distribution with many MB of things ... and to have problems with licenses by redistribution (contact legals on jira if needed). So a first approach could be to have some minimal features inside a jar, and some (OS native) shell scripts to bootstrap this (non-interactive) Shell. And if possible inside pivot-shell jar put some utility classes to scan Pivot classes and cache (in a config file under the profile) all that are runnable as Application, and Applet, etc ... To start I could create it in one of our Apache-Extras projects, and see later if useful to move in a pivot-shell project in core Pivot ... and create a binary distribution for it to go over the Pivot binary distribution, in overlay mode. Comments ? Volunteers ? Of course I know that some proposed things are not possible without a great effort (so probably it's not worth the case), but for the moment I'd like to share these ideas ... What do you think ? Bye, Sandro -- View this message in context: http://apache-pivot-developers.417237.n3.nabble.com/Pivot-Shell-tp3313582p3313582.html Sent from the Apache Pivot - Developers mailing list archive at Nabble.com.
