Port an IDE (e.g. Eclipse) to JNode
| Project: | JNode Distro |
| Component: | Miscellaneous |
| Category: | feature request |
| Priority: | normal |
| Assigned: | Unassigned |
| Status: | postponed |
Jump to:
Description
I would very much like to see not only a compiler, but also an IDE (like Eclipse) working on JNode.
Is there a potential application for this? Like pretty much compile and run directly on JNode?
In my vision it would boost the usability of JNode and speed up development on the platform.
- Login to post comments
Of course we also want to that
Building JNode inside JNode is one of our goal. But there is much things to do to reach this goal (from higher to lower priority) :
- more stability for JNode
- a compiler (the integration of Eclipse Java Compiler, aka ejc is in progress)
- ability to use ant (it's in progress)
- a working filesystem implementation like fat or ext2 (in progress ...)
- for an IDE : a working swing implementation and JNode peers for each component.
- for eclipse : a port of SWT (the widget toolkit used by eclipse) to JNode
- Maybe something else ...
As you can see, there is a lot of work to be done. So, if you want to help us, you are welcome !
Fabien
my blog (in english and french)
Continuing this thread
First I'd like to thank all the JNode Developers for validating most
if not all of my concepts regarding the validity of Java being utilised to develope an OS.
Now I have a couple of questions. When you are discussing using ecj
are we discussing the possibility of utilising the entire Eclipse
IDE Ecosystem ?
Because if that is the case it really would open JNode to a host of
options and developers.
There are a couple of areas of I'd like to look into too. First the Java-Gnome bindings and refactoring JD4X Java Desktop for X-Window.
Now most Unix based OS's have built in ManPages it would be worthwhile
discussing the creation of ManPages for JNode but the concepts behind
that model of ManPages does not take into account Web 2.0 .
If you were paying attention to the Nautilus File Manager for the Gnome Desktop Easel had a concept of NUE-Networked User Enviroment.
What if we were to combine NUE with ManPages ?
We'd have NUE-ManPages some of the features would be to connect end user to Tutorials,End User Groups,IRC,Bugzillia and other Web based Services,Developers see fit.
I do not have enough time right now to share all of my concepts but
I will say this much I'd like to drive Web 2.0 right down into the File System and to have a File System similiar to GLS Cube.
#1
Adding a proper title and changing 'project'
#2
It will be much easier to port NetBeans IDE to JNode. NetBeans is pure JAVA. No need to port any native libraries like SWT.
#3
And the standard response is that this is all hypothetical unless someone actually tries to do a port.
#4
As good of an idea as this sounds. You have to consider the fact that yes, you may be able to port all the apps to build jnode in jnode, but is this even going to be possible given the current state of the core? I personally think it would be mayhem. With the introduction of many new command line apps, and many more on the way, we should be able to actually test jnodes memory/scheduler/compiler to the point where we can find the major bugs and issues, and also be able to tell if what we are doing is having any effect at all.
Before i left for work last fall, i was working on the memory management that needs a lot of attention. The problem we faced was, without creating synthetic benchmarks, there really was no concrete way to tell if what we were doing was having a better or worse effect.
Before we start thinking about large apps that want to pound the resources of its OS, lets start a bit smaller shall we? IDEs are monstrous things, lets start with a text editor, and then work our way up...
#5
Cluster is right. Attempting to port a large, resource hungry application like an IDE would be source of endless frustration given the current state of JNode. The JNode native compiler is slow and outputs poor code, the memory manager is slow and unreliable, JNode has no virtual memory support, the file systems are unreliable, and so on. So even if one succeeded porting an IDE, it would be painful to use compared with running the IDE on (say) Linux. It certainly wouldn't speed up development in the short term!
#6
Not that we want to discourage anyone, but there are hopes and dreams, and then there is reality.
What i would recommend for the short term, is having a full complimentary command line from which to use jnode. The shell is growing more and more mature, and as it does, it is going to make testing jnode, with a full set of commands, much easier. Once the shell and its commands are stable, then the core of jnode can be tested and flushed out to a much better design.
#7
Im also marking this postponed as per the above reasons.