Just want to share this article: http://pureconcepture.blogspot.com/2017/07/goodbye-to-eclipse-end-of-era-for.html
It’s from 2017 and nowadays even more true. Time to switch our main IDE I guess.
Cheers, David
3 Likes
I’m no developer, but maybe it would be better first to complete the existing issues (remerge of eclipse, move to bnd) before adding a mass of other issues. 
3 Likes
I completely agree with Udo. Sadly issues like this are a “don’t care” to the average user and we are already having the impression that something is wrong with the lack of forward progress to the code. I wouldn’t take something like this on until we get back into the release rhythm again. And then try to do so in a way that doesn’t interrupt that release rhythm.
To play devil’s advocate, you can pick any software, technology, and development process and find an almost identical blog posting. I’m not saying that the recommendation to abandon Eclipse is a bad one, but one it two blog posts is not IMHO enough evidence to do so.
Personally, I’ve not been out of development professionally long enough to not have an opinion at all. I’m old enough to remember the vi vs. emacs wars (I was on the emacs side though now find myself defaulting to vim for everything). Developers are passionate about their tools. If OH could be relatively IDE agnostic, maybe defaulting the docs to only one or two and letting the community generate tutorials for setting up the rest, it would be the best if all world’s.
5 Likes
The quoted article is from 2017 and was written at a time when Eclipse core and IDE development was really having problems and some companies were turning away. There was a movement from Eclipse largely to IntelliJ (for Java development) and Visual Studio Code (for other languages like Go).
Eclipse had the same problem as other successful open source projects: there were a lot of users of the good infrastructure but very few contributors to the core development (sounds familiar …).
But a lot has happened since then. Especially Lars Vogel and his team have pushed the development of the IDE in the last two years and AFAIK the JDT and platform projects have merged.
When it comes to development environments for large-scale enterprise software (especially with mainframe backends), Eclipse remains unbeaten. You will find several products on the market that are based on Eclipse (and some also on Visual Studio) but no IDEs based on e.g. IntellJ or Visual Studio Code.
The biggest plus of Eclipse is its architecture that makes the development of powerful plugins very efficient. None of the other platforms can keep up with this at the moment.
But in general, it is good to have an IDE independent development process for add-ins. Currently, I would not use e.g. Visual Studio Code for Java development because of too many missing features. But there are other tasks where it is absolutely sufficient and if somebody prefers IntelliJ for Java development, why not.
1 Like
It is a difficult balance but it is easy to run too fast and end up in a mess that you can never get out of. I chose OpenHAB very recently as my gut feel is that some of the other platforms are running so fast they will trip up.
I would like to help but my gut feel just now is that I would take more time from the team than the value I would add.
Just thanks for all of your efforts and keep up the good work.
As to eclipse it is a side show. If it crumbles something will grow.
1 Like
I‘m not sure the problem really is Eclipse. We should get the stability we had before and maybe move from Maven to Gradle for oh3, not before.
I had a look at IntelliJ and VS Code and I don‘t think they provide better debugging capabilities than Eclipse.
1 Like
It’s about start up time and work flow as well. Debugging with a live running karaf as I have suggested in a different thread works with vscode quite good actually.
Android studio? One of the biggest installation base on earth for Java development I guess. And cloud9 and all other cloud based ide’s are usually visual studio code based. Eclipse is the loosing party here. Personally I have the same feeling like the blog post author, that eclipse feels clumsy and not well integrated.
With gradle we might have at least a better visual workflow for binding developer beginners.
Android studio?
I was talking about large scale Enterprise Development tools like Micro Focus Enterprise Developer, IBM RDz/IDz, Compuware Topaz … They are all based on Eclipse.
And cloud9 and all other cloud based ide’s are usually visual studio code based.
I know and I personally like a lot of the Visual Studio Code concepts like the LSP and Debug protocol. But these concepts are independent of the IDE and at least the LSP can be used with Eclipse.
I am also using VSC for Go development and when I am working with Markdown, Asciidoc, HTML …
However, I am not a fan of web-based IDEs development especially because of security reasons.
For Java development I would miss the refactoring possibilites, the debugger and lots of other features that are available in Eclipse (and also in IntelliJ).
Security is also a reason why I don’t expect that especially large enterprises with high security restrictions will use VSC in the near futures because the npm ecosystem is a security threat.
With gradle we might have at least a better visual workflow for binding developer beginners.
I agree. I am normally using Maven in my daily work but for beginners using Gradle might be easier.
1 Like
I’m a big fan of IntelliJ, and at least for me, it works a lot smoother than eclipse.
If I want to open a project, I just open it.
I don’t have to fight with import dialogs first.
At the end of the day, it boils down to which IDE you can best work with.
Therefore I would rather use the time to make the build system independent of the IDE rather than tieing it to a specific IDE.
The build tool should handle the build related stuff while the IDE interfaces with the build tool to resolve dependencies and manage the build.
The current buildsystem works in all IDEs.
What bugs me is that we have m2e (maven 2 Eclipse) references in our repositories.
Those should be removed IMO (all .project and .classpath files as well as the m2e maven plugin).
And of course moving to Gradle would be even better as they actually care about smooth IDE integration.
1 Like