Re: [thinking out loud] adding refactoring support to the JDE.

[Richard den Adel]

I am using JRefactory right now, it is open source. JRefactory is 
written in Java and currently supports integration with JBuilder and 
some other IDE's.

I am not a lisp wizard, but it might be possible to integrate JRefactory 
with JDE using the beanshell or something simular. JRefactory currently 
supports refactorings based on a UML model generated by JRefactory the 
first time for a package.

You can take a look at it at this page :
http://jrefactory.sourceforge.net/chrisdown.html

JRefactory depends on `pretty printer', which can modify your source 
code according to a code standard.

Richard.

Kevin A. Burton wrote:

> OK.
> 
> I think some of you have heard of refactoring.  It has really taken off recently
> as another technique for improving the quality of existing code.
> 
> While I think that is well and good, the use of refactoring tools can save
> developers TONS of time.  I would really like to see some type of tool support
> for this in the JDE even if we have to write it ourselves.
> 
> A good example of how much time it takes is a standard Java refactor of "rename
> package".  Just look at how much work this is:
> 
>   - create the new directory
>   - update the package statement.
>   - remove the old directory from cvs and add the new directory
>   - go over all files and change their import statement.
>   - now go over all your additional projects that used this package and update
>   - *their* import statements.
> 
> Last night I had to change 5 packages and this took 40 minutes.  A refactoring
> tool could have done this in 3. :(  It is not like this is rocket science but I
> would rather spend time working on algorithms than renaming files and making
> sure everything compiles.  I would estimate that I spend 60% of my development
> time on refactoring efforts.
> 
> I have given a good deal of thought to this over the last few days.  There are
> no decent Open Source/Free Software refactoring tools available.  Even the
> proprietary ones aren't that great.  This is good news as it means we have a
> good opportunity to create a solid tool.
> 
> I think there are also some issues here though:
> 
> - what language do we write it in?
> 
>    - writing it in elisp has some advantages because it would work natively and
>      have support for the JDE.  There are also some drawbacks:
> 
>       - other Java developers on outside IDEs (netbeans) won't be able to use
>         it.  At the very least even if these aren't emacs people they are still
>         smart and should have some feedback and could contribute to
>         development. 
> 
>       - We will have deal with the non-threaded nature of Emacs.  Emacs will be
>         locked on IO when dealing with files.  It would be nice if I could kick
>         off a refactor and then go about reading my e-mail again.
> 
>    - writing it in Java would be good but:
> 
>     - we would have to use a regexp package for some of the files.  The good
>       thing is that a lot of this is already in gnu-regexp.
> 
>     - some logic would have to be dedicated to parsing information from .java
>       files.  Where a method starts, what its params are, etc.  Semantic already
>       give this to use.
> 
> I think I have convined myself that it should be in java with a lisp interface
> into the JDE.
> 
> Anyway.  Is there anyone out there interested in working on this with me?
> Hopefully someone who has the refactoring book too :).  I think we could at
> least work on the architecture in our spare time and then start scratching our
> individual refactoring tool itch.  I want to work on "rename package" first.
> 
> Kevin
>