> -----Original Message-----
> From: Paul Kinnucan [mailto:[EMAIL PROTECTED]]
> Sent: Tuesday, May 02, 2000 2:34 PM
> The following is an example of how the speedbar will look:
>
> + A.java
> - B.java
>   - Package
>     com.mycompany.pkg1
>   - Types
>     - class B
>       - method1 ()
>         - (arg1
>            @ String
>         + |arg2
>         + )arg3
>         @ void
>       + method2 ()
>       + field1 =
>       - field2 =
>         @ BeanInfo
>       + field3 =
>       - class InnerClassA
>         + InnerClass AA
>         + method1 ()
>         + field1 =
>       + class InnerClassB
>       > Frame
>   - Misc
>     import java.lang.String
>     import java.awt.Frame
> + C.java

This looks very groovy and useful. My one big concen is that this looks too
'deep'- with inner classes you get a lot nesting and requiretd
clicking/keystrokes to navigate to where you would like.

How about instead:
+ A.java
- B.java
  - Package
    com.mycompany.pkg1
  - class B
    - method1 ()
      - (arg1
        @ String
      + |arg2
      + )arg3
      @ void
    + method2 ()
    + field1 =
    - field2 =
      @ BeanInfo
    + field3 =
  - class B:InnerClassA
    + method1 ()
      + field1 =
  + class B:InnerClassA:InnerclassAA
  + class B:InnerClassB
  > Frame
  - Misc
    import java.lang.String
    import java.awt.Frame
+ C.java

This way things do not get quite so deep - there is no 'Types', and inner
classes are considered on the same level as the outer class, however with a
prefix to illustrate the heirarchy. The one disadvantage is that the name
for inner classes quickly becomes very long, to the point that the speedbar
'form factor' as a side bar affects things.

Perhaps instead of
  + class B:InnerClassA:InnerclassAA
you can have
  - inner InnerclassAA
and contained element
    + outer B:InnerClassA

>
> The new version of speedbar does not support expansion of
> methods. However,
> Eric and I have discussed the desirability of this capability. The benefit
> I can see would be the ability to navigate to variables and
> classes defined
> within methods. The tradeoff would be greater time to parse the buffer.

Would it be feasable to have this as an option then? If parsing time is
tolerable I would much rather have this feature, but 'tolerable' is a term
that is hard to measure across different people. :)

-David Waite


Reply via email to