[j-devel] J - movePastCloseAndReindent bug

[V. R. <vr...@se...>]

Hi, Peter

The bug: when the point(dot) is just before a closing parenthesis it is =20
ignored and _movePastCloseAndReindent_ passes 2 parenthesis and then =20
reindents.

Basically it looks like this

(a (b<cursor>))   --> (a (b
                         <cursor>))

(a (b<cursor> )) --- (a (b )
                           <cursor>)

and this may be a quick fix (and indeed is)

diff -rN old-j-0.21.4/src/org/armedbear/j/Editor.java \
new-j-0.21.4/src/org/armedbear/j/Editor.java
6762,6769c6762,6774
<         while (pos.next()) {
<             char c =3D pos.getChar();
<             if (c =3D=3D '(')
<                 ++count;
<             else if (c =3D=3D ')')
<                 --count;
<             if (count =3D=3D 0)
<                 break;
---
>         if (pos.getChar() =3D=3D ')') {
>             // found
>             count =3D 0;
>         } else {
>             while (pos.next()) {
>                 char c =3D pos.getChar();
>                 if (c =3D=3D '(')
>                     ++count;
>                 else if (c =3D=3D ')')
>                     --count;
>                 if (count =3D=3D 0)
>                     break;
>             }
6770a6776
>=20

Still _movePastCloseAndReindent_, IMHO remains broken but in a more =20
subtle way. Invoking repeatedly movePast... from the most nested form =20
in this expr:

(a (b (c (d))))

gives us:

(a (b (c (d)
          )
       )
    )

Since I given that matter a bit of thought I guess it can done like =20
this. Pseudocode follows:

if (<lastCommand was movePastCloseAndReindent>) {
   <Undo>
   globalParenCount++;
} else {
   globalParenCount =3D 1;
}
count =3D globalParentCount;
<find count-th closing parenthesis>
<new line and indent>

In my "mental model" of J, it does the right thing. YMMV.

Regards

--
   vasile