It is often useful to restrict the range of commands like s or g to the function one is currently editing. For C/C++ and Java,

:cmap ;tf ?^{??(?,/^}/

maps the keystrokes ;tf (this function) to the range of the function in which the cursor is currently located. It works by searching backwards for a { occurring in the first column, and then for a ( to find the start of the function. A } in the first column ends the function body.

For example, suppose you want to change the name of a function argument in the current function from i to ii:

int foo(int i,
        int j)
{
    // ...
    i++;
    return i;
}

In normal mode, type :;tf followed by the subsitution command s/\<i\>/ii/g. The command line now reads:

:?^{??(?,/^}/s/\<i\>/ii/g

Press Enter to execute the command. The function changes to:

int foo(int ii,
        int j)
{
    // ...
    ii++;
    return ii;
}

The range can be used with other commands too.

This tip assumes that:

• Your source code is indented, so that curly braces on the first column always open or close a function body.
• A function name is followed by a (. This is true in C, mostly true in C++ and Java, and true to a limited extend in Perl.

This tip does not work for constructor definitions in C++ and Java, since they can contain many parentheses, for example:

Foo(int bar, int baz):
    mBar(bar),
    mBaz(baz)
{}

CommentsEdit

I don't know a lot of people that actually put braces on the first column in Java, except for the class.

It is not the case in C++ either if you use namespaces.

It's used in K&R-Style

This is pretty cool if you forget to do va{ (or similar) first, but text-object methods should be discussed as well.