Re: repo to work on python scripting support

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Daniel" == Daniel Jacobowitz <drow@false.org> writes:

Jim> On Tue, Mar 25, 2008 at 11:18:12AM -0700, Jim Blandy wrote:
Jim> I don't know the most Pythonish way of doing things, but it seems to
Jim> me that Python could register not functions, but objects, to be
Jim> callable with $().

Daniel> Hmm, I've worked with some packages that did similar things using
Daniel> mandatory docstrings.

We could use decorators.

    http://wiki.python.org/moin/PythonDecorators

Well, assuming they exist.  I'm not really that up on the current
state of Python.


For this particular thing I would be just as happy with some explicit
constructor thing:

class Strcmp(GdbFunction):
  def __init__(self):
    Strcmp.__init__(self, 'strcmp', 'ee')

  def invoke(expr1, expr2):
    lalala


You could also use a real data structure and named constants:

    Strcmp.__init__(self, 'strcmp', (gdb.EXPRESSION, gdb.EXPRESSION))

.. or what have you.

Aside from the interactive spec, the current gdb-python support for
command objects already works this way.  A command subclass has to
pass in the user-visible command name at construction time.

FWIW here's the code I wrote to override "edit" and turn it into an
annotation-based command:

class Edit(gdb.Command):
    def __init__(self):
        gdb.Command.__init__(self, "edit", gdb.COMMAND_NONE)

    def invoke(self, arg, fromtty):
        if gdb.show('annotate') > -1:
            loc = gdb.decode_line(arg)
            if loc:
                loc = loc[0]
                if loc.symtab.to_fullname():
                    print "\n\x1a\x1asource %s:%d:0:beg:0x0" \
                        % (loc.symtab.to_fullname(), loc.get_line())

# Install it.
Edit()



Just to be clear, I don't really want to consider anything in this
area as fixed.  Advice and input about how everything should fit
together is really helpful.

Tom