[slime-devel] Re: slime-region-for-defun-function

[Madhu]

* Helmut Eller <m2k5d0spr6.fsf at common-lisp.net> :
Wrote on Thu, 25 Sep 2008 08:28:45 +0200:

| * Madhu [2008-09-25 08:00+0200] writes:
|
|> * 2008-03-13  Helmut Eller  <heller at common-lisp.net>
|> |	* slime.el (slime-region-for-defun-function): Deleted.
|> |	(slime-region-for-defun-at-point): Use beginning-of-defun
|> |	and not beginning-of-sexp.
|>
|> This breaks multiple use cases when evaluating s-expressions from lisp
|> file buffers.
|
| Which cases?

Heres One case.

#+nil
(progn
(print 1)
(print 2))

Note the progn and each enclosed sexp all begin at column 0

When I place the cursor at some point inside the sexp and do C-M-x
(say), I want the correct enclosing sexp to be sent to lisp for
evaluation

If I place it within `(print 2)' and do C-M-x I want the expression
`(print 2)' to be sent .

If I place it within `(print 1)' and do C-M-x I want the expression
`(print 1)' to be sent.

If I place it between the `progn' and (print 1) and do C-M-x I want the
whole progn to be evaluated. Instead what gets evaluated is only the
last sexp in the progn which is (print 2)

I can come up with a few other cases, but this is the basic problem.

|
|> `beginning-of-sexp' is the right thing to use here.  What motivated you to
|> change this to `beginning-of-defun'?

slime-region-for-defun-at-point does an end-of-defun . The
implementation of end-of-defun internally calls beginning-of-defun to
figure out the beginning and then goes to the end.

Since we are dealing with sexps, Note that once you go to the end of the
sexp via `end-of-defun', beginning-of-sexp will find the beginning of
the particular sexpression - whose end was determined by `end-of-defun'.

[This is why beginning-of-sexp is the correct function to call after an
end-of-defun]

Calling `beginning-of-defun' again does not find this sexpression that
you want to get a hold of but finds something else.

| beginning-of-defun can be customized with beginning-of-defun-function;
| beginning-of-sexp can't.  I'm using that to find CMUCL's
| (declaim (start-block)) forms.

How would I go about customizing it to get the behaviour desired without
breaking other uses ?

(Haven't looked yet, Thanks)

--
Madhu