Re: Fixing a sample from K&R book using cake static analyser

Kaz Kylheku:

Also:
>
You generally want parallel elements at the same level of
indentation.

Who, me?  As a matter of fact, /I/ do.

The following structure can be grating on the eyes:
>
  if (key == node->key)
    return node;
  else if (key < node->key)
    return tree_search(node->right, key);
  return tree_search(node->left, key);

Yes.

This is just an example; I realize we are not handling the
case of the key being found.
>
This is better:
>
  if (key == node->key)
    return node;
  else if (key < node->key)
    return tree_search(node->right, key);
  else
    return tree_search(node->left, key);

Just a tiny-little-bit better, becase I the last
unconditinoal return makes sense as the default exit route.
I therefore consder the following version more expressive:

  if (key == node->key) return             node;
  if (key  < node->key) return tree_search(node->right, key);
  if (1               ) return tree_search(node->left , key);

To emphasize parallelism, and line length permitting, I
format my if-else chains thus:

       if (key == node->key)  return             node;
  else if (key <  node->key)  return tree_search(node->right, key);
  else                        return tree_search(node->left , key);

The structure being recommended by Anton is most at home
in imperative situations like this:
>
  stmt1;
  stmt2;
  if (condition1)
    return early;
  stmt3;
  stmt4;
  if (condition2)
    return early;
  stmt5;
  ...
When we have multiple cases that are all parallel and of
equivalent importance (they could be in in any order), not
so much.

This structure is perfectly suitable for order-dependent
statements, too.  And I have to use `goto' if some
deinitialisation is required before the function terminates.

--
()  ascii ribbon campaign -- against html e-mail
/\  www.asciiribbon.org   -- against proprietary attachments