J2 decorator grammar

Michael Sparks zathras at thwackety.com
Sun Aug 22 02:20:16 CEST 2004

On Fri, 20 Aug 2004, Robert Brewer wrote:
> Did you ever fix the scope issue?

I've fixed the scoping issue.

I'll clean up the changes I made to get a minimal patch against 2.4a2, and
then put it somewhere handy. Now that I've got a much clearer idea of how
the python source tree hangs together adding in the short form should be
relatively simple.

The current patch uses "decorate" as the keyword. Changing that of course
is utterly trivial - it's also something I think *should* be changed.

The reason for suggesting this change is for one simple reason - python
makes often use of the idiom "decorate-dosomething-undecorate" - eg in
sorting. Using decorate would probably clash heavily with people's code.

The closest word that strikes me that has a similar meaning to
decorate, using and applying would be "decorating". If anyone has
a closer suggestion to a short single word meaning "decorate using",
it'd be great.

> Might it have something to do with:
> static void
> symtable_node(struct symtable *st, node *n)
> {

It certainly did! Your suggested change was necessary, however it was
incomplete - the child to traverse also needed changing. (I really
should've thought of that sooner - of course it's no longer zero! :)

ie this
>                       symtable_node(st, CHILD(n, 0));
>                       symtable_node(st, CHILD(n, 4));

So THANK YOU for pointing out this part of the code :)

Currently I have a file that looks like this, which is completely legal,
and does what you would expect for this syntax:
def memoise(f):
   return f

class Foo:
   def Hoo(who):
      print "You", who
      if 1:
         def Print(what):
            print "Ya", what



No matter what syntax we end up with, all of these syntactic sugar changes
rely on the original decorator code, and I've been very pleasantly
suprised at just how clean (and well documented) the python source is for
a project of it's size.

As I say I'll post the cleaned up patch later.



More information about the Python-list mailing list