Newbie can't figure out documentation practices

[Fernando Perez]

Sean Ross wrote:

> Here's another way:
> 
> class dotdict(dict):
>     # credit: name 'dotdict' taken from Alex Martelli
>     def __getitem__(self, key):
>         try:
>             instance, attrname = key.split('.')
>         except ValueError:
>             return dict.__getitem__(self, key)
>         return getattr(dict.__getitem__(self, instance), attrname)
> 
> class MyClass:
>     def __init__(self, z):
>         self.z = z
>     def test(self, other):
>         info = \
> """
> var x = %(x)s
> var y = %(y)s
> var self.z = %(self.z)s
> fun x+y+self.z = %(x)s+%(y)s+%(self.z)s
> var z from other object = %(other.z)s
> z from self plus other = %(self.z)s + %(other.z)s
> """
>         x, y = 3, 4
>         infodict = dotdict(locals().copy())
>         infodict.update(dict([('self', self), ('other', other)]))
>         print info % infodict

But this fails the moment you need self.a.b, or self.a.b.c...  (unless you
manually update for those as special sub-cases).

I still don't see a generic way of accessing member data without jumping
through hoops.  If nothing else, a newbie should be able to expect that if 

print "x=%(x)s" % locals()

prints x's value, then

print "self.x=%(self.x)s" % locals()

should similarly work.  Principle of least surprise, "Simple is better than
complex" and all that...

Cheers,

f.

ps.  As I was going to hit send, I got this from J. Collins by mail:

class evaldict:
   def __init__(self, dict):
       self.mydict = dict
   def __getitem__(self, key):
       return eval(key, self.mydict)

This one works generically for arbitrarily nested stuff.  This is, I think,
a variation on a trick I've seen before.  It's definitely a clean, nice
solution, but considering the 'wartiness' of this issue (IMHO ;), I'd
rather have a canonical solution for it in the language.

But I've beat this horse already in the past with no success,  so I'll go
back to my cave :)