Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.python > #45586 > unrolled thread

Re: mutable ints: I think I have painted myself into a corner

Back to article view | Back to comp.lang.python

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: mutable ints: I think I have painted myself into a corner Cameron Simpson <cs@zip.com.au> - 2013-05-20 15:57 +1000

#45586 — Re: mutable ints: I think I have painted myself into a corner

On 19May2013 09:01, Peter Otten <__peter__@web.de> wrote:
| Cameron Simpson wrote:
| 
| > TL;DR: I think I want to modify an int value "in place".
| > 
| > Yesterday I was thinking about various "flag set" objects I have
| > floating around which are essentially bare "object"s whose attributes
| > I access, for example:
| > 
| >   flags = object()
| >   flags.this = True
| >   flags.that = False
| > 
| > and then elsewhere:
| > 
| >   if flags.that:
| >     do that ...
| > 
| > Nice and readable, but I thought to myself: so bulky!
| 
| Plus, it doesn't work:

Yeah, sorry. My "old" code was a trite subclass of object.
The new broken code is a subclass of int.

| > But setting up a flags object? What I _want_ to write is code like this:
| > 
| >   Flags = BitMask('this', 'that')
| > 
| >   # set default state
| >   flags = Flags()
| >   flags.this = False
| >   flags.that = True
| >   ... iterate over some options ...: flags.this = True
| > 
| > and there's my problem. This would modify the int in place. There's
| > no way to do that. For the base type (int) this makes perfect sense,
| > as they're immutable.
| > 
| > Before I toss this approach and retreat to my former "object"
| > technique, does anyone see a way forward to modify an int subclass
| > instance in place? (That doesn't break math, preferably; I don't
| > do arithmetic with these things but they are, after all, ints...)
| 
| No, but you could make
| 
| flags = Flags(this=False, that=True)

Yes, that is true. Flags would need to be a factory function computing
the corresponding bitmask value and then returning my new int, but
it would work. It still wouldn't let me go the whole way of modifying
the flags after instantiation.

Cheers,
-- 
Cameron Simpson <cs@zip.com.au>

Carpe Daemon - Seize the Background Process
        - Paul Tomblin <ab401@freenet2.carleton.ca>

[toc] | [standalone]

Back to top | Article view | comp.lang.python

csiph-web