Path: csiph.com!usenet.pasdenom.info!aioe.org!news.stack.nl!newsfeed.xs4all.nl!newsfeed3.news.xs4all.nl!xs4all!post.news.xs4all.nl!not-for-mail
MIME-Version: 1.0
In-Reply-To: <CAAu18hevvTDWizx8Gt2S4MUv04KQ0uw7sZaOk-pi-ffsOtG8nw@mail.gmail.com>
References: <lcgtpf$tui$1@ger.gmane.org> <lcgut2$b8r$1@ger.gmane.org> <858utviwgs.fsf@benfinney.id.au> <lchce7$44q$1@ger.gmane.org> <52EC3C40.7080402@stoneleaf.us> <mailman.6251.1391224639.18130.python-list@python.org> <52ec84bc$0$29972$c3e8da3$5496439d@news.astraweb.com> <roy-5740C1.00250401022014@news.panix.com> <bl81skFh14iU1@mid.individual.net> <mailman.6313.1391383680.18130.python-list@python.org> <roy-C454F0.18405902022014@news.panix.com> <CAPTjJmo6qpfP74YF+2xKV1y-qUNt6rG+8kHkoYJZXbb3HED5pg@mail.gmail.com> <CAAu18hevvTDWizx8Gt2S4MUv04KQ0uw7sZaOk-pi-ffsOtG8nw@mail.gmail.com>
Date: Tue, 4 Feb 2014 04:57:13 +1100
Subject: Re: __init__ is the initialiser
From: Chris Angelico <rosuav@gmail.com>
Cc: "python-list@python.org" <python-list@python.org>
Content-Type: text/plain; charset=UTF-8
Precedence: list
Newsgroups: comp.lang.python
Message-ID: <mailman.6349.1391450237.18130.python-list@python.org>
Lines: 49
NNTP-Posting-Host: 2001:888:2000:d::a6
Xref: csiph.com comp.lang.python:65352

On Tue, Feb 4, 2014 at 12:50 AM, Nicholas Cole <nicholas.cole@gmail.com> wrote:
>> There have been occasional times I've wanted an "explicit destruction"
>> feature. Rather than the facetious exception I listed above, it'd be
>> better to have all those references (including the original one in a,
>> since there's nothing special about that) turn into some kind of "null
>> state" - either None, or a special object that marks itself as a
>> destructed/destroyed (terminology debates aside) object. With custom
>> types, I can mark them off with a special flag, and check that all the
>> time; but I can't, for instance, have a dict that maps some lookup
>> keyword to its output file, and then destroy output files to remove
>> all their references from everywhere in the dict. (I have had
>> something along these lines, a bit more complicated than this, but not
>> in Python.)
>
> Can't you get close to that using weakrefs?  I'll admit that care is required.

Weakrefs are a related tool, but solving a different problem. What I
wanted here was an easy way to force all references to a particular
file to be wiped out, based on one of the existing references. Here's
a concocted setup that's broadly similar to what I was doing, which
might illustrate the issue:

log_files = {}

def add_log_file(fn, *keywords):
    f = open(fn, "w")
    for kw in keywords: log_files[kw]=f

for line in process_me_line_generator():
    kw = line.split()[0]
    f = log_files.get(kw)
    if not f: continue
    f.write(line+"\n")
    if 'quit' in line:
        # Okay, let's now close this file.
        destruct(f)


In this particular case, I could use "f.close()" and "if not f or
f.closed: continue", but that requires that the object cooperate in
this way, and I'm not entirely sure about resource usage. (I was
actually working with a database connection object, IIRC, which didn't
offer me a way to inquire if it was still open or not.) To do this
with weak refs, I'd have to have some other source of strong refs, and
closing would be done by digging through that list, disposing of it
from there, and then triggering garbage collection. I suppose it could
be made to work, but it feels like going about everything backwards.

ChrisA