[Patches] [ python-Patches-643443 ] dict.fromkeys()
noreply@sourceforge.net
noreply@sourceforge.net
Tue, 26 Nov 2002 23:36:15 -0800
Patches item #643443, was opened at 2002-11-25 05:29
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=305470&aid=643443&group_id=5470
Category: Core (C code)
Group: Python 2.3
>Status: Closed
Resolution: None
Priority: 5
Submitted By: Raymond Hettinger (rhettinger)
Assigned to: Raymond Hettinger (rhettinger)
>Summary: dict.fromkeys()
Initial Comment:
Implements dict.sequpdate() as discussed on python-
dev:
def sequpdate(self, iterable, value=True):
for k in iterable:
self[k] = value
return self
----------------------------------------------------------------------
>Comment By: Raymond Hettinger (rhettinger)
Date: 2002-11-27 02:36
Message:
Logged In: YES
user_id=80475
After more code review, added PyDict_Check.
Also, discussion on python-dev showed a need to rename
the method to indicate that keys were being used rather
than items. The sequence part of the name was dropped
because any iterable will do as an argument.
Checked-in as:
Misc/NEWS 1.548
Objects/dictobject.c 2.132
Lib/test/test_types.py 1.41
Doc/lib/libstdtypes.tex 1.111
----------------------------------------------------------------------
Comment By: Guido van Rossum (gvanrossum)
Date: 2002-11-26 11:40
Message:
Logged In: YES
user_id=6380
I'm ok with this, but I haven't had time to review the code.
----------------------------------------------------------------------
Comment By: Just van Rossum (jvr)
Date: 2002-11-26 05:25
Message:
Logged In: YES
user_id=92689
Btw. I was looking for more info on METH_CLASS and came
across this thread on python-dev:
http://mail.python.org/pipermail/python-dev/2002-April/023566.html
which resulted in this bug report:
http://www.python.org/sf/548651
which has not been resolve. I'll add a comment there, too.
----------------------------------------------------------------------
Comment By: Raymond Hettinger (rhettinger)
Date: 2002-11-26 05:04
Message:
Logged In: YES
user_id=80475
Okay, fixed decrefs, eliminated i; used for(;;).
Left in PyObject_CallObject because it's clean.
Thanks.
----------------------------------------------------------------------
Comment By: Just van Rossum (jvr)
Date: 2002-11-26 04:42
Message:
Logged In: YES
user_id=92689
dang, that should've read "you should NOT incref d at the end".
----------------------------------------------------------------------
Comment By: Just van Rossum (jvr)
Date: 2002-11-26 04:40
Message:
Logged In: YES
user_id=92689
- I'm not sure whether PyObject_CallObject() is the right
idiom here. I would have expected to call cls->tp_new (if
not NULL), but I'm not sure either way.
- I would get rid of the loop counter: you're not using it.
for (;;) will do nicely
- you're XDECREF'ing values of which you know they're not NULL
- idem XINCREF. Also: I'm pretty sure you should incref d at
the end: I think it currently leaks.
----------------------------------------------------------------------
Comment By: Raymond Hettinger (rhettinger)
Date: 2002-11-26 04:20
Message:
Logged In: YES
user_id=80475
Fixed news item.
----------------------------------------------------------------------
Comment By: Thomas Heller (theller)
Date: 2002-11-26 04:13
Message:
Logged In: YES
user_id=11105
It seems the code and the docs are out of sync in
fromseq2.diff.
The docs say "fromseq(sequence, value=True)" while the
code does "fromseq(sequence, value=None)"
----------------------------------------------------------------------
Comment By: Raymond Hettinger (rhettinger)
Date: 2002-11-26 03:53
Message:
Logged In: YES
user_id=80475
Done! Revise patch attached.
It is now a class method and uses cls for the type of the
new instance. Added tests to demonstrate meaningful
inheritance.
----------------------------------------------------------------------
Comment By: Guido van Rossum (gvanrossum)
Date: 2002-11-25 20:15
Message:
Logged In: YES
user_id=6380
Hm, but the convention for "alternate constructors" is that
the cls argument *is* used to determine the type of the new
instance. This way, the constructors are inherited in a
meaningful way by subclasses of dict.
----------------------------------------------------------------------
Comment By: Raymond Hettinger (rhettinger)
Date: 2002-11-25 19:38
Message:
Logged In: YES
user_id=80475
Attached revised patch incorporating all of the review
comments. Changed default to None. Made a static
method instead of a classmethod because cls was not
used. Doesn't return self.
Tests, docs, and news item revised accordingly.
Ready for BDFL pronouncement.
----------------------------------------------------------------------
Comment By: Just van Rossum (jvr)
Date: 2002-11-25 09:26
Message:
Logged In: YES
user_id=92689
One more thing: I'd say the default value should be None.
This makes the idiom also useful to initialize dicts which
later get filled with proper values. This doesn't take
anything away from your original use case.
----------------------------------------------------------------------
Comment By: Just van Rossum (jvr)
Date: 2002-11-25 09:09
Message:
Logged In: YES
user_id=92689
I see no reason why d.sequpdate() should return self.
d.update() doesn't either.
Also, as I suggested on python-dev, a fromseq() constructor
(as a class method) would be nice:
>>> dict.fromseq('End Quit Stop Abort'.split())
{'End': True, 'Quit': True, 'Stop': True, 'Abort': True}
>>>
----------------------------------------------------------------------
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=305470&aid=643443&group_id=5470