On 4/23/2013 12:57 PM, Blind Anagram wrote:
So, all I was doing in asking for advice was to check whether there is an easy way of avoiding the slice copy,
And there is.
not because this is critical, but rather because it is a pity to limit the performance because Python forces a (strictly unnecessary) copy in order to perform a count within a part of a list.
Python does not force that. You have been given several simple no-copy alternatives. They happen to be slower *with CPython* because of the speed difference between Python code and C code. If the same timing tests were done with any of the implementations that execute python code faster, the results would likely be different.
I thing str/byte/bytearray.count have more need for optional start,stop boundary parameters because a) people search in long texts and subtexts, more so I think that for other sequences, b) they search for substrings longer than 1 and hence c) the generic no-slice alternatives do not work for counting substrings.
That said, I do see that tuple/list.index have had start, stop paramaters added, so doing the same for .count is conceivable. I just do not remember anyone else asking for such. The use case must be very rare. And as I said in my other post, .count(x) applies to any collections, but start,stop would only apply to sequences.
In other words, the lack of a list.count(value, limit) function makes Python less effective than it would otherwise be.
Untrue. The alternatives are just as *effective*. -- http://mail.python.org/mailman/listinfo/python-list
