"Myles Strous" <[EMAIL PROTECTED]> escribi� en el mensaje news:[EMAIL PROTECTED]
satisfy some handy properties, the first of which being: l[:n] + l[n:] = l
I don't think l[:5] + l[5:] = l is a handy property and to me is clearly counterintuitive. Further,
It can be quite useful for inserting something into a list (or string), after finding the position where you wish to insert it. improvedList = l[:n] + [new stuff] + l[n:]
As I answered in another post this is not more useful than writing l[:n-1]. Of course, I'm aware of the case where n=0, but this would require only a bit of extra code (and, after all, I'm just saying that half-open ranges are not a panacea and that I don't like their side effects).
I vaguely remember hearing at one stage that the sequence[position:position+length] notation is also potentially useful for indexing into large strings or buffers.
Right, to some extent it's useful, but at the cost of introducing tricky syntaxes for very specific cases, like this one, and unexpected off-by-one errors in other cases, which is my point. For example, recently I had to get a value from a list preceded by the two previous values: lst[n-2:n+1] and not the more logical (at last to my eyes) lst[n-2:n].
Instead of giving further examples I would like to cite three cases:
1) You have a starting point (s) and a length (t): lst[s:s+t]. 2) You have an ending point (e) and a length: lst[e-t+1:e+1]. 3) You have a starting point and an ending point: lst[s:e+1].
What's odd is that Python applies the syntax of case 3 to the logic of case 1. While something like lst[s:s+t-1] for the first case could be explained in simple mathematical terms (in other words, it's an integral part of the algorithms), I cannot find a way to explain the e+1 in cases 2 and 3 (and the inconsistency with e-t+1 in case 2 vs. s+t in case 1) except the Python syntax.
While this may be an interesting philosophical (or should that be philological) discussion, since Python has worked this way for donkey's years, and since a change would break 30% of the existing codebase, you clearly can't be advocating change.
So, what's the point of this thread now?
regards Steve -- Steve Holden +1 703 861 4237 +1 800 494 3119 Holden Web LLC http://www.holdenweb.com/ Python Web Programming http://pydish.holdenweb.com/
-- http://mail.python.org/mailman/listinfo/python-list
