Message221288
All contributions are subject to final commit review. I looked at the patch and it is a *lot* of code for little benefit. I think the better solution would be an informative error message: "Currently, islice arguments must be less than {} on {}-bit systems".format(n, k).
Since I posted nearly 4 years ago, I have become more aware of the important differences between 3.x range as a sequence class whose instances are non-iterator *(re)iterables* and count as an iterator class whose instances are one-time *iterators*. To me, arbitrarily large indices now seem more appropriate for virtual sequences that can do O[1] indexing rather than iterators where indexing is O[n].
A recent proposal on python-ideas, which as I remember amounted to enhancing count to be more like range, tripped over this difference. I suggested that a new infinite sequence class Count (like range but without necessarily having a stop value) was a more sensible idea for what the OP wanted to do. |
|
Date |
User |
Action |
Args |
2014-06-22 18:04:06 | terry.reedy | set | recipients:
+ terry.reedy, loewis, rhettinger, thomasguest, AlokSinghal |
2014-06-22 18:04:06 | terry.reedy | set | messageid: <1403460246.48.0.310124305282.issue6305@psf.upfronthosting.co.za> |
2014-06-22 18:04:06 | terry.reedy | link | issue6305 messages |
2014-06-22 18:04:06 | terry.reedy | create | |
|