I've found it very useful, even if it is O(n) complexity to get the n-th element, to allow indexing.
For a database, if you have a packed record (for example, like SQLite's row format), and you only need to access a single field, it's better to be able to do just that, and *not* have to unpack the entire record just to do that (skipping fields can be many many times faster than unpacking them, plus the amount of time wasted on allocating objects for all of the fields of records scanned and then GCing them). I think it would be better to have a trait that you could find out if an indexable structure is directly indexable or not, kind of like strings used to have) (or does that exist now, with all of the changes for Arraymageddon?) On Thursday, June 23, 2016 at 3:35:06 PM UTC-4, Mauro wrote: > > My recollection is that part of the indexing interface in Julia (just by > convention) is that indexing should be of O(1) (or close to that) > complexity. Iterable things, in general, have O(n) complexity to access > the n-th element, because you have to traverse the data to get there > (the classic example are linked lists). Thus, it makes sense that > indexing is not supported, instead you have to call collect first if you > want to index. > > On Thu, 2016-06-23 at 20:43, Davide Lasagna <[email protected] > <javascript:>> wrote: > > Hi Jacob, > > > > In my view, in principle, all "iterators" should be indexable, (at least > > read-only), *unless *the underlying data is not indexable by nature, > e.g. > > with data that comes from a stream... Doing `zip([1, 2], [2, 3])[1]` > > should probably just work. > > > > I also think that for `Zip`s if the zipped collections are not > indexable, > > then the "outer" getindex method should fail internally with a > MethodError, > > on the "inner" getindex calls not implemented for the non-indexable > > collections. > > > > On Thursday, June 23, 2016 at 7:23:34 PM UTC+1, Jacob Quinn wrote: > >> > >> Sorry, to clarify a little: > >> > >> The things you're zipping are not necessarily indexable (i.e. other > >> iterators), so it's not safe to assume you can always index a Zip. > >> > >> On Thu, Jun 23, 2016 at 2:21 PM, Jacob Quinn <[email protected] > >> <javascript:>> wrote: > >> > >>> Most "iterator" types are not indexable, AFAIK. The typical > >>> recommendation/idiom is to just call `collect(itr)` if you need to > >>> specifically index. > >>> > >>> -Jacob > >>> > >>> On Thu, Jun 23, 2016 at 2:18 PM, Davide Lasagna <[email protected] > >>> <javascript:>> wrote: > >>> > >>>> Is there any particular reason why `Zip` objects are iterable but not > >>>> indexable? Python allows that. > >>>> > >>>> From previous discussion on the topic (2014 topic at > >>>> https://groups.google.com/forum/#!topic/julia-dev/5bgMvzJveWA) it > seems > >>>> that it has not been implemented yet. > >>>> > >>>> Thanks, > >>>> > >>>> > >>>> > >>> > >> >
