On Fri, Nov 30 2018, Vladimir Nikishkin wrote: > I have made a first attempt to add docstrings to functions and sent it > as a merge request. Maybe some of my comments can be picked. > Unfortunately, only commenting the public interface turned out to be > harder than I expected, (although it should have been more evident), > because the public interface is dependent on the internals, so I had > to first understand those.
excellent, i'll review the merge request as soon as possible (hopefully this weekend). > I have one more meaningful question. In the geiser-base.el, there is a > function geiser--del-dups, which is currently working in O(n^2), since > is toes a dolist (O(n)) times member (O(n)). > Is there an easy way to, maybe, do it in O(nlogn)? Like a hashset, or > something like this. there are ways, yes, but maybe it's not it: using a hash is going to be faster only when n is bigger than some threshold (otherwise, constant factors associated to create the hash just override the lookup time gains). we only call that function in ./geiser-completion.el57, so i would first test when that becomes a bottleneck... jao
