Yeah, looks like very similar arguments to 9147. Although not as easy to circumvent as using zeros() or the like.
One potential "fix" could be a potential third argument to resize with the "default / fill" value, which doesn't seem like a viable solution to the constructors discussed in that thread. Cameron p.s. Thanks, Josh, but I think that solution would create a temporary array (from zeros() call) on every resize. Perhaps successive push!(vec, 0.0) might also be a good way (I know push allocates in chunks, resize might not -- I haven't looked). On Tue, Oct 27, 2015 at 10:36 AM, Stefan Karpinski <ste...@karpinski.org> wrote: > I've encountered this too – it might make sense to zero out the grown > memory. Of course there's the whole "to zero out uninitialized memory or > not" discussion: https://github.com/JuliaLang/julia/issues/9147. I feel > like this falls into that same category of questions. > > On Mon, Oct 26, 2015 at 3:59 PM, Josh Langsfeld <jdla...@gmail.com> wrote: > >> You could do 'append!(vec, zeros(i-n))'. >> >> >> On Monday, October 26, 2015 at 3:32:56 PM UTC-4, Cameron McBride wrote: >>> >>> Hi All, >>> >>> What's the best julian way to do the following: >>> >>> function vecadd!(vec, i, v) >>> n = length(vec) >>> if n < i >>> resize!(vec, i) >>> vec[n+1:i] = 0.0 >>> end >>> vec[i] += v >>> end >>> >>> This seems somewhat typical for a growing array that's not just >>> incremental (i.e. not just a push!() operation), so I feel like I missed >>> something. Better suggestions? >>> >>> Thanks! >>> >>> Cameron >>> >> >
