Andy Dienes
Andy Dienes
I'm assuming this is stale? (and from the discussion above, it seems this was unlikely to be a Julia bug in the first place)
I just hit something similar: would it be at all feasible for `BitIntegers.@define_integers 2048` to make `widen(UInt1024) == UInt2048` ?
does the "help wanted" here refer to performance optimizations? or are there other, deeper, darker problems. that is --- would it be accurate to say that this PR implements all...
I have some proposed updates in this branch here https://github.com/adienes/julia/pull/1 what would be the recommend way to merge (or at least compare) to this pr? should I draft a PR...
besides the needed bugfix (in comment), this PR looks like a good idea to me and I'd be happy to continue to review it
just spitballing: the fact that both `dropdims` and `insertdims` forward to `reshape` suggest that maybe `reshape(::PermutedDimsArray)` should be specialized rather than each of `dropdims` and `insertdims`. also `reshape` is not...
downvote. this adds more instances where `convert(T, x)` returns something not of type `T`, which I think should be avoided unless for some reason absolutely necessary.
sorry, my mistake. I forgot that `AbstractRange
personally speaking I don't think I'm a fan of this. the ("wat" potential) : (cuteness of syntax gained) ratio is too high
essentially a duplicate of https://github.com/JuliaLang/julia/issues/18312, although they examine different consequences so maybe both should be left open