Tom Clune

In the mean time I have a project https://github.com/Goddard-Fortran-Ecosystem/gFTL which provides (by far less elegant means) a generic container system. Currently it supports Vector and Map (ala C++ STL), but...

@misael-diaz Encapsulating a C implementation is a reasonable approach for lists of (most) intrinsic types. However, it does not help much with linked-lists whose elements are non C-interoperable derived types....

The set (tree) implementation still has to do pretty much the same thing, just not in the same order. It has to traverse down to each leaf node for the...

It would be fairly straightforward to add List (STL name for doubly-linked list) to my package. Just a bit tedious though, so not something I'm likely to get to in...

I'm sorry I misread your email above. You should be able to delete a doubly-linked list in O(N) steps. Your description suggests that your implementation is O(N^2). You can start...

2.4 GHz 8-Core Intel Core i9 (Recent vintage MacBook Pro)

@misael-diaz You may also have inadvertently helped to identify a bug in my set implementation. I tried the analogous O(N^2) clean up process and it crashes my code for the...

May not be a bug in the container itself after all. Same code works fine with NAG compiler. If it's a compiler bug, it might be real fun to fix...

Not sure the array case is very relevant here - it has different properties. Even my Set example was only relevant in that it has a similar complexity for deletion....

I suspect no Fortran compilers do this. For multidimensional cases, the risk of oversubscribing memory is too great, and even in the 1D case, the compiler should not risk overallocation...