Automatic arrays and intrinsic array operations: to use or not to use?

[zaikunzhang]

@RonShepard

I’d guess that other languages don’t need to worry about this the same way we do in fortran because they are already so inherently inefficient, it doesn’t matter if they use heap for everything, or if they attempt to use stack, fail, and fall back to heap allocation. All of that happens with less effort than parsing the next line of user code, so it is all hidden by the nature of interpreted languages. Instead these languages rely on library routines, often written in fortran ironically, to do the heavy lifting.

Thank you very much for the insights.

To everybody, I would like to ask the following question: Which one do you prefer, (capable of) being efficient at the expense of occasional crashes, OR being robust at the expense of occasional inefficiency?

Recall that one of the slogans of Julia is “Fast as Fortran, Beautiful as Python”, or even “Faster than Fortran, cleaner than Numpy”. There are two possibilities.

Possibility 1. The Julia people are lying. Efficiency and elegance simply cannot coexist. They eat each other.

Possibility 2. It is possible to achieve efficiency (with the help of Fortran) while keeping elegant, maybe not as much as the Julia people claim. If this is the case, why could other languages achieve such a nice balance with the help of Fortran, but Fortran itself could not? Is there a paradox here?

Disclaimer: I am not a Julia user. My knowledge about this language comes completely from the current forum. I use only Fortran and MATLAB, both heavily.