Book: Perl 7: A Risk-Benefit Analysis

Very unfortunate. I cut my teeth with Perl 5, and built some fantastic stuff with it. For all its faults, I could be writing messy one-off scripts in the morning, and cranking styleguide-strict production grade code in the afternoon. I used to maintain some CPAN modules.

With Perl 6 stalled, people moved on to Python, or Ruby (which looked very close to the Perl 6 vision, with simpler sigils), or more exotic stuff - Elixir, Go... and bash matured a bit (but not enough).

COBOL is, itself, the next COBOL, to be joined by Perl and Java as peers. If this evokes visions of the Lava Flow anti-pattern writ large, we are in agreement. COBOL's next generation will need to be facile in all three.

Certainly, most new projects that would once have been done in COBOL are not, anymore. It looks like Java inherited that mantle. New projects are still started in Java, hard as Oracle tried to kill it. Perl, Python, and for a while Ruby picked off the top, but a great deal of such work migrated to web tech and Javascript.

We can't identify Javascript as a COBOL inheritor yet because nothing appears poised to supplant it for new development, howsoever lucky we would all be if at least Typescript were.

Meanwhile, for systems programming, nothing will touch C++ for at least twenty years(*), present Rust enthusiasms notwithstanding. Rust has passed the point where it might have Jumped the Shark, but it can still become Erlang just by not growing its user base quite fast enough. It is not yet growing at anywhere near a sufficient rate, but still might, and must.

Rust enthusiasts should seek new users from among the best Java and Go coders, frustrated at those languages' limitations, even though most on those could never make the jump. C coders would be prime candidates (and we would all be better off if they switched) but any who might have jumped ship did long ago.

(*) Provided civilization even lasts so long. Not looking good.

Modern Android versions also precompile DEX classes into native code.

For example, in October of last year, the United States Air Force announced that it had just figured out how to launch a nuclear missile without the use of 8-inch floppy disks.

To give an example more obviously* related to accountants, the financial system of the United States is still critically dependent on a variety of hilariously outdated systems. They have the unusual property of being so old that they are paradoxically secure against the internet's "background radiation" of opportunistic non-targeted scans and attacks (because they use EBCDIC instead of ASCII). Or at least, they would be, if the banks were foolish enough to let them anywhere near a live internet connection in the first place. Instead, my understanding is that they are entirely confined to a series of air-gapped private networks, with limited internetworking for ACH processing etc.

* The (discretionary) US military budget for fiscal year 2019 is approximately $686 billion. For context, that is substantially greater than the entire GDP of Belgium (as of 2018).