Hacker News Reader

Comment by mst | Hacker News Reader

The "essentially no dependencies" thing always fascinates me.

Certainly, complex C code is a risk across OS versions but ... perl5 version 8.1 was released just over 18 years ago and after multiple years of annual major releases of perl we're now on version 34. Many of my chosen cpan dependencies produce the exact same output when on a 5.8.1 install as they do on a 5.34.0 install, and I can move them to a new machine/OS via rsync or tar of the local::lib structure (virtualenv like thing configured by setting a couple of environment variables) and it Just Works.

Am I weird here? Is the author? Is python unusually bad? Is perl unusually good? This all seems quite strange to me.

View on HN

perl is unusually good in this regard, as I'm sure you already knew.

I do suspect that, but my fingerprints are on enough of cpan these days that I would worry about bias if I simply asserted it.

I guess it’s the language maintainers who decide if they want to keep backwards compatibility. And you can’t even say that’s due to language age, necessarily. PHP 8 (Nov 2020) had many breaking changes over past PHP versions and that language is from the mid 1990s. On the other hand, Java also dates to the mid 1990s and code written for Java2 still compiles and executes in Java 17 (2021).

Java is not my cup of anything but I do absolutely respect that about it.

I have a long standing semi-shitpost opinion that if faced with a choice of technologies, it's often best to pick the one that's been "declared dead/dying" because that means it's stable and in wide use.

Maybe some code does, but I've run into JARs that I've had to grab an older version of the JRE to run.

I am with you here. That being said, I believe the author refers to « weaker » libraries (higher-level, more recent etc. )

I have noticed the same thing regarding Python and Perl. It's sometimes difficult to get the same program to even run on adjacent minor versions of Python. There are multiple xkcds about sorting out your Python environment: https://xkcd.com/1987/ https://xkcd.com/2510/ (see title text)

Meanwhile, I have Perl programs deployed that have been running with no changes for over 15 years.

Reply on news.ycombinator.com