[enriquto]

> Legacy systems have legacy dependencies, and run in legacy environments.

In my experience, a sort of opposite of this sentence tends to be true.

Modern systems have too many dependencies, most of the time needlessly so, but still required. I recently used a Python package that required a specific version of the pandas behemoth to run, but it was only used to parse a date. Absolutely ridiculous.

On the other hand, the 40-year old Fortran code that everybody uses to compute special functions compiles and runs OK in a modern computer, and is completely standalone with 0 dependencies apart from the compiler.

[Dave3of5]

My experience is the opposite. You python code all it needed was a Requirements.txt and a pip install and the tooling takes care of the rest, been like that for years.

Any non trivial 40-year Fortran code though works with a specific compiler good luck getting that to run on a modern computer.

Also to note is that date parsing is complex. I can see why most would prefer to use a properly tested lib to do it. I've seen far too many handcrafted date parsers that suck and break in crazy ways over the years.

[swiftcoder]

I don't think I've ever used the phrase "legacy system" to something from the 90's, even. Mostly it's software in that twilight zone of 5-10 years old. Built with "modern" expectations and dependencies, but every system around it has migrated to a new and shiny tech stack, and all the subject matter experts are long gone...

[enriquto]

> every system around it has migrated to a new and shiny tech stack, and all the subject matter experts are long gone

This seems to be an unfortunate property of software related to web development and similar stuff. Fortunately, scientific computation moves at a more reasonable pace. There are new algorithms every day, but no new "frameworks"!