3 ms·

> Today, I type "npm install json2csv" and I'm done. You have to do a little bit more: You have to search for it and decide which of the n converters you will

> Today, I type "npm install json2csv" and I'm done. You have to do a little bit more: You have to search for it and decide which of the n converters you will use (licence, other dependecies, ..) But I get your point.

routerl 4y ago
You just made something click for me. My parent was an electrical engineer. While their skillset was varied, and heavily weighted towards physics and math, their day job was mostly about knowing how to pick parts out of a catalogue, which catalogue to use when, and maintaining good relations with the suppliers behind those catalogues. I always saw this as a weird disjunction. Until now. I've just realized that while my core skillset is in math, logic, computer science, and system design, I do spend a lot more of my time researching repositories, packages, licenses, etc, than actually coding. But with the new perspective your comment gave me, I finally saw that as totally in line with other, older engineering disciplines. So, to address the original point of this thread, it seems as if programming has finally evolved into a semi-mature engineering discipline, wherein our ability to gauge the quality of existing supplies is more important than our ability to produce new supplies.
- rileyphone 4y ago
  Brad Cox had a concept of Software ICs, in the chip sense, that align with this. He envisioned these as sharable objects rather than just libraries, but either way that’s the present we’ve been given accidentally.
  - spc476 4y ago
    Motorola developed the MC6839 which allowed one to do floating point math with the MC6809. The MC6839 was not a floating point CPU, but an actual 8K ROM of 6809 machine code that implemented floating point math (with position independent code so it could be mapped anywhere in the memory map). This is about the closest I have ever seen of a "software IC" in my life.
  - anon25783 4y ago
    I mean. we do call them "shared object files"

routerl 4y ago

You just made something click for me. My parent was an electrical engineer. While their skillset was varied, and heavily weighted towards physics and math, their day job was mostly about knowing how to pick parts out of a catalogue, which catalogue to use when, and maintaining good relations with the suppliers behind those catalogues. I always saw this as a weird disjunction. Until now. I've just realized that while my core skillset is in math, logic, computer science, and system design, I do spend a lot more of my time researching repositories, packages, licenses, etc, than actually coding. But with the new perspective your comment gave me, I finally saw that as totally in line with other, older engineering disciplines. So, to address the original point of this thread, it seems as if programming has finally evolved into a semi-mature engineering discipline, wherein our ability to gauge the quality of existing supplies is more important than our ability to produce new supplies.

rileyphone 4y ago
Brad Cox had a concept of Software ICs, in the chip sense, that align with this. He envisioned these as sharable objects rather than just libraries, but either way that’s the present we’ve been given accidentally.
- spc476 4y ago
  Motorola developed the MC6839 which allowed one to do floating point math with the MC6809. The MC6839 was not a floating point CPU, but an actual 8K ROM of 6809 machine code that implemented floating point math (with position independent code so it could be mapped anywhere in the memory map). This is about the closest I have ever seen of a "software IC" in my life.
- anon25783 4y ago
  I mean. we do call them "shared object files"