more packages != better
Agreed.
But not all packages are equal, and not all packages have the same ‘importance’.
A library that is used by > 1000 programs …
You could probably calculate the ‘importance’ of package, from the number of things that depending on it, the number of users using it etc etc etc. There must be a set of metric that could define this.
This could be used as a ‘bonus’ multiplier, for bounties (time. nix coins, dollars, kudos, reputation)
Ibm used to Klocks as a metric for paying programmers, for every 1000 lines of code they wrote they got paid more.
Seems like a good idea , but it made programmers write large sprawling programs, rather the smaller optimised programs… i classic case of the management not understanding what software was, how to craft it.
Maybe software organisations (which are sometime an individual or < 2 actual people) , should pay programmers not to developer code, not add features, but pay for refactoring, pay for documentation, pay bonuses for tests. Rather than a feature based economy, which has probably got us into this ‘software complexly’ crisis. Nix doesn’t address that, it just manages that complexity.
How could one measure the complexity of a derivation (package / module).
Many nix should come with a way of submitting (voluntary) stats on what your machine has installed. I can see whay Macrohard do this ‘data collection’, so they know what to spend their time 'make that ‘thing’ work.
Like a CVE score but for ‘importance’.
The more i think about it, the more is, How does open source work so well at all?
The kernel has it easier, its a much smaller project compared to nix/OS .
I’m going to wander off for a while and mull this over… food for thought. Thanks you for all your comments.