Time between “I need to address this problem” and “solution is designed and implemented”:
- g-expressions: 4y
- system services: 2y
- channels: 3y
- Git checkout authentication: 4y
- records & packages: 0.5y, with several iterations over 14y
Time between “I need to address this problem” and “solution is designed and implemented”:
- g-expressions: 4y
- system services: 2y
- channels: 3y
- Git checkout authentication: 4y
- records & packages: 0.5y, with several iterations over 14y
I guess I’m slow (and lucky to have had conditions that allowed me so spend so much time on this!), but really, during that time nothing would happen until the design would click in my mind.
@civodul Thank you for all these great contributions that we all benefit from so much, Ludo! 🙏🙏🙏
@fnat Heh, thanks!
To be clear, my point was not to show off but to emphasize that fine design issues can take time before they trickle down into something workable.
I guess the same holds for any non-trivial software system that has lasted a bit (maybe with the exception of Unix :-)), and that’s probably a blind spot in current discourse on software engineering and productivity.
Current slopogoloy I see every day at work - never look forward, rewrite all from scratch, don't use shared code, spend time to reimplementing common practices on and on and on
But that is work, and today is Friday
During hobby time I enjoy gardening Guix packages :heart_fire:
@civodul Hey, don't worry, I had no doubt this wasn't meant as showing off. 🙏
Re the discourse around software engineering, there's always been a myth of 10x productivity, even before AI. Now we see extreme manifestations of it - and I can see how problematic this may be.
On the other hand, as an AI user myself, I do see some truth in some of the pro-AI claims. Not all GenAI-produced or assisted projects are "slop".