Working now at a bigco, the most impactful thing I do these days is in guiding projects away from building the first thing they think will solve a problem. People don't spend much time doing thought experiments of how changes will evolve in the future or with adjacent scopes. After thinking in that mode for a while you realize that there are concepts here that could and should be separated. A small tweak here and there, changing some naming/terminology goes a long way to saving tons of refactoring/cleanup down the road.
If you mean single-handedly, kinda hard to say. I also rewrote chunks of a retail FX app written in Java1/awt -> Java5+/Swing. Right now I'm enjoying using my own HN viewer (hackerer.news). I'd like to make an SQL-oriented library so people don't have to settle for JPQL/Hibernate--started but not done/promoted.
A recent stroke of luck was working on a small team building buy-online-pickup-instore for thousands/millions of merchants, that completed just before the pandemic hit.
Sure, SafeQL[0] SQL library. It's been sitting in a close but not enough to promote state for a while. The main thing I wanted was to have generation of the bindings to existing DB schema. I also want to combine using Moja[1] datatypes for uniform handling of single/multi, normal/async, and errors.
I suspect I'm mostly driven by the unknown and as I get close to practically solving a problem start to lose the curious itch.
If you mean single-handedly, kinda hard to say. I also rewrote chunks of a retail FX app written in Java1/awt -> Java5+/Swing. Right now I'm enjoying using my own HN viewer (hackerer.news). I'd like to make an SQL-oriented library so people don't have to settle for JPQL/Hibernate--started but not done/promoted.
A recent stroke of luck was working on a small team building buy-online-pickup-instore for thousands/millions of merchants, that completed just before the pandemic hit.