nuky

7 ms·

1.
by nuky 8mo ago
Oh yeah tree-sitter it's a great foundation for semantic structure. What I'm exploring is more about what we do with that structure once someone/smth starts generating thousands of changed lines: how to compress change into s
2.
by nuky 8mo ago
Agreed - that trap is very real. The open question for me is what we do when atomic, 5min readable diffs are the right goal but not realistically achievable always. My gut says we need better deterministic signals to reduce noise before hum
3.
by nuky 8mo ago
This is exactly the gap I'm worried about. human review still matters, but linear reading breaks down once the diff is mostly machine-generated noise. Summarizing what actually changed before reading feels like the only way to keep rev
4.
by nuky 8mo ago
This came out of reviewing a lot of large refactors and AI-assisted changes. The goal isn't to replace code review (ofc) or relax standards, but to make the consequences of changes visible early — so reviewers can decide where to focus
5.
A tool to see what changed in your code (beyond the diff)(gitlab.com)
1 points by nuky 8mo ago|1 comments
6.
by nuky 8mo ago
It was precisely because this was going too far that I thought the consequences of the active adoption of LLM tools could be made visible. I'm not saying LLM is completely bad—after all, and not all tools, even non-LLM ones, are 100% d
7.
by nuky 8mo ago
fair — that’s what I do as well)
8.
by nuky 8mo ago
Yeah difftastic and similar tools help a lot with formatting noise really. My question is slightly orthogonal though: even with a cleaner diff, I still find it hard to quickly tell whether public API or behavior changed, or whether logic ju
9.
by nuky 8mo ago
That matches my experience too - tests and plans are still the backbone. What I keep running into is the step before reading tests or code: when a change is large or mechanical, I’m mostly trying to answer "did behavior or API actually
10.
by nuky 8mo ago
Makes sense. What about latency? for typical and hard queries
11.
by nuky 8mo ago
Nice. And how do you handle candidates near H3 boundaries?
12.
by nuky 8mo ago
Sounds interesting. What's the stack under the hood?
13.
by nuky 8mo ago
Just to clarify - this isn’t about replacing diffs or selling a tool I ran into this problem while reviewing AI-gen refactors and started thinking about whether we’re still reviewing the right things. Mostly curious how others approach this
14.
Ask HN: Are diffs still useful for AI-assisted code changes?
7 points by nuky 8mo ago|17 comments