RU version is available. Content is displayed in original English for accuracy.
Advertisement
Advertisement
⚡ Community Insights
Discussion Sentiment
67% Positive
Analyzed from 561 words in the discussion.
Trending Topics
#code#review#local#generated#tools#before#making#projects#commit#gonna

Discussion (18 Comments)Read Original on HackerNews
it doesnt help that these projects have AI generated README's. Not AI-assisted, or AI-curated, but fully generated. it's a bunch of sentences that say nothing. "the diff is whats reviewed". okay, what about per-commit reviews? does it keep context of that? just diffs isnt enough to review a PR, it just sounds efficient but youre leaving a lot of context out, so the reviewer will have trouble confidently reviewing your changes.
Not for criticism, but is this targeting that “air gapped” git diff scanning area?
...But that seems like a very odd thing to be concerned about. Plus, unless I'm missing something, the program will crash with an an ImportError before that block can ever run.
AIs tend to have a debilitating phobia of any kind of exception. Claude especially will throw try/except and weird coersion everywhere in the code.
This bites you because it hides legitimate errors, even when the program is in such an unrecoverable state that it doesn't make logical sense to continue. It becomes impossible to debug, and requires the test suite to be absolutely perfect because you end up throwing away the only way for the language to warn you that something went wrong.
So while yes, it's relatively minor, like many code smells it is indicative that other things may be wrong.