Post #1544884
2026-03-07 11:19 UTC
Replies (1)
-
@ginabythebay@hachyderm.io 2026-03-07 11:28
@vaurora I’m someone who enjoys crafting nice code and occasionally it lasts but really not that often. That said, the last week or so I finally gave the jetbrains llm a try since it was bundled into my subscription. I’ve been paying close attention to what it does, treating it like a first draft. Generally I edit whatever it did, sometimes a lot. But also it has shown me some corners of python I didn’t know about, eg NamedTuple, so that is nice. Its test generation seems less good. It kind of goes crazy trying to test every last possible thing and I don’t want to be stuck maintaining massive piles of test code if they aren’t useful. Also I keep finding tests that it writes that don’t test anything. Like they do all the test setup, execute the code that needs testing and then, nothing, no check that the code did what it was supposed to. I keep hearing that Claud is better, including in the essay you linked