Content
@
0 reply
0 recast
2 reactions
Ryan J. Shaw
@rjs
The thing I'm enjoying most about Cursor isn't that it's perfect (it's not - it's nowhere near 100% accurate), but rather that it seems to confirm the approach I'm taking to something when I'm working in an unfamiliar domain (dagster + python)... I upgraded to pro after my trial ended this morning. I wonder if I will keep my $32/mo JetBrains sub, or my $20/mo ChatGPT subs...
6 replies
0 recast
24 reactions
adnum
@adnum
I tried it the first time it became trendeing (several months ago) and it was awful. It made to many mistakes and it changed style of programming to many times in the same project. Now I read everybody is praising it but I am sceptical and do not want to try it eventhough is still installed on my mac Is it really that good now?
1 reply
0 recast
1 reaction
Ryan J. Shaw
@rjs
It makes mistakes, you need some experience to know when it's off-course. But I've been building a proper Dagster pipeline for the first time (I did 2 spikes previously, now I'm making properly classed resources, proper directory structure, tests, analysis scripts, etc.) and it's been helpful because it knows where I'm going next and that saves me mental energy. It does make mistakes like it suggested I could mark something as "backfillable=False" which is actually what I was looking for, but doesn't exist... but then I was digging deeper into the Dagster implementation to confirm it, so it kinda helped me figure out how to check for that sort of stuff too (I have like < 1 week python experience in my entire life)
1 reply
0 recast
1 reaction
adnum
@adnum
The last time I quit using it was because it took me too much time debugging the errors it generated. I felt like a was reviewing code (that I created) that was not mine. It was stranger to me and I had to dig deep in the project to understand and debug the code From what i have seen in videos , now its a tottaly different product and it suggests only pieces of code. Hope there is stil a trial period for me so I can give it a another test run I was a huge advocate of AI. But the more I used it (before summer) the more problems I encountered and and more unreliable the results were. I had the feeling that the quality degraded during usage.
1 reply
0 recast
1 reaction
