heya, not sure if this is the right place to leave...
# report-bugs
v
heya, not sure if this is the right place to leave this sort of feedback, but here goes. I noticed a new check summary for the first time, so I hovered over it and read the fairly wordy description - and it was wrong. It's somewhat frustrating to have the same UI element both tell you that there is a build failure, but also that nothing is wrong. Particularly because the logs unambiguously show that there was an error. I totally get that genAI features will be wrong sometimes and helpful other times. It's very hard to design and build around that unpredictability. So I'm just leaving this as my experience report - it was frustrating to have an incorrect summary presented to me. There is already a lot competing for my attention on the graphite PR page, tbh I don't really want genAI text interleaved with everything else
p
(If applicable) To help us repro, please send the following if you haven't already: - PR links, screenshots, and/or debug output if applicable - Repro steps - Does this happen consistently? Just to you or others on your team also? - Is this a regression? Is the problem new? - If you're having an issue with the CLI, please send us debug logs using
gt feedback --with-debug-context <any message>
Hey Victor — this is a fine place for feedback! Sorry for the friction here. I've just shared this with the team so we can take a closer look. We're working diligently on improvements for this 🙏
The team is wondering what you might expect the description to say?
v
I think I'd prefer the truncated description at the top level PR view just not be there at all. the important thing is the name of the build step that failed. I find the PR view kinda busy with lots of text, so subtracting would be good for me (I do think the redesign improved things somewhat)
👍 1
for the un-truncated summary, it might be useful to just say something like "one failed test" or "lint error"
I think I could have finished scanning through my whole build output in the time it takes to read the paragraph that is there
p
Got it, thanks!
v
i recognize its a tricky design problem, a CI build step can have lots of different types of output, different types of failures. but as a user i'm very familiar with the structures of my CI output and can scan CI logs very quickly and spot familiar shapes. But i can't quickly scan a paragraph of AI text
🫡 1
np!