Hi Junio, On Mon, 28 Feb 2022, Junio C Hamano wrote: > Junio C Hamano <gitster@xxxxxxxxx> writes: > > > FWIW, CI run on "seen" uses this series. > > Another "early impression". I had to open this one today, > > https://github.com/git/git/runs/5367854000?check_suite_focus=true > > which was a jarring experience. It correctly painted the fourth > circle "Run ci/run-build-and-tests.sh" in red with X in it, and > after waiting for a while (which I already said that I do not mind > at all), showed a bunch of line, and then auto-scrolled down to the > end of that section. > > It _looked_ like that it was now ready for me to interact with it, > so I started to scroll up to the beginning of that section, but I > had to stare at blank space for several minutes before lines are > shown to occupy that space. During the repainting, unlike the > initial delay-wait that lets me know that it is not ready by showing > the spinning circle, there was no indication that it wants me to > wait until it fills the blank space with lines. Not very pleasant. > > I do not think it is so bad to say that it is less pleasant than > opening the large "print test failures" section and looking for "not > ok", which was what the original CI UI we had before this series. > But at least with the old one, once the UI becomes ready for me to > interact with, I didn't have to wait for (for the lack of better > phrase) such UI hiccups. Responses to looking for the next instance > of "not ok" was predictable. Let me again state my goal clearly, because some readers seem to be confused and believe that I want to improve the developer experience of veterans of the Git mailing list who are more than capable of finding their way through the build failures. My goal is instead to make new contributors' lives easier. It is a pretty high bar we set, expecting a new contributor faced with a build failure to figure out how to fix the breakage. It's as if we _wanted_ to instill impostors' syndrome in them, which is probably not intentional. In that respect, a relatively responsive page that utterly fails to direct the reader to the culprit is far, far worse than a slightly bumpy page that _does_ lead the reader in the right direction. In any case, thank you for integrating the patches into `seen` so that the impact of the patches can be "seen". Ciao, Dscho
