Kristian H. KristensenI have another thing to the list of hardest things in computer science, maybe even a new #1: reproducing the regression outside CI.
Martin Roukala (né Peres)@krh Yeah.... that's why with the CI infra I've been developing (#CI_tron), we run containers directly from an initrd (#boot2container). It doesn't solve the reproducibility of results across kernels and hardware, but at least the userspace is covered!
@mupuf When I see a failure in gitlab CI, I just want a "Shell Here Now" button.
@mupuf Dont make me learn docker, don't make me download images, just give me a shell, right now, right here, in the browser where it failed.
@krh Oh, yeah. That's something gitlab supports with the SSH executor, but not with the docker one.
I am planning to support that in CI tron because it is such a useful feature indeed!