IMHO is not even resembling: there (if this is about 0 byte "marker" file) a _state_ was recorded (and kept) on disk, making the disk carrying the "state" that determined the outcome of the build. here, a file is being read by JDK, and it is read successfully on many, and failed to read at Elliotte workstation.
Unsure what he tried to build in the first place, as it is possible that source ZIP is corrupted, I have no idea. But one thing for sure, if this is the case, then 1.9.8, 1.9.7, 1.9.6 releases should be -1 according to him, as there is no change in 1.9.9 related to anything in this area. T On Wed, Apr 26, 2023 at 4:52 PM Michael Osipov <micha...@apache.org> wrote: > Am 2023-04-26 um 16:44 schrieb Tamás Cservenák: > > Howdy, > > > > Am fine with that, what I don't understand is: > > 1. how are both CIs on 3 different JDKs not catching this? > > This is logically identical to MSITE-960, not catched by me and > partially not in CI. Took me hours and there was one serious bug AND a > subsequent failure due to the broken setup dene previously. > Rather check twice than fail in production. > > M > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org > For additional commands, e-mail: dev-h...@maven.apache.org >
