Hi, that's because the PR is a based on a pre-loom version of master.
Best regards Christoph > -----Original Message----- > From: build-dev <build-dev-r...@openjdk.java.net> On Behalf Of Philip Race > Sent: Montag, 23. Mai 2022 17:14 > To: Aleksey Shipilev <sh...@redhat.com>; Ioi Lam <ioi....@oracle.com>; > build-dev@openjdk.java.net > Subject: Re: pre-submit tests for github PRs > > BTW I am wondering whether I should believe this explanation since it does > pass sometimes, > > eg : > > https://github.com/openjdk/jdk/pull/8820 > > How is that possible ? > > -phil > > On 5/23/22 8:05 AM, Aleksey Shipilev wrote: > > On 5/23/22 16:53, Ioi Lam wrote: > >> On 5/23/2022 4:41 AM, Alan Bateman wrote: > >>> On 22/05/2022 23:58, David Holmes wrote: > >>>> > >>>> GHA tests a range of OpenJDK ports, not just the "mainstream > >>>> platforms". > >>>> > >>>> The existing linux-86 failures (and others) are due to the Loom > >>>> integration which only fully supports a couple of platforms and > >>>> which broke all the other ports upon initial integration. Some > >>>> folks in the community have been working through things to fix that. > >>> > >>> GHA is configured to cross build for most of these > >>> ports/configurations and they should pass. Running them is another > >>> matter of course. GHA does run some tests on linux-x86 and the > >>> x86_32 port does need work before it can run the tests with --enable- > preview. > >>> Aleksey Shipilëv seems to be making good progress on it. > >>> > >>> -Alan > >> > >> Maybe it makes sense to disable the x86 GHA runs before the > >> underlying issue is fixed? There's too much noise. > > > > There is a slightly better way to do it: > > https://github.com/openjdk/jdk/pull/8843 > >
