If a test works for EFI, it seems like it also qualifies for x86_64. Maybe the exception to this is for booting, where the x86_64 column indicates x86_64+BIOS, and EFI indicates x86_64+EFI. Right?
Chris Murphy
On Tue, 2013-10-08 at 13:43 -0600, Chris Murphy wrote:
If a test works for EFI, it seems like it also qualifies for x86_64. Maybe the exception to this is for booting, where the x86_64 column indicates x86_64+BIOS, and EFI indicates x86_64+EFI. Right?
No, not really.
Well.
The intent is that the EFI column is greyed out where it is extremely unlikely EFI would ever behave differently, and white where it is possible that it may. I more or less pulled the list out of my ass, though. If you disagree with any of the calls I made in doing it, do suggest a modification, but in terms of the *general approach*, it is that the 'x86_64' column represents x86_64 BIOS, and UEFI represents x86_64 UEFI.
On Oct 8, 2013, at 1:59 PM, Adam Williamson awilliam@redhat.com wrote:
but in terms of the *general approach*, it is that the 'x86_64' column represents x86_64 BIOS, and UEFI represents x86_64 UEFI.
OK good. Followup question: Is Mac EFI a suitable substitute for UEFI? In a sense Mac EFI is more limited, so any failure there doesn't necessarily mean broader failure with UEFI. Mac EFI is sortofa canary in a mine. If it works, then UEFI should work.
Chris Murphy
On Tue, 2013-10-08 at 14:41 -0600, Chris Murphy wrote:
On Oct 8, 2013, at 1:59 PM, Adam Williamson awilliam@redhat.com wrote:
but in terms of the *general approach*, it is that the 'x86_64' column represents x86_64 BIOS, and UEFI represents x86_64 UEFI.
OK good. Followup question: Is Mac EFI a suitable substitute for UEFI? In a sense Mac EFI is more limited, so any failure there doesn't necessarily mean broader failure with UEFI. Mac EFI is sortofa canary in a mine. If it works, then UEFI should work.
Well, there's always a slight element of subjectivity in filling out the matrix. If you have an extremely unusual bug that happens to affect your test configuration but would likely not affect anyone else's, you probably won't mark the box as 'fail' (perhaps 'warn'). I'd say that applies in this case: if the failure was one likely to affect other UEFI configs I'd mark it as 'fail', if it was exclusive to the config being tested, I'd say 'warn', it it works, 'pass'. If it's something that likely only affects EFI Macs, that's an...interesting question, but we do seem to mostly be on the 'block for Macs' side of the question at present, so probably 'fail'.
Ultimately we can have multiple results for a matrix entry and the entries are only guidance for human evaluation anyway, so it's not too terribly important: the most important thing is that all potential blocker bugs are nominated as blockers.