Bug#593241: flavours "486 686" by default for i386 needs testing
When I talked to Mark about this, I thought he had tested the default
(i.e. no --linux-flavours specified). It turns out he had not. He had
explicitly specified "486 686". But my reading of
/usr/share/live/build/functions/defaults.sh at line 554 (live-build
2.0~a23-1) is that the wrong default of "486 686" will be used here.
Since the installer only provides a 486 kernel, at least the default
case should be only "486" when an installer is included on the image (I
didn't check this -- is this true for the launcher *and* the install
boot option or just the launcher?) Furthermore, it may even be
appropriate to issue a warning if flavours other than what the installer
supports are specified (or even an error? what's the valid use case for
I see some other cases in defaults.sh where multiple kernel flavours are
specified for other architectures. Do they all match what's provided by
d-i for those architectures, or do we have a similar problem there, too?