Hi Jan.
>> Just curious: usr/src/lib/liborchestrator/Makefile: line 108:
>>
>> This Makefile is in a directory which looks platform generic, yet I see
>> a change to .../proto/root_i386/usr/snadm/lib. I see the same on line
>> 115, which was there before... (I assume this is the new installer as
>> this is a bugzilla bug not bugster bug.) Makes sense in that the new
>> installer runs only on X86...
>>
>> So how come this file isn't in a more platform-specific directory?
>>
>
>
> Thank you for pointing this out. You are right that this code relates to
> the new installer, which for time of being runs only on x86. But sparc
> platform is not out of scope - we will need to resurrect sparc at some
> point - according to the Draft PSARC slides Dave has sent out recently,
> it seems that it might be delivered with Indiana 2H 2008.
>
> According to this, liborchestrator will address both platforms and thus
> using 'i386' specific path should be avoided - I will take advantage
> of macro $(ROOTADMINLIB) which will do the right thing in this case.
>
Looks good.
>> - - -
>>
>> om_is_upgrade_target_valid() is now a stub. While removing references
>> to it may be beyond the scope of this bug, is it appropriate to open
>> another bug (or is there one already open) to remove all references to
>> it from other places?
>>
>
> om_is_upgrade_target_valid() is called from GUI function
> upgrade-screen.c:validate_upgrade_target() for validation of Solaris
> instance to be upgraded.
>
> In Dwarf Caiman project it took care of validation of Solaris instance
> installed on UFS for purposes of in-place upgrade.
>
> Since in-place upgrade is no longer supported, the appropriate code
> was removed from om_is_upgrade_target_valid().
>
> As time of being, GUI installer supports only initial installation,
> thus you are right that any code related to upgrade feature is not
> in-use for now.
>
> But since it seems probable that support for upgrade feature might
> be reenabled in GUI installer (UFS->ZFS migration, Snap upgrade ?),
> then om_is_upgrade_target_valid() would implement new code for validation
> of Solaris instance (either UFS or ZFS) to be upgraded.
>
> If you agree, I would keep it for now and if we are not sure about
> upgrade support in GUI I might file bug in order to be aware of this.
> If we later decide that only initial installation will be supported
> from GUI installer, all upgrade related code might be then removed from
> orchestrator as well as from GUI.
>
> Please let me know, what you think.
>
Seems reasonable to me.
Does it make more sense to have om_is_upgrade_target_valid() return
false instead of true since upgrading is not supported with the new
installer at this time?
Thanks,
Jack
>>> Webrev is available at
>>> http://cr.opensolaris.org/~dambi/bug-683/
