[OmniOS-discuss] On pkg(1) behavior in r151022

[Jim Klimov]

21 февраля 2017 г. 3:24:58 CET, Bob Friesenhahn <bfriesen at simple.dallas.tx.us> пишет:
>On Mon, 20 Feb 2017, Dan McDonald wrote:
>
>>
>>> On Feb 20, 2017, at 6:17 PM, Bob Friesenhahn
><bfriesen at simple.dallas.tx.us> wrote:
>>>
>>> If someone forgets to use -r the first time around, can the same
>command be invoked again with the -r and then the zones get the updates
>as well?
>>
>> Yes.  The pkg5 test suite explicitly covers that, so I know it works.
>
>This is very useful.  Often the zones are located outside of the root 
>pool.  After updating just the root pool and seeing that it is working 
>well, then the zones can be updated.  This reduces the chance of 
>problems when updating the zones.
>
>Bob

I'm in favor of keeping options consistent across the ecosystem, so scripts and habits are portable going forward. So '-r' behavior and option should be explicit as a new addition.

It does make sense to add a run-time warning for the old usecase(s) which changed behavior - like "This command used to implicitly recurse into local zones, now you must specify -r to get that behavior." - and that should serve to recover the least surprise approach.

Also mark this in release notes for people upgrading from older non-bloody releases (there's a chapter for deprecations and big changes, right?) and you're golden.

Jim 
--
Typos courtesy of K-9 Mail on my Samsung Android