Moinak Ghosh wrote:
On Mon, Apr 6, 2009 at 10:53 PM, Shawn Walker <swal...@opensolaris.org> wrote:
Moinak Ghosh wrote:
  Which means those packages will require system reboot not unlike
  Windoze!
First, how else can you ensure the scripts correctly run without having the
new environment loaded? (i.e. account for newly added users, hardware, etc.)


   That is very true for OS upgrades, but not necessarily for applications or
   some other layered software. Of course sw can be changed to do some

Right, but then you have to leave it up to the package creator to determine whether it can run in the current context, and that is where most SVR4 packages failed (think diskless clients, etc.).

By not allowing the package creator to determine the context, new contexts can be created (xvm, zones, etc.) without having to change the package to account for them, by allowing the packaging systems itself to do so.

   config at first start but it is not always possible to do that
practically unless
   the sw was designed with such things in mind. If scripting was being totally
   banned (I do find arbit scripting messy) then an alternative configured state
   should have been in the package design and facilities to do common config
   tasks should have been of priority. Depending on SMF IMHO is simply bad
   practice.

There are numerous avenues for configuration being researched, but by forcing no scripting, it has allowed the discovery:

* of a significant number of packages didn't need any custom scripting at all (including SMF) to deliver the software

* of gaps in provided action functionality

* of areas that still need to be accounted for

Remember that the focus is on delivering OpenSolaris, so some things like advanced custom configuration handling have not been a priority because they are not necessary to deliver OpenSolaris.

Cheers,
-Shawn
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org

Reply via email to