Hi,

Philip Brown wrote:
> The "issue" of "dont allow scripting at ALL", is invalid, to me,  and 
> others, such as Moinak.
> To reduce the number of common errors in postinstall scripts, just requires 
> that sun provide pre-canned utilities and class action scripts.

There are standard class action scripts and things like /usr/bin/pkgcond 
were introduced so that people can work out what the appropriate action is 
in a postinstall script. Ok there is still user scripting involved but they 
don't have to work out everything.

Now I love the idea of getting rid of scripts. Even when a developer writes 
a perfect script for their application it can break when some new feature 
is introduced to the OS - zones being the best example; so getting rid of 
that possibility seems great.

A big problem though, in SVR4, is that users always will come up with 
something that needs to be done that we haven't thought of and are free to 
write their own scripts. A corollary for that is that when we give them 
limited options for what they can do then they may not be able to do what 
they need.

And thats before we even tackle the many patches that have special install 
instructions where the procedures have not / can not be scripted.

I guess my fear is that if applications cannot do what they need to do with 
the use of actuators then they will rely on posting manual steps in release 
notes or something and giving the user more work and a lousy experience. 
Worst case the user will complain that he didn't have to do these steps 
when he added the patch on s10 but does in opensolaris!

Cheers,
~Al


-- 
Albert White - Sun Ireland
[EMAIL PROTECTED]
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to