Brian,

I wouldn't do an EOF case for this change.  The biggest
reason is because this has not been in a revenue release.
Simply create package history files so that the appropriate
thing happens when upgrading.

Thanks,

John

Brian Cameron wrote:
> 
> John:
> 
>> Support for an interface has no barring on the EOF process.
>> For Committed and Uncommitted interfaces one must go through
>> the EOF process.  For Volatile interfaces it is a little bit
>> more difficult to discern when to follow the process and not.
>> For example, if the security team were to remove OpenSSL they
>> would need to go through the EOF process because they have
>> so many projects depending upon them.  If sfw decided to
>> remove the animate binary from /usr/sfw then they could
>> probably do so without the process.
>>
>> Which interface(s) are you concerned about? 
> 
> We are talking about the libsexy package names, SUNWlibsexy
> and SUNWsexy-python.  Package names are typically defined to
> be Uncommitted.  These package names are defined in the Compiz
> LSARC case (2008/115).
> 
> We know the GNOME community plans to merge libsexy's library
> functionality into GTK+, and the libsexy library will eventually
> go away.  So, we have been discussing whether it would make more
> sense to put libsexy in a package like SUNWgnome-base-libs.  This
> way we do not have to remove the Uncommitted SUNWlibsexy and
> SUNWsexy-python packages later.
> 
> The reason our team tends to bundle multiple modules in generic package
> names like SUNWgnome-media, or SUNWgnome-base-libs because this gives us
> some flexibility to add/remove content from the packages without
> affecting package names.  Since package names are typically Uncommitted,
> this means we would need to go through the EOL/EOF process if we remove
> the package, or if we replace one module with another (like replacing
> gnome-cd with sound-juicer).  Isn't this correct?
> 
> If we decide to put libsexy modules in SUNWgnome-base-libs or some other
> generic package, then do we need to do anything to inform ARC that we
> will not be adding the package names as specified in the LSARC
> 2008/115 case?
> 
> Your thoughts?
> 
> Brian

Reply via email to