Liane Praza wrote: > I'd still like to see the actual dependencies for installing single-user > patches enumerated rather than encode more infrastructure around the > current ad hoc mechanism.
Unfortunately, they are very hard to enumerate. The true dependencies (and anti-dependencies or whatever you call the "must not be running" requirements) are dependent on the particular patch. Pretty much it's "with the system as quiet as you get get it", but with enough stuff mounted so that the patch software and the software to be patched are mounted. Preferably there is absolutely no background activity, because one of the primary fears is that some newly-updated kernel module will get loaded while we're working and will be incompatible with the running kernel. > I'm also wary of doing a bunch here now if the actual goal is to make > single-user patches go away because of how IPS works. Sun will be supporting S10 for about another ten years. My group supports versions of our software back as far as S8. Life is a little different in the patch world than it is on the bleeding edge.
