On Mon, Jun 30, 2008 at 09:37:34AM -0700, Jordan Brown wrote:

> While the record of what pkg install / image-update actually does may be 
> interesting for diagnosis and as a midpoint check, it does not seem like 
> it's the right thing to be looking at to determine whether the operation 
> was done correctly end-to-end.  After all, the list of files updated is one 
> of the things that could be wrong.  It seems like a test like this should 
> compare the expected contents of the package with the actual contents of 
> the package and, moreover, should compare the expected contents of the 
> package with what's actually on the system.

Again, "pkg verify" will do just that, checking hashes with -f.  We haven't
forgotten to install the man pages along with everything else, have we?

If Albert wants to perform such a test outside the pkg(5) framework, that's
fine, but he'll need to be careful not to reimplement everything, and
probably do the test based on a hardcoded set of data.

Danek
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to