Dan Price wrote: > On Wed 18 Nov 2009 at 04:01PM, Danek Duvall wrote: > > Richard Lowe wrote: > > > > > http://cr.opensolaris.org/~richlowe/pkg_11609 > > > > Looks okay to me. > > Can I ask: what are the implications here? Will existing binaries fail > verification after this change?
If they're out of date with respect to the file hash, then yes, they'll fail to verify. But we do check the file hash when the content hash fails and don't fail verification if the file hash is okay. See the comment on line 277 of file.py. We put this in when we stopped hashing .SUNW_dof some time ago. We don't get quite as lucky in terms of verification noise as we did back then, unfortunately, but it's better than broken DTrace. Note that the first time you upgrade to a system published with this changeset, you'll be fine again, as you'll have re-downloaded all the bits that would have verification problems. And since most folks won't be using pkg(5) bits with this changeset until that happens, they generally won't see any problems. > Will I be able to install software from e.g. extra/ without republishing > with the new hash algorithm? I'm not sure that any software in extra/ has .SUNW_ctf sections, but as long as the file hash matches, it doesn't matter what the content hash is, so you'll only see problems on upgrade where we decided not to re-download because the content hash was the same. Danek _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
