svn commit: r344567 - in head: etc/mtree include sbin sbin/veriexec

[Warner Losh]

On Tue, Feb 26, 2019, 6:50 AM Alexey Dokuchaev <danfe at freebsd.org> wrote:

> On Mon, Feb 25, 2019 at 10:19:45PM -0800, Rodney W. Grimes wrote:
> > > Author: sjg
> > > Date: Tue Feb 26 06:17:23 2019
> > > New Revision: 344567
> > > URL: https://svnweb.freebsd.org/changeset/base/344567
> > >
> > > Log:
> > >   Add verifying manifest loader for mac_veriexec
> > >
> > >   This tool will verify a signed manifest and load contents into
> > >   mac_veriexec for storage
> > >
> > >   Sponsored by:     Juniper Networks
> > >   Differential Revision:    D16575
> >
> > Just a small nit, for future reference, from the template:
> > > Differential Revision:    https://reviews.freebsd.org/D###
> > (*full* phabric URL needed).
>
> IMHO we should just fix the scripts to accept D#### alone, without
> the URL.  We don't do that for PR (bugzilla) references, and the fact
> that we do for the phab is both needless and confusing.  Also, that
> URL might change one day while we could probably keep the old numbers
> if we move.


When this came up before, it was an upstream decision to require the full
path. The reasoning is that Dxxxx space isn't unique and the review may
happen in a different instance of phab than the default one. When those
exact objections were raised, the experience of moving to bugzilla was
cited to show the URL can remain stable. And if we went to another
system's, we could keep the old in place and then we would need a new URL
to disambiguate. There were good reasons we decided this before.

So this has been litigated before, and the consensus was we needed to break
with the practice of not putting URLs in commit messages for these reasons.

Warner

>