On 08/13/10 06:14 PM, Brock Pytlik wrote:
On 08/12/10 10:31 PM, Shawn Walker wrote:
...
src/modules/actions/signature.py:
lines 94-97: Since you do an os.stat on path anyway on line 104,
it seems logical to do the stat once at the beginning and then
use stat.S_IS* checks on the result to determine whether to
raise exceptions, etc.
I find using the exists/isdir etc methods much clearer and easier to
maintain. Unless you're claiming a significant performance advantage to
the stat approach, I'd prefer to stick with how it is.
Danek has made the same suggestion to me in the past, so I mentioned it
here.
...
src/modules/client/api.py:
============================================================
lines 2288-2290: I don't really understand the need to revoke and
unset CAs during the *initial* addition of a publisher to an image.
At the very least, the unset seems odd. Can you expound?
Unset seems unlikely when adding a new publisher. Revoke you do need to
be able to do since a user may want to revoke a certificate the
publisher will set before it's ever used. Since there's no way to tell
early on whether a user is adding a new publisher or modifying an
existing one and unsetting a cert can't do any harm, I prefer to leave
this as is.
That's not true. There's an *add*_publisher method, and an
*update*_publisher method. add_publisher() can only be used for
publishers that don't already exist.
-Shawn
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
