On Wed, Jul 08, 2009 at 10:55:46PM -0500, Shawn Walker wrote: > I didn't know that you wanted to limit it to working within the > refresh_seconds interval of time. I figured we didn't want to do that > since refresh_seconds will eventually come from the repository metadata > instead of being defaulted to four hours. > > My personal belief has been that if an implicit refresh (such as that > done by install, image-update, etc.) can't be performed due to network > connectivity issues, then we should not fail or require --no-refresh. > The failure should occur later when we attempt to retrieve a manifest, > etc.
I've changed the test to remove the last_refreshed file, and to install with --no-refresh. You and I can argue about whether this error should be suppressed if it happens in the --no-refresh case some other time. IMO, the test shouldn't be performed if we're refreshing the catalog and don't care about the result, but currently that concept is too complicated to encode in the transport, unless we're willing to run the captive portal test multiple times. I'd rather not, and consequently that requires us to keep this simple, at least for now. -j _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
