Re: hg diff --subrepos

"Stephen J. Turnbull" <stephen(a)xemacs.org&gt; writes: > Mats Lidell writes: > > >>>>> Michael Sperber <sperber(a)deinprogramm.de&gt; writes: > > > > > That's dangerous, as it doesn't necessarily give you the > > > *consistent* set that Norbert is managing. > > > > Yes if you mix up doing a consistent pull with doing a pull on all > > subrepos. So in that sense onsub might be evil since it makes that > > error easier to make. > > As I wrote before, it's not an error (bug). The problem is that an > onsub pull gives you a workspace with *no consistency guarantees at > all* in our current workflow. What we're trying to work out here is > "what consistency guarantees should we provide to the smoke test?" > Not to mention, Norbert. :-) I think you're right and I was wrong: Package authors may want to use the smoketest to find out if their packages builds outside their own environment, and push something there that hasn't gone through Norbert. If that's true (and I think it is), Mats should indeed to an onsub-pull to feed the smoketest.