On 06/04/2018 11:33 AM, Esteban Lorenzano wrote: > hi, > >> On 4 Jun 2018, at 19:11, Martin McClure <[email protected]> wrote: >> >> On 06/04/2018 05:34 AM, Esteban Lorenzano wrote: >>> hello Martin, >>> >>> This is becacuse metacello finds another version already registered. >> How is it already registered? This is a fresh image. The only thing that >> has happened is that it is already cloned, since I had loaded it in a >> different image. Deleting the cloned repository makes it work, but fails >> again on the next clean image until I delete the repository again. >> This does not seem correct. It was working correctly a week earlier >> (which is, I think, two Iceberg versions ago?) > it was not working properly. > what happened before is that iceberg was silently taking the version already > present. > this may be the preferred behaviour but in other contexts is plain wrong, > hence the need to force you to choose. Hi Esteban, Thanks for the explanation. I guess that makes sense. The terminology is a bit confusing -- "Loaded" is what you have on disk locally (but not in the image)? and "Incoming" is what's in the Git remote repository? > > this is why we introduced the #onUpgrade: protocol. > > btw, even if is not an “upgrade”, iceberg will fail to detect a reference to > (for example) tag v1.0.0 is the same as a commit SHA (we are working on this, > but it will not be anytime soon). Ah, so if I'm on, say, branch "master" and my local repo is on the same SHA as the latest commit to "master" on the upstream, it will still think it's an upgrade? Nice to know that this will (someday) be smarter. I know these things take time and effort, which is appreciated.
> > So, I guess you need to go for the #onUpgrade: mechanism. OK, I'll give that a try. Thanks again, -Martin
