Shawn Walker wrote: > I also think it would be pretty silly to move the location of the image > version in the structure as that would defeat the whole purpose of adding > a versioning mechanism.
True, and I hope that what we have is sufficient for all time, but I tend to think that very few things are. > I felt it was best for the client to try as hard as possible to make the > image useable rather than simply fail. But if you want me to change the > path where I'm falling back to trying to detect the type of image based > on structure, and the image looks like it's a version 4 (or possibly > newer) image, and have the client fail with an exception, I'm willing to > compromise and do that. I'm completely fine using the image structure to determine an image's version, as long as the version we get is v2/v3 (or whatever versions we've already produced that didn't have explicit versions). I'm much less comfortable using image structure to determine the image version if it's a version we know should have a version number stored somewhere, but we couldn't find it. In particular, trying to determine the difference between a broken image (which we could then try to repair) and an image which is too new for us to recognize (at which point we should throw up our hands) seems dicey. If you can explain why you're comfortable with that, perhaps I could be persuaded. Danek _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
