Richard Lowe wrote: > Richard Lowe <richlowe at richlowe.net> writes: > > Just a reminder that the below still needs review. > > -- Rich >
The changes look good to me as they are. -dvd > >> Hey all, >> >> Please review: >> 6742650 cdm restore error could be more useful for insufficiently >> updated destination >> >> Webrev: >> http://cr.opensolaris.org/~richlowe/onnv_6742650 >> >> This expands upon the error you receive when attempting to restore >> from cdm backups into a workspace which is insufficiently up-to-date, >> or is not related to the workspace from which the backup was taken. >> (these cases are indistinguishable, to us). This has been a source of >> questions for us (from two or three people, but that's still a >> reasonable percentage of the questions I've seen on IRC). >> >> Prior to these changes (such as with the present tools), you get an >> error like so: >> >> adding changesets >> transaction abort! >> rollback completed >> abort: 00changelog.i at 4bc1ca7b6767: unknown parent! >> >> With these changes, you get some extra text indicating the likely >> cause of this issue: >> >> adding changesets >> transaction abort! >> rollback completed >> abort: Error restoring workspace: >> couldn't restore committed changes: >> 00changelog.i at 4bc1ca7b6767: unknown parent >> Your workspace is either not sufficiently up to date, >> or is unrelated to the workspace from which the backup was taken >> Workspace may be partially restored >> >> We can't provide more information about the node we were missing (as >> we don't have it to look at...), only indicate likely reasons that it >> is missing. >> >> It is possible that we catch too much in the way of error here, in >> concept you may fail a node lookup in something other than the >> changelog (indicating a far larger problem). >> >> The only we could differentiate in this case would be to parse the >> '00changelog...' message, which is a string including verbatim in the >> exception raised by Mercurial. That seems rather brittle to me, so I >> chose not to do so. I will do so if advised. (I would add that, to >> the best of my knowledge, if you receive this error about something >> other than the changelog, the chances are that your workspace is >> corrupt, and nothing we can say would help you.) >> >> Testing: >> >> Attempted to restore backup into an unrelated workspace (this >> workspace into an scmtest workspace). >> >> Attempted to restore into an artificially down-rev workspace. >> >> Both cases worked as I expected, and gave me the new message as >> above. >> >> pbchk: >> pbchk -q is silent. >> >> Thanks, >> >> -- Rich >> > _______________________________________________ > scm-migration-dev mailing list > scm-migration-dev at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/scm-migration-dev >
