Merge lp:~maxb/udd/import-ordering-fix-take-2 into lp:udd
Status: | Merged |
---|---|
Approved by: | James Westby |
Approved revision: | 451 |
Merged at revision: | 452 |
Proposed branch: | lp:~maxb/udd/import-ordering-fix-take-2 |
Merge into: | lp:udd |
Diff against target: |
135 lines (+44/-58) 2 files modified
icommon.py (+32/-49) tests/test_package_to_import.py (+12/-9) |
To merge this branch: | bzr merge lp:~maxb/udd/import-ordering-fix-take-2 |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
James Westby | Approve | ||
Review via email: mp+63335@code.launchpad.net |
Description of the change
Jelmer, you were so very right when you suggested import ordering could be better done based on versions at the bzr sprint in London.
What neither of us realized at the time was that importing sid first in all cases only holds true as a correct method if sid contains all older releases in its ancestry .... which is true in theory, but not in Launchpad's database.
In fact, Launchpad's "sid" only goes back as far as lenny, and woody, sarge, and etch are imported by the UDD importer via package indices from archive.debian.org.
Fortunately, this miscalculation of mine only causes suboptimal history to be produced when doing fresh imports of package versions around the etch / lenny boundary, so should not have had much if any impact.
The comments are much clearer now too, thanks.
James