Hello, On Mon 04 Mar 2019 at 06:03PM +00, Ian Jackson wrote:
> AFAIAA there is no way for dgit-infrastructure or dgit to know whether > something was dcut or rejected. Right. >> Shouldn't --deliberately-include-questionable-history be required only >> once? Shouldn't dgit-infrastructure learn that the commits it thinks >> are tainted aren't actually tainted? > > Each time the thing is cut and repushed, presumably there are new > commits that might be tainted. > > Ie, > push ...A > cut now ...A is tainted > push ...A+B needs reassurance that A is OK > cut now ...B is tainted > push ...A+B needs reassurance that A+B is OK > etc. > > Don't you think ? Once I --deliberately the 1.2-1 upload, and dgit sees that it was ACCEPTed, can't it safely untaint everything in the 1.2-1 history? That way, eventually, all commits will be marked as untainted and I won't have to keep passing the option until the end of time. If not, I think there needs to be some moderately standard way to request that the whole current history of a package on dgit-repos be marked as untainted. -- Sean Whitton
signature.asc
Description: PGP signature
