On 14/04/2011 20:45, Matthias Kilian wrote:
On Thu, Apr 14, 2011 at 11:57:51AM -0400, Edward Z. Yang wrote:
The remote repository has been reset to 9eebc6dec9a2271a5179, removing
the two merge commits that Simon pushed. If you do not have any commits
more recent then that in your working copy, you don't have to worry about this.
And if someone already has local commits?
Please, never ever do something like this again to the main public
repository from which almost all people are pulling.
Given that some (or even many) people are still unfamiliar with git
(including myself), if an accidents like commiting to the wrong
branch happens, it should be fixed by another commit reverting the
changes, not by fiddling with the upstream repository.
Maybe I don't grok git enough, but IMHO the correct handling would
have been to merge Simon's generics fix
(ce2ea8274f72199ac32d5219fcadb0aaeb968707) from head into the
generics branch, then backout it from head (by committing a reverse
diff).
We considered doing it this way, but came to the conclusion that there
would be problems either way, so we opted for the solution that left the
repo in the cleanest state.
Whether this was the best decision or not I'm not completely sure, not
being a git expert, but fortunately we did have two git experts on hand
to deal with it (ezyang and tibbe), so thanks to them for sorting it all
out.
Cheers,
Simon
_______________________________________________
Cvs-ghc mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/cvs-ghc
