The idea is to use "Git magic" to make it all work; there is no cherry picking of patches. Instead a "hack on Git merges" is used, which preserves the SHA1 valu

The idea is to use "Git magic" to make it all work; there is no cherry picking of patches. Instead a "hack on Git merges" is used, which preserves the SHA1 values as they are in the existing stable branches. While that is really good from an operational POV, one of the main features (the main feature?) of the SHA1 was that it is dependent cryptographicly on the preceding commit and recursively to the initial commit, giving assurances that the whole tree is valid and unchanged. It seems like "hacking on Git merges" like it would break the cryptographic assurances.