| 1 |
On 07/03/2015 05:19 PM, Andrew Savchenko wrote: |
| 2 |
> |
| 3 |
> As I see from git docs only commits and tags may be signed. There |
| 4 |
> is no way to sign a push. |
| 5 |
|
| 6 |
This was new to me, but check out the "--signed" flag of git-push (1). |
| 7 |
|
| 8 |
|
| 9 |
> Moreover there is no need to sign each |
| 10 |
> commit, see what Linux says on that: |
| 11 |
> http://git.661346.n2.nabble.com/GPG-signing-for-git-commit-td2582986.html |
| 12 |
> |
| 13 |
> '' |
| 14 |
> Btw, there's a final reason, and probably the really real one. |
| 15 |
> Signing each commit is totally stupid. It just means that you |
| 16 |
> automate it, and you make the signature worth less. It also doesn't |
| 17 |
> add any real value, since the way the git DAG-chain of SHA1's work, |
| 18 |
> you only ever need _one_ signature to make all the commits |
| 19 |
> reachable from that one be effectively covered by that one. So |
| 20 |
> signing each commit is simply missing the point. |
| 21 |
> '' |
| 22 |
|
| 23 |
I think the next sentence is relevant: |
| 24 |
|
| 25 |
IOW, you don't _ever_ have a reason to sign anything but the "tip". |
| 26 |
|
| 27 |
My interpretation is that it doesn't make sense to sign commits one |
| 28 |
through nine if you're going to sign the tenth before pushing. But most |
| 29 |
of our commits are small and self-contained so it's probably easier to |
| 30 |
automate the signing with repoman than it would be to come up with a |
| 31 |
to-sign-or-not-to-sign guide a mile long. |
Archives