| 1 |
On 6 July 2015 at 08:01, William Hubbs <williamh@g.o> wrote: |
| 2 |
> Once we have a version of git stable that allows this, can someone fill |
| 3 |
> me in on why we would need to sign commits if we sign pushes? If we have |
| 4 |
> a signature on the push, we know where that came from, so it seems to be |
| 5 |
> overkill to sign the commits as well. |
| 6 |
|
| 7 |
The TL;DR of "why" though is basically: It allows a verifiable record |
| 8 |
of which user set BRANCH/$ID to == $COMMITSHA1. |
| 9 |
|
| 10 |
That is all. |
| 11 |
|
| 12 |
It doesn't verify the commits themselves, only where they are visible. |
| 13 |
( ie: so if there's experimental dangerous stuff in a side branch, a |
| 14 |
malicious dev can't point master to there and blame the person with |
| 15 |
the signed commit for breaking the tree ) |
| 16 |
|
| 17 |
|
| 18 |
|
| 19 |
I noted that even though I can sign pushes now, upon experimentation I |
| 20 |
found the server couldn't respond to signed pushes. |
| 21 |
|
| 22 |
Seems you need to upgrade git on the receiving side *first* to make |
| 23 |
the feature even an optional thing. ( Github for instance does not |
| 24 |
support signed pushes yet either ). |
| 25 |
|
| 26 |
So we'd have to get it "working" in an optional state long before we |
| 27 |
could get it to be mandatory. |
| 28 |
|
| 29 |
|
| 30 |
-- |
| 31 |
Kent |
| 32 |
|
| 33 |
KENTNL - https://metacpan.org/author/KENTNL |
Archives