Oh yes, very good to do this. Not sure about the migration woes! I think it's OK to assume that operators should have saved their secret already because we have warned them that this secret can never be viewed again (without serious work-around hacks). I don't imagine many people are using this recipe yet, so it might be OK to flag that this is a big breaking change, give it some time and then release it. I myself have my secrets stored so can bring it down / re-release and re-wire it up to match the new config.
Oh yes, very good to do this. Not sure about the migration woes! I think it's OK to assume that operators should have saved their secret already because we have warned them that this secret can never be viewed again (without serious work-around hacks). I don't imagine many people are using this recipe yet, so it might be OK to flag that this is a big breaking change, give it some time and then release it. I myself have my secrets stored so can bring it down / re-release and re-wire it up to match the new config.
Not sure what happened yesterday to cause that issue but it's gone now, maybe I had a git repo desync issue or something.
Tested the migration steps and also tested the fresh install steps in the readme and both work!
Not sure what happened yesterday to cause that issue but it's gone now, maybe I had a git repo desync issue or something.
Tested the migration steps and also tested the fresh install steps in the readme and both work!
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
By using two user-controlled secrets, we can support rotation of the encryption keys as described in the docs 🎉
Also enabled HTTP -> HTTPS redirection
Hmm, when trying to test the migration steps...
Oh yes, very good to do this. Not sure about the migration woes! I think it's OK to assume that operators should have saved their secret already because we have warned them that this secret can never be viewed again (without serious work-around hacks). I don't imagine many people are using this recipe yet, so it might be OK to flag that this is a big breaking change, give it some time and then release it. I myself have my secrets stored so can bring it down / re-release and re-wire it up to match the new config.
Not sure what happened yesterday to cause that issue but it's gone now, maybe I had a git repo desync issue or something.
Tested the migration steps and also tested the fresh install steps in the readme and both work!
One minor comment but otherwise LGTM.
Thanks!
@@ -11,3 +11,2 @@SECRET_ENC_KEYS_VERSION=v1ENC_KEY_ACTIVE=""SECRET_ENC_KEYS_A_VERSION=a1 # generated=false❓ toolshed/abra#665 (comment) ❓
Good catch!
LGTM!
@@ -2,0 +5,4 @@generate_enc_keys() {KEY_A="$(openssl rand -base64 32)"KEY_B="$(openssl rand -base64 32)"abra app secret insert "$APP_NAME" enc_keys_a a1 "$KEY_A" --chaosReminds me of toolshed/abra#571 btw
The migration was a bit more painful than originally imagined 🙈
#11