| 1 |
>>>>> On Fri, 6 Jul 2018, Marc Schiffbauer wrote: |
| 2 |
|
| 3 |
> * Michał Górny schrieb am 06.07.18 um 11:33 Uhr: |
| 4 |
>> If you don't see it for 5 years, how can you be sure that it is |
| 5 |
>> even still there? |
| 6 |
|
| 7 |
> Are you serious? Who tells you that I do not check from time to |
| 8 |
> time? |
| 9 |
|
| 10 |
> I am sure there will always be some scenario which makes a key |
| 11 |
> unacessible in some way. I do not disagree with that. Its a matter |
| 12 |
> of propability. |
| 13 |
|
| 14 |
> And for the worst case there is a revoke-Certificate which can be |
| 15 |
> used. |
| 16 |
|
| 17 |
Note that the revocation certificate is still listed under |
| 18 |
recommendations only, so devs need not create one. Making this a |
| 19 |
requirement would be a real improvement, IMHO. |
| 20 |
|
| 21 |
Instead, the GLEP draft is focusing on short expiration times. |
| 22 |
It won't help much if your compromised key will expire within one |
| 23 |
year, but you cannot revoke it. |
| 24 |
|
| 25 |
Suggestions: |
| 26 |
- Change the minimum requirement for key expiry to at most 3 years |
| 27 |
(which is what in version 1 is recommended). |
| 28 |
- Recommend at most 15 months of key expiry, to be renewed at least |
| 29 |
2 weeks before the expiry date. |
| 30 |
- Make creation of a revocation certificate (and storing it in a place |
| 31 |
separate from the key) mandatory. |
| 32 |
|
| 33 |
Ulrich |
Archives