From 407a35e6e9390ce0b4f1ca16c3e422298de23b79 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Vladim=C3=ADr=20=C4=8Cun=C3=A1t?= Date: Sun, 28 May 2023 08:37:19 +0200 Subject: [PATCH] 0149-cache-key-rotation: add some alternatives --- rfcs/0149-cache-key-rotation.md | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/rfcs/0149-cache-key-rotation.md b/rfcs/0149-cache-key-rotation.md index a38f2a4..795b347 100644 --- a/rfcs/0149-cache-key-rotation.md +++ b/rfcs/0149-cache-key-rotation.md @@ -58,7 +58,13 @@ Why should we *not* do this? # Alternatives [alternatives]: #alternatives -What other designs have been considered? What is the impact of not doing this? +- change nothing, obviously +- also resign old `*.narinfo`. Maybe it's not too hard. + It would help people wanting ot use old builds. +- double-sign `*.narinfo` for some time. (also not an exclusive alternative) + I don't know if consumers support multiple signatures. + It doesn't seem to give us significant advantage though; + acceptance of multiple keys seems more advantageous. # Unresolved questions [unresolved]: #unresolved-questions