[deps] bump Bouncy Castle to 1.86 - #375
chadlwilson wants to merge 1 commit into
Conversation
Signed-off-by: Chad Wilson <29788154+chadlwilson@users.noreply.github.com>
|
I'll look into branch. |
|
The test matrix also seems totally out of control to me, which makes this take absolutely forever (and quite the waste of energy)
Anyway, seems there are some other problems here, so needs work. |
|
I'm going to roll with JRuby releases with just latest 0.16.x for now since these are fairly recent (no CVSS and some don't show up in databases yet) and this needs more time to clean up. @kares We'll want to spin a 0.16.x with this update too, but I'm not sure where to branch from. |
|
@chadlwilson The matrix could definitely be pared down. No non-LTS Java (except perhaps most recent), no JRuby releases prior to 9.3 or 9.4 (10 would make it impossible to do any extended 9.4 support), etc. |
Yeah, that makes sense. I haven't actually analyzed whether the CVEs are relevant to jruby-openssl usage, but I'd suggest that in a world of AI, we're gonna need to find a way to streamline the process for getting updates through the process into a jruby release, particularly for vendored gems like jruby-openssl, since they make breaking changes alongside non-backported security fixes. (presumably unless you're a paying customer) |
|
When we can loop in @kares we'll have a chat about long term management of jruby-openssl and BC. I have pushed a "0.16" branch based on the "v0.16.2" tag. |
Many more CVE fixes here: https://www.bouncycastle.org/resources/new-release-bouncy-castle-java-1-86/
Seems we need a 0.16.x branch, as we need a way to get security patches out into released jruby versions and master is already on
0.19.x?IMHO there is a truly excessive amount of boilerplate here just to add a new version to a matrix...