feat: bump bindle to v0.9.0-rc.1#786
feat: bump bindle to v0.9.0-rc.1#786FrankYang0529 wants to merge 3 commits intospinframework:mainfrom
Conversation
499e7d0 to
5371607
Compare
vdice
left a comment
There was a problem hiding this comment.
Thanks for this much-needed PR @FrankYang0529! I'm excited to test.
The updates look good/correct at a high level. The one question I had was around the scenario when spin (specifically its bindle client) doesn't share the keyring file with the bindle server, eg when the bindle server is fully remote. Do we have logic here to fetch the public key from the bindle server in this case? (... speaking of, I wonder if it would be worth adding a new test for this scenario as well -- I love seeing how we already have existing tests that have been updated!)
|
Hi @vdice, I agree we need to fetch the public key. However, even if I use different folders for client and server, I still can run Setup Push spin app Run spin app |
I know the reason! We didn't update bindle package version to |
5371607 to
b2109b8
Compare
Hi @vdice, I updated |
|
I will resolve conflicts tomorrow. |
👍 I'll be away next week (10/3-10/7) but hopefully another maintainer can take a look during that time as well. I do know that we'll probably want to wait to merge until after the next v0.6.0 release. Thanks! |
5a37fe9 to
3915f41
Compare
d0f4edb to
978787a
Compare
|
Resolve conflicts with #794. Add Setup Push spin app Run spin app Start Hippo Login and deploy |
|
@FrankYang0529 Thanks so much for updating to work with the latest spin deploy changes! I hope to revisit this PR asap. One other item I was thinking about... thoughts on how we'd surface spin-bindle incompatibilities? For instance, if a user is attempting to deploy using spin v0.5.0 and the bindle server is 0.9.0+, how best to relay that this combination isn't supported. Conversely, when we release a version of spin updated to talk to 0.9.0+, how best to make it obvious that the newer spin client can't talk to previous bindle server versions? |
Yeah, this is a good question. If we can determine bindle-server version, we can give a proper message to users. Do you think it's good to return bindle-server version in |
Indeed, if bindle-server doesn't yet return its version in another endpoint, that seems like a good candidate. I'll create an issue there just in case other bindle maintainers have input. |
978787a to
fb85df1
Compare
fb85df1 to
1ba8e0e
Compare
Signed-off-by: Frank Yang <yangpoan@gmail.com>
Signed-off-by: Frank Yang <yangpoan@gmail.com>
Signed-off-by: Frank Yang <yangpoan@gmail.com>
1ba8e0e to
48066a4
Compare
|
The CI will fail since we haven't updated the bindle version in CI pipeline, but I add the bindle health check first. I will change the bindle version in |
|
@FrankYang0529 thanks so much for your effort here. It's looking like Spin moving towards using OCI registries as its default form of artifact support. Full details can be seen in SIP-008: Using OCI Registries and a working (albeit in-progress) prototype can be seen here. It remains to be seen whether Spin deprecates support for using Bindle or not. However, I'm anticipating that we'll be sticking with Bindle 0.8.x for the forseeable future, at least while we gain experience and collect feedback around OCI support. Lastly, if we do indeed update Bindle at some point, we'll most likely wait until a non-rc release of the next version is available and hardened. With the notes above, I suggest we close this PR. However, I wanted to check in with you first before doing so. |
|
As this was suggested to be closed in January, and there's been no feedback to keep it open, I'm going to close it. If anyone wants it furthering then do feel free to reopen though! |
resolve #689
Setup
Test
spin bindle prepareTest
spin bindle pushTest integration with hippo