-
Notifications
You must be signed in to change notification settings - Fork 36
Conversation
9f7a1b7
to
882f636
Compare
Note blockstore depends on raw-multihash-enabled datastore already, therefore this cannot be merged until that change is unlocked, or needs to be merged into a pre v1.0.0 branch. |
882f636
to
4b8b354
Compare
Pending ipfs/go-ipld-format#68 |
d6b4f4b
to
0d3fe7e
Compare
0d3fe7e
to
af197de
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
Note that this breaks code using |
@orpheuslummis yes, that why this code has been released with a new minor version (not patch). 🙂 If you see ipfs/kubo#7074 (comment) you can find most of the related PRs that updates various go-ipfs modules. |
Thanks for the reference to the PRs. :) |
We usually use Under This modules somehow, have a I'll raise this issue with the maintainer team, but my personal opinion is that we do a semver like where we do |
@orpheuslummis a really good point has been raised by someone else. Golang's way of importing V2 of packages require adding Having to bump all imports paths and not just go.mod files for such trivial change sounds like a useless suffering. |
I agree. I suggest considering switching to using v0.x.x versioning to avoid future confusions and to be semver-compatible. |
@orpheuslummis I don't like that idea, golang also plays very badly with that, every time people would Btw this package got a |
Ah yes. Then I suggest indicating in the README the versioning detail of the repo because it diverges from semver and other go-ipfs projects. |
@orpheuslummis thx for your report, this has been openned: |
could we in the short term put in an alias of
|
The new version takes a parameter (the cid that wasn't found) so an alias is insufficient. |
|
No because the main issue is that code that use
However that code is broken now because even if that the same error (what the checks want to do), your proposition would only match both identicaly errors and CIDs. (so an |
iirc there is intent here, in that people should switch to ipld.IsNotFound() checkers etc. |
Yeah, realized as posting that the blockstore approach doesn't work. I think the unfortunate thing i'm hitting is that there are blockstore interactions with |
I think there is a good argument to make about moving Then we can deprecate and add aliases in |
Rationale: ipfs/kubo#7074