Upgrade go-ipld-cbor, boxo and friends

[rvagg]

See failures in #13196 coming out of go-state-types where we now have boxo@v0.32.0. It comes in from filecoin-project/go-state-types#394.

Unfortunately there's a bunch of breakage, so we either need to deal with that, or wind back go-state-types so we don't get caught when we really need an upgrade.

[rjan90]

Unfortunately there's a bunch of breakage, so we either need to deal with that, or wind back go-state-types so we don't get caught when we really need an upgrade.

Ugh, sorry for introducing this. Wish there was an easier way to see if version bumps causes breakages. I'd vote for reverting that PR bump for now. Seems cleaner to get the #13196 fix in first, then tackle the boxo@v0.32.0 bump breakages as a separate effort.

[eddieperez88]

May I ask what this is

…

On Fri, Jul 4, 2025 at 5:47 AM Phi-rjan ***@***.***> wrote: *rjan90* left a comment (filecoin-project/lotus#13197) <#13197 (comment)> Unfortunately there's a bunch of breakage, so we either need to deal with that, or wind back go-state-types so we don't get caught when we really need an upgrade. Ugh, sorry for introducing this. Wish there was an easier way to see if version bumps causes breakages. I'd vote for reverting that PR bump for now. Seems cleaner to get the #13196 <#13196> fix in first, then tackle the ***@***.*** bump breakages as a separate effort. — Reply to this email directly, view it on GitHub <#13197 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BLBXDFDJUPXJEBIASPASWPL3GZZUJAVCNFSM6AAAAACAZHRU2SVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTAMZWGEZTMMJTGI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[rvagg]

Seems cleaner to get the #13196 fix in first

Maybe, we'll have to assess the extent of the breakage. It's being forced on us in all directions anyway because boxo is an uber-package that impacts everything, so we have to deal with this at some point.

[rvagg]

#13202 has an upgrade, but it also has bitswap failures, so that's no good. We might have to wind back go-state-types until someone has the time and inclination to work this out because we can't introduce more bitswap problems than we already have.

[rvagg]

@gammazero can I ask you to parachute in here and help us diagnose this? As it stands we don't really have the resources to grok what's going on here and will default to reverting the cascaded upgrades coming out of boxo to make this work. #13202 has an upgrade for lotus. https://github.com/filecoin-project/lotus/actions/runs/16133716904/job/45525929661?pr=13202 is one of the two failures (same failure across two different machines). I'm having a hard time figuring out what warnings or errors are relevant in those logs when it's trying to sync but it seems to get to a point where it can't find peers to talk to and at some point it bails (blockstore closed errors suggesting a context cancellation from a failure somewhere). Maybe it's not bitswap related like it seems? If you have any suggestions related to us bumping these libraries that would be very helpful.

[aschmahmann]

@rvagg pretty sure the issues there are just that you've delegated the AI bots to upgrade across minor versions and they don't seem to be able to handle release notes. Maybe they'd be better with fewer releases at a time due to context window sizes 🤷‍♀️.

The thing quite literally passed an option to disable the bitswap server bitswap.WithServerEnabled(false) 🙃.