build failures with `no entry found for key` panics

[scottlamb]

Crate names

• off-rs 0.1.3
• retina 0.3.7
• rp2040-pac 0.3.0
• num-prime 0.1.0
• holochain 0.0.124
• kube 0.68.0
• themelio-stf 0.9.7

Build failure links

• https://docs.rs/crate/retina/0.3.7/builds/505117
• https://docs.rs/crate/holochain/0.0.124/builds/505120

Additional details

Looks like there's been a recent regression in nightly rustdoc that affects more than just my crate. These look related:

• rustdoc panics when resolving intra doc link rust#93428
• rustdoc: Pre-calculate traits that are in scope for doc links rust#88679 (comment)
• rustdoc ICE: 'no entry found for key' rust#93476
• rustdoc ICE: 'no entry found for key' with static referencing external type rust#93698

Is it possible/worthwhile to pin docs.rs to an older nightly until this is fixed?

[Nemo157]

There is not a noticeably higher percentage of build failures over the last day or two, so it seems like it's not a very common failure. There's a pretty small fix PR already posted, so we can queue rebuilds for the crates impacted in a couple days when that is released to nightly.

[Nemo157]

(Speaking of, it'd be useful to have some scripting to grep recent build logs so we can find all impacted crates/versions).

[cmpute]

I guess I'm affected by the same problem: https://docs.rs/crate/num-prime/0.1.0/builds/501949

[kpreid]

The original issue was marked fixed, but there's still a panic with rustdoc 1.60.0-nightly (08df8b81d 2022-01-30) and some discussion of it on maybe-duplicate #93476

[Nemo157]

So it sounds like there are multiple sources of this ICE, just in case the specific issues these crates hit are fixed I've requeued builds for retina 0.3.7, off-rs 0.1.3, rp2040-pac 0.3.0 and num-prime 0.1.0. Then once 93476 is closed we can try again.

[Nemo157]

Only off-rs appears to have succeeded. I've put details for all the crates hitting this and the new rust issue in the OP.

[Nemo157]

I did a little grepping and grabbed all other builds that hit this issue in the last few days

all-is-cubes 0.4.0
all-is-cubes-content 0.4.0
all-is-cubes-gpu 0.4.0
arti-client 0.0.4
arti-config 0.0.4
asn1-codecs 0.1.3
asn1-rs 0.3.0
cargo-deny 0.11.1
cargo-raze 0.14.2
cdbc 0.1.11
deno_runtime 0.43.0
deno_webstorage 0.30.0
der-parser 6.0.1
hls 0.0.9
holochain 0.0.124
holochain_cascade 0.0.24
holochain_cli 0.0.25
holochain_cli_sandbox 0.0.21
holochain_conductor_api 0.0.24
holochain_p2p 0.0.24
holochain_state 0.0.24
holochain_wasm_test_utils 0.0.24
httpflv 0.0.6
jtag 0.0.2
melwallet-client 0.4.1
menmos 0.2.2
menmos-client 0.1.0
menmos-interface 0.1.0
menmos-protocol 0.1.0
miio-proto 0.1.4
oberon 1.1.1
rtmp 0.0.12
sqlite2parquet 0.3.0
themelio-nodeprot 0.10.10
tor-dirmgr 0.0.4
xiu 0.0.10
zephyrus 0.2.4

[thejpster]

Hopefully the upstream bug is fixed now?

[Nemo157]

Requeued the crates that have been mentioned (retina, rp2040-pac, num-prime, holochain, kube). If they all succeed then we might requeue all the other crates that have hit this issue too.

[Nemo157]

rp2040-pac, num-prime and kube appear to have all worked; retina and holochain failed again and I've added their new build logs to the OP. I'll open a new upstream issue and wait on requeuing unasked for crates (if you run into/have published a crate that hit this issue, feel free to ask for a rebuild).

[syphar]

@Nemo157 there is a new failure reported (by @TheDarkula )
https://docs.rs/crate/themelio-stf/0.9.7/builds/502388

is it worth requeuing?

[Nemo157]

Yep, that build was before the last fix so might succeed, I've requeued it.

[TheDarkula]

@syphar @Nemo157 Thank you!
Is there somewhere I can view the queue?