Multiple cyclic dependencies

[0x7CFE]

Recent versions lead to ra_lsp_server consuming 100% of a single core when trying to parse Substrate repository.

RA version:

commit 4576e6597e0778ecb3f204ff781011610fba3631
Merge: 419eec3d 7a9d7dd6
Author: bors[bot] <26634292+bors[bot]@users.noreply.github.com>
Date:   Thu Sep 5 20:12:19 2019 +0000

Console output:

�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-executor -> substrate-client�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-offchain -> substrate-client�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-offchain -> substrate-transaction-pool�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-client-db -> substrate-client�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-client-db -> substrate-executor�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-offchain-primitives -> substrate-client�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-network -> substrate-client�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-network -> substrate-consensus-babe-primitives�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-test-client -> substrate-client�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-test-client -> substrate-executor�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-application-crypto -> sr-io�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-test-runtime-client -> substrate-test-client�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-test-runtime -> sr-io�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-test-runtime -> srml-babe�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-test-runtime -> srml-executive�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-test-runtime -> srml-system�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-test-runtime -> srml-timestamp�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-test-runtime -> substrate-application-crypto�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-test-runtime -> substrate-client�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-test-runtime -> substrate-consensus-babe-primitives�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-test-runtime -> substrate-executor�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-test-runtime -> substrate-session�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-test-runtime -> substrate-test-runtime-client�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-consensus-aura-primitives -> substrate-application-crypto�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency substrate-consensus-aura-primitives -> substrate-client�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency srml-support -> sr-io�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency srml-support -> srml-system�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency sr-primitives -> sr-io�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency sr-primitives -> substrate-application-crypto�[0m
�[1;38;5;196mERROR�[0m [ra_project_model] �[1;38;5;196mcyclic dependency node-testing -> node-executor�[0m

[matklad]

THe 100% CPU is probably the same as #1684 : 2^n code path inside the chalk. The cyclic dependency error seems unrelated but weird.

[lnicola]

The cyclic dep errors still happen, but that's probably #2414. We still use a lot of CPU (in syntax_bridge?), but I don't see any slowness.

[Veykril]

Dupe of #2414