Closed philderbeast closed 3 months ago
The lock check output when there's a difference.
diff --git a/stack.yaml.lock b/stack.yaml.lock
index fe80246..3856c80 100644
--- a/stack.yaml.lock
+++ b/stack.yaml.lock
@@ -5,12 +5,12 @@
packages:
- completed:
- hackage: rest-rewrite-0.4.2@sha256:4afd57aaad716827485240428c1deb5a60b81a3c069968f3b0f27a3cb9faf247,3929
+ hackage: rest-rewrite-0.4.3@sha256:915fb98b8c0a0f518c1a4b75bcf3ae27a3ebd5b10d60cc16a216c2fca2148ab0,3929
pantry-tree:
- sha256: a72c2400f09a609d891198b8df1360dd9b50d2fd9d3628a887b04b24636e6401
+ sha256: 4de44fbeb7bd655caeafe605406c0bc5fe1788d9cf37cdac2f18b018ee4cf5e6
size: 4075
original:
- hackage: rest-rewrite-0.4.2
+ hackage: rest-rewrite-0.4.3
- completed:
hackage: smtlib-backends-0.3@sha256:a947aead99f6a314833bddca9b502d5faea8d3bd2fc76ffb53c34d5c5b7557bc,1211
pantry-tree:
@@ -40,12 +40,19 @@ packages:
original:
hackage: store-0.7.18
- completed:
- hackage: store-core-0.4.4.6@sha256:37d4a628785b72dd26f0d8565895d01f36b4a7b2539100d735c239cf12d76b53,1489
+ hackage: store-core-0.4.4.7@sha256:a2ea427ff0dde30252474dcb0641cb6928cb8a93cd5ee27d4c22adba8e729683,1489
pantry-tree:
- sha256: da7e56dce68e291ea431b552a5cbc7cb0736296f9436e339931b193d48072e68
+ sha256: 67828df739695d14f81cd572a3085c76f65ac75f8417095afef8dfb2815f523e
size: 271
original:
- hackage: store-core-0.4.4.6
+ hackage: store-core-0.4.4.7
+- completed:
+ hackage: tasty-1.5@sha256:8da3f47fff790714f7d676692f1207aac156b41f705c55f14d1d8147a751264b,2787
+ pantry-tree:
+ sha256: d139b379075ff0204ad2aa2e20ba668534b21d03f8d712077e03e6d9294b2b8f
+ size: 1944
+ original:
+ hackage: tasty-1.5
snapshots:
- completed:
sha256: 876a5c75d90718add42e1ad36d66000bf35050ce1c66748119897a32df613186
Error: Process completed with exit code 1.
The resolver check when it fails:
I went ahead and checked the resolver and the lock file.
Actually, I'm not sure I understand the purpose of the lock file in CI.
If the lock file is only needed to generate an accurate cache key, it could be generated with --dry-run
before fetching the cache (similar to this setup). Is there a good reason to keep it version controlled?
The stack.yaml.lock
file is commited to source control to help humans. As for the lock file in CI, I've done a bit more reading of the Stack docs, on lock files and the --lock-file
option. You're right, I need to review the caching for this pull request, especially in light of https://github.com/haskell-actions/setup/issues/86.
hm, the jobs after merging don't find the cache, even though they use the same key used to upload it.
I have some recollection of caches being kept separate per branch. Probably rerunning the jobs in develop will find them.
Update: restarting finds the caches, but there's still no speed up. Much dependencies seem to be rebuilt anyway.
Minor glitch with #689 that slipped by me, the
stack.yaml.lock
file was out of date or rather I was testing a matrix ofGHC = {9.8.2, 9.6.5, 9.4.8}
but had forgot to update the resolver to match instack.yaml
that was forghc-9.8.1
.~@facundominguez do you think we need to check for this in CI?~ I went ahead and checked the resolver and the lock file.