Conversation
|
I've done basic dumb work to update the code and CI for 9.8 but there are some actual failures that need to be looked into. Help appreciated. |
ad35664 to
a187d98
Compare
|
I fixed the actual regressions (due to -Wx-partial) locally and now I'm wrangling with allow-newer. |
0d8adfb to
2a9814e
Compare
2a9814e to
c27ebb6
Compare
c27ebb6 to
ee45982
Compare
|
Timed out while fixing all those heads and tails. Converting to draft for now. |
2528de3 to
d13ae95
Compare
|
All right, I issue an official call for help: both in Matrix and here. The failing test is here: https://github.com/haskell/cabal/actions/runs/6536363720/job/17747928752?pr=9330 It's about a hash mismatch in cabal/Cabal-tests/tests/UnitTests/Distribution/Utils/Structured.hs Lines 28 to 33 in 9f37325 I had a cursory look over what this |
d13ae95 to
3ed2fdb
Compare
|
Gershom greenlights adding some CPP to handle this apparent change in how GHC 9.8 codegens data types. |
3ed2fdb to
217d76e
Compare
|
More of the same |
|
Wow, and this is one of those times when cache reset does not help (or the dice rolled wrong again). |
a3354e0 to
764ed85
Compare
|
It was a long ride… I'll double check that commit history makes sense (I edited it several times) and then apply the merge label. |
b17e6e0 to
fc82767
Compare
c0d182e to
e026e17
Compare
e026e17 to
c155656
Compare
…tured It seems, GHC 9.8 changed something in the code generation for data types. Structured class is supposed to catch such cases.
And bump Cabal's "supported version" of GHC
|
Took only two months… 🥲 |
Template Β: This PR does not modify
cabalbehaviour (documentation, tests, refactoring, etc.)Include the following checklist in your PR: