Repository navigation
Conversation
6d57dc6 to
219ee89
Compare
|
This MR attempts to regain sharing after-the-fact during serialisation. I think that's papering over the issue in an incomplete and non-robust way. As far as I understand, the regression was due to the addition of A similar win in this area is to stop |
219ee89 to
ab7be0d
Compare
Every unit carried its package's GenericPackageDescription, so the plan caches grew with the square of a package's component count. The in-library configure now reads it from the package's .cabal file, as a Setup script does. Fixes haskell#12463. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ab7be0d to
a42643c
Compare
|
done: dropped |
|
Yes, much better approach, thanks. We will need to think through what it means for error messages for users, e.g. ensure that it doesn't cause duplicate error messages. |
Fixes #12463.
drops
elabGPkgDescription; the in-library configure reads the.cabalfile instead, asSetupdoes. unpacking already writes revisions there.no-op
cabal build all, one package with N public sublibraries; master 538d465, ghc 9.14.1, medians of three:improved-plansignificance: significantin the changelog file.QA Notes
build a package with internal libraries, and one from Hackage with a revision; both configure as before.