Resolve conflict with Simon's lovely addUsedGRE changes
Oct 30 2015
Rebase on latest HEAD (with record pattern synonyms), separate from OverloadedLabels
Crumbs. The code is substantially LONGER than before. Why?
Still, it is at least modularly separated.
Oct 29 2015
Refactoring and minor bug fixes
- Handle record selectors properly in Template Haskell
- Refactor disambiguateRecordBinds and make errors more uniform
- Handle non-fields appearing in updates
- More refactoring and fixing of comments
- Refactor to make lint happy
Oct 28 2015
Oct 22 2015
- Tweak lexer to accept hash at the beginning of a line
- Add OverloadedLabels to GHC-only extensions in T4437
Oct 19 2015
Thanks, will do.
Oct 16 2015
Bring up to date with master
The bit I am worried about is the record selector generation. I don't know how this is implemented but if there class instances generated then there seems to be the possibility of generating orphan instances.
I am very scared of the rebase which is going to have to happen after D761 gets merged.
This builds fine on Travis but not on Phab, I suspect because the diff needs to be updated to apply cleanly to the latest HEAD. Is it okay for me to go ahead and merge this? I'll resolve whatever conflicts there are and do a final validate before merging.
Changes following Simon's review
Oct 15 2015
Oct 12 2015
Jul 24 2015
Jul 23 2015
Jul 20 2015
I've made the change to DuplicateRecordFields.
Jul 16 2015
Squash changes into one commit on top of master
Jul 15 2015
Hmm, I don't understand why the build is failing. Clearly addUsedSelector is used in RnEnv by lookupSubBndrOcc, which is itself exported. Have I done something silly in updating the diff?
Jul 14 2015
- Merge remote-tracking branch 'origin/master' into wip/orf-reboot
- Get rid of extendRecordFieldEnv, do it in getLocalNonValBinders instead
- Introduce FieldOcc for occurrences of fields in HsExpr
- Use FieldOcc for ConDeclField names; hsLTyClDeclBinders et al become a bit nicer
- Simplify hsLTyClDeclBinders still further
- Use first datacon name instead of tycon name when mangling selector names
- Get rid of dfid_rep_tycon and the whole assignInstDeclNames malarkey
- Fix silly bug in ifaceConDeclFields for empty data types
- Big refactoring to distinguish HsRecField from HsRecUpdField
- Do not store (lexical) parents in hsRecUpdFieldSel
- Extra tests for odd corner cases
- Merge branch 'wip/orf-reboot' of git://git.haskell.org/ghc into wip/orf-reboot
- Use a map when renaming ConDeclFields
- Do a global lookup when renaming fields in updates
- Return FieldOccs from lookupOccRn_overloaded
- Store just selector Name in FieldOcc, not FieldLbl
- Make parser report use of .. in record updates
- Fix unused imports/declarations/constraints and missing deriving Typeable
- Use Maybe Name instead of Parent where that's what we mean
- Update haddock submodule
- Get rid of OverloadedRecordFields for now; add ADRF to T4437
- Tweak test output for T7145b