1. 13 Jun, 2005 9 commits
    • simonmar's avatar
      [project @ 2005-06-13 14:12:59 by simonmar] · f96c7c1a
      simonmar authored
      Implement :tags command
      
      Patch supplied by Claus Reinke, with some modifications by me.
      Ideally we'd like this to be a command line option too, and we'd like
      to drop the restriction that all the source files must be interpreted,
      but that needs some work elsewhere (interface files have to store
      definition source locations).
      f96c7c1a
    • simonmar's avatar
      [project @ 2005-06-13 13:46:39 by simonmar] · 7004f764
      simonmar authored
      unused imports
      7004f764
    • simonmar's avatar
      [project @ 2005-06-13 13:44:48 by simonmar] · 58b4a155
      simonmar authored
      - Eliminate some warnings, remove dead code
      - export PrintUnqualified, alwaysQualify
      58b4a155
    • simonmar's avatar
      [project @ 2005-06-13 12:29:48 by simonmar] · b07f3876
      simonmar authored
      Block allocator performance fix: instead of keeping the free list
      ordered, keep it doubly-linked, and introduce a new flag BF_FREE so we
      can tell when a block is free.  We can still coalesce blocks on the
      free list because block descriptors are kept consecutively in memory,
      so we can tell based on the BF_FREE flag whether to coalesce with the
      next higher/lower blocks when freeing a block.
      
      This (almost) make freeChain O(n) rather than O(n^2), and has been
      reported to help a lot when dealing with very large heaps.
      b07f3876
    • simonmar's avatar
      [project @ 2005-06-13 10:21:09 by simonmar] · 15e00848
      simonmar authored
      Use perl rather than sed to munge the package.conf file; avoids
      portability problems with sed (previous version didn't work on Solaris).
      15e00848
    • wolfgang's avatar
      [project @ 2005-06-13 02:44:22 by wolfgang] · c26a153c
      wolfgang authored
      Darwin:
      Deal with the fact that the code and data in a .o file might not be
      16-byte aligned even if it contains instructions that require 16-byte
      alignment.
      This fixes floating point related crashes on Darwin/x86; there might
      also have been problems on Darwin/ppc if people load .o files with
      AltiVec code into GHCi.
      c26a153c
    • wolfgang's avatar
      [project @ 2005-06-13 02:03:42 by wolfgang] · 4302b12e
      wolfgang authored
      Darwin/x86:
      Instead of aligning all literals at 16 bytes, reuse the hack from x86-64.
      4302b12e
    • wolfgang's avatar
      [project @ 2005-06-13 00:16:44 by wolfgang] · cd22d3c7
      wolfgang authored
      Darwin/x86: support GHCi.
      cd22d3c7
    • wolfgang's avatar
      [project @ 2005-06-13 00:15:58 by wolfgang] · 59177a1b
      wolfgang authored
      Darwin/x86:
      Fix two causes for crashes:
      *) align literals to 16 bytes, because Apple's GCC likes to use
         SSE instructions which require this alignment.
      *) In print_doctored, recognize global jumps that masquerade as
         local jumps by going via Lfoo$stub or Lfoo$non_lazy_ptr.
      59177a1b
  2. 10 Jun, 2005 2 commits
  3. 09 Jun, 2005 4 commits
  4. 08 Jun, 2005 1 commit
  5. 07 Jun, 2005 5 commits
  6. 06 Jun, 2005 3 commits
  7. 03 Jun, 2005 4 commits
  8. 02 Jun, 2005 2 commits
  9. 01 Jun, 2005 1 commit
  10. 31 May, 2005 4 commits
    • simonmar's avatar
      [project @ 2005-05-31 14:14:26 by simonmar] · ae658c18
      simonmar authored
      Make more error messages from the downsweep into ErrMsg exceptions.
      ae658c18
    • simonmar's avatar
      [project @ 2005-05-31 13:10:39 by simonmar] · 31f289ba
      simonmar authored
      oops, fix updating the module graph
      31f289ba
    • simonmar's avatar
      [project @ 2005-05-31 12:45:03 by simonmar] · aa6eb36c
      simonmar authored
      Fix some reporting of errors in the GHC API: errors during the
      downsweep were thrown as exceptions; now they're reported via the
      (Messages->IO ()) callback in the same way as other errors.
      
      getModuleInfo no longer prints anything on stdout.  It does ignore
      error messages and return Nothing, however - we should fix this and
      return the error messages at some point.
      
      The ErrMsg type can now be thrown as an exception.  This can be a
      convenient alternative if collecting multiple error messages isn't
      required.  We do this in the downsweep now.
      aa6eb36c
    • simonmar's avatar
      [project @ 2005-05-31 09:11:27 by simonmar] · 004ed82c
      simonmar authored
      Fix the names of a few fields.
      004ed82c
  11. 27 May, 2005 1 commit
  12. 26 May, 2005 1 commit
    • simonpj's avatar
      [project @ 2005-05-26 21:37:13 by simonpj] · 6a18febc
      simonpj authored
      MERGE TO STABLE
      
      Put back in a missing case for higher-rank types. When the
      definition is
      	a) non-recursive
      	b) a function binding
      	c) lacks a type signature
      we want to *infer* a perhaps-higher-rank type for the RHS,
      before making a monomorphically-typed Id for the LHS.
      
      E.g. 	f = \(x :: forall a. a->a) -> (x True, x 'c')
      
      This case got lost in the transition to 6.4
      
      tc194 tests it
      6a18febc
  13. 25 May, 2005 3 commits