GitLab

Updating TH; not finished

Remove redundant param from allUses

Tiny code simplification (allUses)

Add info about compiling the Main module

Fix a bad consequence of the new story for the generic toT/fromT functions
derived from data types declarations. The problem was that they were being
generated and then discarded by the simplifier, because there was nothing
keeping them alive.

This commit
  * Adds a field tcg_keep to the TcGblEnv, which records things
    to be kept alive;

  * Makes the desugarer pin the keep-alive flag on each binding
    (it's actually a call to setIdLocalExported)

  * Removes that job from updateBinders in SimplCore


It's somewhat tiresome, but not really difficult.

Put to and from the correct way round

Print info about orphan rules and instances

Squish bogus warning

don't recognise (| or (# if followed by a symbol character, e.g. (||)

good for STABLE, I hope

Declaration splices should shadow the top-level environment

Add support for .bss (zero-initialized) sections on Mac OS X

Merge to stable.

Wibbles about argument variance

Fix to super-class loop avoidance code; commented in the file; tcrun020 tests

Improve duplicate-instance reporting; swap inst-env param order

Comments only

Always qualify in defaultErrStyle

Improve error message when iface decl not found

Eliminate bogus warning

Remove unused function dontAddErrLoc

Initialize hp_file for heap profiling (code stolen from Profiling.c).
This bug might suggest some general reviewing of this code-path...

Closes: SF bug [ 827485 ] Heap profile w/ debugging RTS dumps core
http://sourceforge.net/tracker/index.php?func=detail&aid=827485&group_id=8032&atid=108032

Fix bugs in previous commit.

When allocating a large object in gen 0, update the n_large_blocks
count.  I think this is just an accounting issue, and doesn't actually
cause a space leak, but it does result in an assertion failure when
running with sanity checking on.

Remove a comment that appears to contradict the code.

Set the mut_link of a MUT_ARR_PTRS_FROZEN to NULL.  This might be the
cause of the stage2 crash; I'll check as soon as the build completes.

Add missing -optF flag.

Fix markup

Another HsForAllTy wibble

Add missing table of -opt<blah> options.

Wibble for HsForAllTy

Do foralls right in the renamer

Fix a nasty bug in the GC mutable list handling, which shows up when
an array is frozen and then unsafeThaw#'d.  The array could end up on
the mutable list twice.

Fixes SourceForge bug #819116.

- In GHCi & Template Haskell, give a warning for things mentioned
  on the command line that we don't recognise, and then ignore them.
  In batch mode, these are still passed to the linker as before.

- Clean up & rename the filename-suffix recognising functions in
  DriverPhases. There's probably a better place for these, but I've left
  them where they are for now.

thread_stack(RET_BCO): Grab the pointer to the BCO *before* threading
it.  Fixes crashes when using compacting GC with GHCi.

I noticed while debugging this that compacting GC is horrendously slow
now, which might have something to do with the new eval/apply PAP
layouts.  That's something to investigate later.

mips64-sgi-irix6.5 works unregisterised

Wibbles

1. A tiresome change to HsType, to keep a record of whether or not
   the HsForAll was originally explicitly-quantified.  This is
   solely so that the type checker can print out messages that
   show the source code the programmer wrote.  Tiresome but
   easy.

2. Improve reporting of kind errors.

Show generic-ness when printing a data decl

Wibble to reporting duplicate instance decls

Report duplicate instance declarations

Save one fcntl-call in setNonBlockingFd when Fd is already non-blocking.

Please merge.