GitLab

test case is driver062.5

On FreeBSD FD_SETSIZE is unsigned. Cast it to a signed int
for portability.

addDLL returns const char*, not just a char*.
Fix compiler warning

`set -o igncr' does not work on non-cygwin-systems.
Fail silently if this command does not work, instead
of aborting the build.

MERGE TO STABLE

Lexing of the doc options pragma was changed, but but no change was 
made to the parser to reflect that. This patch fixes this problem.

MERGE TO STABLE

- This fixed the failures of GMapAssoc and GMapTop for optmising ways

MERGE TO STABLE

MERGE TO STABLE

MERGE TO STABLE

MERGE TO STABLE

  
  It turns out the newtype handling code in :print
  was slipping non mutable Tyvars in the types reconstructed.
  The error message eventually produced was rather obscure:
  
  [src/Tp.hs:75:28-64] *MainTp> :p x
  *** Exception: No match in record selector Var.tcTyVarDetails
  [src/Tp.hs:75:28-64] *MainTp>
  
  Due to non mutable tyvars, unifyType was failing.
  A well placed assertion in the unifyType code would have made
   my life much easier.
  Which reminds me I should install a -ddump-* system in the 
  RTTI subsystem, or future hackers will run away in swearing.


MERGE TO STABLE

  
  In Os X for instance, by default we have TERM=xterm-color 

MERGE TO STABLE

This actually didn't break anything, as the shell expanded $pkgid to the
empty string, but it was still wrong.

further attempt to fix #1391 on MacOS

Using -stubdir together with hierarchical modules, -fvia-C, and --make
is essentially broken in 6.6.x.  Recently discovered by Cabal's use of
-stubdir.

Test cases: driver027/driver028 (I've updated them to use -fvia-C, in
order to test for this bug).

Patch from Clemens Fruhwirth

It turns out I didn't get it quite right in the case of compacting
collection.  This should make it work again.

Document the rules for pragmas that must occur at the top of a file.

Please merge this patch

After a couple of abortive attempts, I think I've got this right.
When the GC sees a chain of the form 

   snd (_, snd (_, snd (_, ...)))

it can now deal with an arbitrary nesting depth, whereas previously it
had a depth limit which was necessitated by the use of recursion.  Now
we chain all the selector thunks together in the heap, and go back and
update them all when we find the value at the end of the chain.

While I was here I removed some old cruft in eval_thunk_selector()
which was carefully manintaing invariants that aren't necessary any
more, the main one being that evacuate() can safely be passed a
to-space pointer now.

In addition to validate I've tested building a stage3 compiler with
and without +RTS -c, so I'm reasonably sure this is safe.

Based on a patch from Lemmih

Resolving this bug is a bit tricky, it boils down to the question: Should the
Haddock links between packages include the package version or not?

Pro: We can differentiate between various versions of the same package,
installed all at once. (How often does this really happen in practice?)

Cons: When package A refers to a package B, and B is later upgraded, links
in A's documentation will break. Furthermore, if an *additional* version of
B is installed, which version should A refer to?

Because IMHO it is not clear what to do when version numbers are included,
let's leave them out. If somebody has a better idea, feel free to submit a
better patch.

MERGE TO STABLE

The previous hack to include Parser.hs in source distros broke the possibility
of doing a "make dist" in a fresh tree, i.e. one which has just been checked
out and configured, but *not* built. Of course you will need Happy for such a
source distro later, but at least the freedom to do this is important.

The ultimate goal should be that something like "make dist" will work in a
freshly checked out tree, with no prerequisite steps (this is very common in
most projects). We should move towards that goal, not away from it...


MERGE TO STABLE