GitLab

Change expectations a bit for the non-i386-world.
cc004 and ffi012 use stdcall and is therefore expected to fail on non-i386.
ffi009, arith008 and arith012 expected to fail in optasm because of an
i386-specific floating-point accuracy problem. It now only expects failure on
i386.

MERGE TO STABLE

Test recursive dictionaries

Add test for indirection-zapping

accept output

Add another fundep test

Add a functional-dependency test

Accept output

bogon on test

accept output

Add a desugaring test involving existentials

Add test for kind errors in record update; and synonyms in iface files

add type tidying test

accept output

More tests

Add infix type contructor test

Add a functional-dep test

Add data2tag test

- update for changes to InstalledPackageInfo syntax
- new test: ghcpkg02: for every package we know about, pipes the
  output of ghc-pkg describe into ghc-pkg update, to test
  pretty printing/parsing of InstalledPackageInfo

add do parsing test

Update comment, this is nothing to do with requiring an expression as
the last statement in a do.

Add further commentary

rename module name too

Rename Rn017.hs to avoid clash on non-case-sensitive file systems

Bump the stack size for conc019

read027 is an expected failure

computers have got a little fast for this test

Add some tests for <pkg>-*

accept output

Add a basic ghc -M test

accept output

add a test for recompilation bug in GHCi

Fix up for recent syntax changes, and add some dependencies for testing

Disable the use of the timeout wrapper on Windows for now; it runs the
command using the wrong shell.

Add a module-import warning test

Flush output in runIO

Accept output

slight correction to runCmd to fix exit code values.

more tests

rn011 should be a multimod test too

Add a timeout to test runs, using a wrapper program (written in
Haskell, using System.Process of course!).