GitLab

updated

Verify that 'as' clause without 'qualified' brings both qual and unqual'ed name into scope

Test which verifies that special ids are correctly handled by the front-end

FFI decl tests

If -noC is used, don't compile the foreign-export C stubs

* On foreign decls, "ext_name"s are now optional. If missing, the ext_name
  is made equal to the Haskell name.
* Half a dozen special-ids were incorrectly handled when occurring as tyvars.

Fix default Enum methods

Fix preInlineUnconditionally and postInlineUnconditionally, so they respect the black list

Small changes to concurrency documentation

Adjust library imports to get fromInt/toInt from PrelNum

Num( fromInt ) and Integral( toInt ) are not Haskell 98, and so they
should be exposed only by GlaExts, not by Prelude.

Fix typo

synch output.

typo

update sample output

missing stdout file.

syslib misc now depends on syslib concurrent because of Memo.hs.

When deciding if a module is a "library" module, allow
	C:/
	C:\
as prefixes of the path, not just
	/

We need to improve this hack, but this fix makes things
work on Windoze.

wibble

Fix imports.

oops, export threadWaitRead etc.

Move the rest of the Concurrent Haskell documentation into the
GHC/Hugs libraries document, so it now all lives in one place.  Also
update and tidy it up somewhat.

threadDelay etc. should be defined in PrelConc.

Added missing #include.

Fixed typo UNUSED => STG_UNUSED

enable non-blocking I/O.

Add select code for thread{WaitRead,WaitWrite,Delay}.

Support for thread{WaitRead,WaitWrite,Delay}.  These should behave
identically to the 3.02 implementations.

We now have the virtual timer on during all program runs, which ticks
at 50Hz by default.  This is used to implement threadDelay, so you
won't get any better granularity than the tick frequency
unfortunately.  It remains to be seen whether using the virtual timer
will have a measurable impact on performance for non-threadDelaying
programs.

All operations in the I/O subsystem should now be non-blocking with
respect to other running Haskell threads.  It remains to be seen
whether this will have a measurable performance impact on
non-concurrent programs (probably not).

Little test for Eq and Ord instances of ThreadId.

Eq and Ord instances for ThreadId.

Add -D__HASKELL__=98

IdInfo.copyIdInfo was wrong.  It didn't copy a NOINLINE pragma
when the occurrenc analyser "shorted out" a binding; as a result
the following program didn't respect the NOINLINE pragma:

	plus :: Int -> Int -> Int
	{-# NOINLINE plus #-}
	plus = (+)

	f x = plus x (plus x x)

oops, back out some other changes that made it in by mistake.

An inaccurate constraint on the inline assembly code for mulIntC#
caused gcc 2.95 to overoptimise around it, leading to bogus multiply
code for Integers.

Document the -i effect on recompilation

Record all orphan modules (including "library" ones)
in the usage info in an interface file.  Leaving these out
prevents propagation of orphan module information.

Make primOpHasSideEffects return True for the various Write primops.
Amazingly, this wasn't done, and that meant that Write operations
could be silently discarded!  This broke the GHC typechecker
(actually the implementation of checkNoErrsTc, showed up in
deriving code)

When recompiling with GHC 4.02 and Keith's new type representations,
it seems that I need UVar to be in Var.hi-boot.  I don't think this
will be the case with the better handling of SOURCE import in
4.04, but for now I'm just going ahead and adding the UVar synonym
to the Var.hi-boot files.

-fno-cpr ==> -fno-cpr-analyse

back out rev. 1.115 - -fno-cpr-analyse works fine.