simonpj@microsoft.com [Fri, 12 Oct 2007 16:23:25 +0000 (16:23 +0000)]
Fix DoCon: Another try at getting extractResults right
For some reason TcSimplify.extractResults is quite difficult to get right.
This is another attempt; finally I think I have it.
Strangely enough, it's only Sergey's DoCon program that shows up the
bug, which manifested as a failure in the Simplifier
lookupRecBndr $dGCDRing{v a1Lz} [lid]
But it was due to extractResults producing multiple bindings for
the same dictionary.
Please merge this to the stable branch (after previous patches to
TcSimplify though).
Simon Marlow [Thu, 11 Oct 2007 13:57:36 +0000 (13:57 +0000)]
mention what SCC stands for
Simon Marlow [Thu, 11 Oct 2007 13:55:05 +0000 (13:55 +0000)]
Add a proper write barrier for MVars
Previously MVars were always on the mutable list of the old
generation, which meant every MVar was visited during every minor GC.
With lots of MVars hanging around, this gets expensive. We addressed
this problem for MUT_VARs (aka IORefs) a while ago, the solution is to
use a traditional GC write-barrier when the object is modified. This
patch does the same thing for MVars.
TVars are still done the old way, they could probably benefit from the
same treatment too.
Simon Marlow [Wed, 10 Oct 2007 15:32:44 +0000 (15:32 +0000)]
we need to #include "Stg.h" first, we can't rely on GHC to inject it
This fixes the unreg build, and in general building the RTS code
via-C. I'm not sure at what stage this was broken, but I think it
was working accidentally before.
simonpj@microsoft.com [Thu, 11 Oct 2007 12:34:26 +0000 (12:34 +0000)]
Fix Trac #1680; check for unboxed tuples in TcType.marshalableTyCon
simonpj@microsoft.com [Wed, 10 Oct 2007 16:47:31 +0000 (16:47 +0000)]
Fix Trac #1759: do not let ticks get in the way of spotting trivially-true guards
GHC spots that an 'otherwise' guard is true, and uses that knowledge to
avoid reporting spurious missing-pattern or overlaps with -Wall.
The HPC ticks were disguising the 'otherwise', which led to this failure.
Now we check. The key change is defining DsGRHSs.isTrueLHsExpr.
Test is ds062
simonpj@microsoft.com [Wed, 10 Oct 2007 15:02:50 +0000 (15:02 +0000)]
Fix Trac #1755; check for stage errors in TH quoted Names
There are a number of situations in which you aren't allowed to use
a quoted Name in a TH program, such as
\x -> 'x
But we weren't checking for that! Now we are.
Merge to stable branch.
Test is TH_qname.
simonpj@microsoft.com [Wed, 10 Oct 2007 12:40:13 +0000 (12:40 +0000)]
checkWellStaged: reverse comparsion (no change in semantics), plus some comments
simonpj@microsoft.com [Wed, 10 Oct 2007 12:39:33 +0000 (12:39 +0000)]
Add traceTc in tcSimplifyDefault
simonpj@microsoft.com [Wed, 10 Oct 2007 12:37:26 +0000 (12:37 +0000)]
Improve pretty-printing of splices in HsSyn
simonpj@microsoft.com [Wed, 10 Oct 2007 14:57:05 +0000 (14:57 +0000)]
Fix Trac #1678; be more careful about catching and reporting exceptions in spliced TH monadic computations
Many of the new lines are comments to explain the slightly-convoluted
in which exceptions get propagated out of the Q monad.
This fixes Trac 1679; test is TH_runIO (as well as the exising TH_fail).
Please merge
simonpj@microsoft.com [Wed, 10 Oct 2007 14:56:46 +0000 (14:56 +0000)]
Comments only
Simon Marlow [Wed, 10 Oct 2007 10:37:04 +0000 (10:37 +0000)]
FIX BUILD (when compiling base via C): declare n_capabilities
Simon Marlow [Wed, 10 Oct 2007 09:32:41 +0000 (09:32 +0000)]
GHCi: use non-updatable thunks for breakpoints
The extra safe points introduced for breakpoints were previously
compiled as normal updatable thunks, but they are guaranteed
single-entry, so we can use non-updatable thunks here. This restores
the tail-call property where it was lost in some cases (although stack
squeezing probably often recovered it), and should improve
performance.
Simon Marlow [Wed, 10 Oct 2007 08:58:20 +0000 (08:58 +0000)]
FIX #1681: withBreakAction had too large a scope in runStmt
Simon Marlow [Tue, 9 Oct 2007 14:50:02 +0000 (14:50 +0000)]
tiny refactoring
Simon Marlow [Tue, 9 Oct 2007 14:53:05 +0000 (14:53 +0000)]
small reworking of the loop-breaker-choosing algorithm
Previously inline candidates were given higher preference as
non-loop-breakers than constructor applications, but the reason for
this was that making a wrapper into a loop-breaker is to be avoided at
all costs. This patch refines the algorithm slightly so that wrappers
are explicitly avoided by giving them a much higher score, and other
inline candidates are given lower scores than constructor
applications.
This makes almost zero difference to a complete nofib run, so it
amounts to just a tidyup.
Manuel M T Chakravarty [Wed, 10 Oct 2007 10:18:40 +0000 (10:18 +0000)]
Fix warnings when build w/o readline
Clemens Fruhwirth [Wed, 10 Oct 2007 07:44:15 +0000 (07:44 +0000)]
Update documentation for win32 DLL linking
simonpj@microsoft.com [Wed, 10 Oct 2007 09:33:34 +0000 (09:33 +0000)]
FIX: tidy up TcSimplify following equality constraints additions
The combination of "type refinement" for GADTs and the new equality
constraints has made TcSimplify rather complicated. And wrong:
it generated bogus code for cholewo-eval.
This patch is still far from entirely satisfactory. There are
too many places where invariants are unclear, and the code is
still a bit of a mess. But I believe it's better, and it passes
the regression tests! So I think it's good enough for the 6.8 release.
Please merge.
The main changes are:
- get rid of extractLocalResults (which was always suspicious)
- instead, treat the 'refinement' along with 'givens', by
adding a field to RedEnv, red_reft which travels with red_givens
- I also reworked extractResults a bit, which looked wrong to me
This entailed changing the Given constructor in Avail to take
an Inst rather than a TcId
simonpj@microsoft.com [Wed, 10 Oct 2007 09:30:58 +0000 (09:30 +0000)]
Improve pretty-printing for HsSyn
simonpj@microsoft.com [Sat, 29 Sep 2007 10:44:06 +0000 (10:44 +0000)]
Fix Trac #1746: make rule-matching work properly with Cast expressions
The Cast case of the rule-matcher was simply wrong.
This patch fixes it; see Trac #1746.
I also fixed the rule generation in SpecConstr to generate a wild-card
for the cast expression, which we don't want to match on. This makes
the rule more widely applicable; it wasn't the cause of the bug.
simonpj@microsoft.com [Sat, 29 Sep 2007 10:43:09 +0000 (10:43 +0000)]
Small comment only
Simon Marlow [Tue, 9 Oct 2007 14:27:01 +0000 (14:27 +0000)]
export n_capabilities, see #1733
Simon Marlow [Tue, 9 Oct 2007 14:25:54 +0000 (14:25 +0000)]
FIX #1743, create a fresh unique for each Id we bind at a breakpoint
Simon Marlow [Tue, 9 Oct 2007 13:09:42 +0000 (13:09 +0000)]
remove vestiges of way 'u' (see #1008)
Simon Marlow [Tue, 9 Oct 2007 12:24:09 +0000 (12:24 +0000)]
also call initMutex on every task->lock, see #1391
Simon Marlow [Tue, 9 Oct 2007 12:23:38 +0000 (12:23 +0000)]
remove the "-unreg" flag and the unregisterised way, see #1008
Simon Marlow [Tue, 9 Oct 2007 10:51:38 +0000 (10:51 +0000)]
warning removal
Simon Marlow [Wed, 3 Oct 2007 17:00:05 +0000 (17:00 +0000)]
warning removal
Simon Marlow [Wed, 3 Oct 2007 16:35:36 +0000 (16:35 +0000)]
refactoring only: use the parameterised InstalledPackageInfo
This required moving PackageId from PackageConfig to Module
Simon Marlow [Wed, 3 Oct 2007 17:40:16 +0000 (17:40 +0000)]
warning removal
Simon Marlow [Wed, 3 Oct 2007 17:34:48 +0000 (17:34 +0000)]
warning removal
Simon Marlow [Wed, 3 Oct 2007 17:32:02 +0000 (17:32 +0000)]
warning removal
Simon Marlow [Wed, 3 Oct 2007 17:27:15 +0000 (17:27 +0000)]
warning removal
Simon Marlow [Wed, 3 Oct 2007 09:08:04 +0000 (09:08 +0000)]
remove most warnings
Simon Marlow [Tue, 2 Oct 2007 11:49:17 +0000 (11:49 +0000)]
mkIfaceExports: sort the children of AvailTC
This fixes a problem with spurious recompilations: each time a module
was recompiled, the order of the children would change, causing extra
recompilation.
MERGE TO STABLE
Simon Marlow [Mon, 8 Oct 2007 13:49:58 +0000 (13:49 +0000)]
error message fix (#1758)
Thorkil Naur [Fri, 5 Oct 2007 14:41:05 +0000 (14:41 +0000)]
FIX validate for PPC Mac OS X - RegAllocStats.hs
Thorkil Naur [Fri, 5 Oct 2007 14:36:07 +0000 (14:36 +0000)]
FIX validate for PPC Mac OS X - RegAllocLinear.hs
Thorkil Naur [Fri, 5 Oct 2007 14:49:08 +0000 (14:49 +0000)]
FIX validate for PPC Mac OS X - Linker.c
Thorkil Naur [Fri, 5 Oct 2007 14:44:54 +0000 (14:44 +0000)]
FIX validate for PPC Mac OS X - Evac.h
Simon Marlow [Mon, 8 Oct 2007 13:13:05 +0000 (13:13 +0000)]
FIX #1748: -main-is wasn't handling the case of a single hierarchical module
test case is driver062.5
jochemberndsen@dse.nl [Thu, 27 Sep 2007 13:26:49 +0000 (13:26 +0000)]
FIX BUILD FD_SETSIZE signed
On FreeBSD FD_SETSIZE is unsigned. Cast it to a signed int
for portability.
jochemberndsen@dse.nl [Thu, 27 Sep 2007 13:26:19 +0000 (13:26 +0000)]
FIX BUILD addDLL returns const char*
addDLL returns const char*, not just a char*.
Fix compiler warning
jochemberndsen@dse.nl [Wed, 26 Sep 2007 20:37:50 +0000 (20:37 +0000)]
FIX BUILD `set -o igncr'-issue on FreeBSD
`set -o igncr' does not work on non-cygwin-systems.
Fail silently if this command does not work, instead
of aborting the build.
Simon Marlow [Mon, 8 Oct 2007 11:57:40 +0000 (11:57 +0000)]
comment-out "use vars" in 3 places (see #1739)
David Waern [Tue, 2 Oct 2007 14:38:49 +0000 (14:38 +0000)]
Change DOCOPTIONS pragma to DOC_OPTIONS
MERGE TO STABLE
David Waern [Tue, 2 Oct 2007 14:37:13 +0000 (14:37 +0000)]
FIX: parsing of doc options
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
David Waern [Tue, 2 Oct 2007 14:34:59 +0000 (14:34 +0000)]
FIX: add missing case to OccName.isSymOcc
Manuel M T Chakravarty [Tue, 2 Oct 2007 13:07:36 +0000 (13:07 +0000)]
Remove warnings from WwLib
Manuel M T Chakravarty [Tue, 2 Oct 2007 13:04:07 +0000 (13:04 +0000)]
FIX: mkWWcpr takes open alg types into account
- This fixed the failures of GMapAssoc and GMapTop for optmising ways
MERGE TO STABLE
Manuel M T Chakravarty [Mon, 1 Oct 2007 15:43:43 +0000 (15:43 +0000)]
FIX #1738: KPush rule of FC must take dataConEqTheta into account
MERGE TO STABLE
Manuel M T Chakravarty [Sat, 29 Sep 2007 12:26:24 +0000 (12:26 +0000)]
FIX #1729: Don't try to expand syn families with -XLiberalTypeSynonyms
MERGE TO STABLE
Manuel M T Chakravarty [Sat, 29 Sep 2007 12:19:41 +0000 (12:19 +0000)]
Some more traceTcs
Manuel M T Chakravarty [Fri, 28 Sep 2007 22:55:41 +0000 (22:55 +0000)]
FIX: Make boxy splitters aware of type families
MERGE TO STABLE
Pepe Iborra [Thu, 27 Sep 2007 15:13:00 +0000 (15:13 +0000)]
Finally, I managed to squash an infamous bug in :print
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
Pepe Iborra [Thu, 27 Sep 2007 14:15:49 +0000 (14:15 +0000)]
Be a bit more flexible in terminal identification for do_bold
In Os X for instance, by default we have TERM=xterm-color
MERGE TO STABLE
Ian Lynagh [Thu, 27 Sep 2007 13:04:27 +0000 (13:04 +0000)]
html_installed_root shouldn't contain $$pkgid
This actually didn't break anything, as the shell expanded $pkgid to the
empty string, but it was still wrong.
simonpj@microsoft.com [Thu, 27 Sep 2007 11:08:42 +0000 (11:08 +0000)]
Comments and debug output only
Simon Marlow [Thu, 27 Sep 2007 10:25:39 +0000 (10:25 +0000)]
further stub filename fix: I managed to break non-stubdir -fvia-C compilation
Simon Marlow [Thu, 27 Sep 2007 09:13:31 +0000 (09:13 +0000)]
also acquire/release task->lock across fork()
further attempt to fix #1391 on MacOS
Simon Marlow [Wed, 26 Sep 2007 13:45:39 +0000 (13:45 +0000)]
FIX -stubdir bug: the .hc file was #including the wrong _stub.h filename
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).
Ian Lynagh [Wed, 26 Sep 2007 12:26:37 +0000 (12:26 +0000)]
Add STANDARD_OPTS to SRC_HC_OPTS in rts/Makefile so we get -I../includes for .cmm files
Patch from Clemens Fruhwirth
Simon Marlow [Wed, 26 Sep 2007 10:07:32 +0000 (10:07 +0000)]
fix #1734, panic in :show modules after load failure
Clemens Fruhwirth [Wed, 26 Sep 2007 08:48:02 +0000 (08:48 +0000)]
Remove current package from preloaded package set
andy@galois.com [Wed, 26 Sep 2007 05:53:31 +0000 (05:53 +0000)]
Fixing #1340, adding HPC Documentation
Simon Marlow [Tue, 25 Sep 2007 13:26:00 +0000 (13:26 +0000)]
fix to previous fix to THUNK_SELECTOR machinery
It turns out I didn't get it quite right in the case of compacting
collection. This should make it work again.
simonpj@microsoft.com [Mon, 24 Sep 2007 16:28:47 +0000 (16:28 +0000)]
Be more specific about file-header pragmas
Document the rules for pragmas that must occur at the top of a file.
Please merge this patch
Simon Marlow [Mon, 24 Sep 2007 10:33:23 +0000 (10:33 +0000)]
comments only: point to relevant bug reports
Simon Marlow [Mon, 17 Sep 2007 15:18:34 +0000 (15:18 +0000)]
FIX #1038: failure of selector-thunk machinery to do its job
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.
Simon Marlow [Fri, 14 Sep 2007 14:55:19 +0000 (14:55 +0000)]
attempt to fix #1391, hold locks across fork() and initialize them in the child
Ian Lynagh [Sun, 23 Sep 2007 20:49:40 +0000 (20:49 +0000)]
Put packages in ../$$pkgid not ../$$pkg
Ian Lynagh [Sun, 23 Sep 2007 18:16:20 +0000 (18:16 +0000)]
Notice when C modules have changed when deciding whether or not to link
Based on a patch from Lemmih
Ian Lynagh [Sun, 23 Sep 2007 17:42:42 +0000 (17:42 +0000)]
Whitespace changes only
Ian Lynagh [Sun, 23 Sep 2007 16:36:40 +0000 (16:36 +0000)]
Remove remaining bits of bindist "make in-place"
Ian Lynagh [Sun, 23 Sep 2007 13:18:45 +0000 (13:18 +0000)]
Add an entry for strings treated as ISO-8859-1 to the users guide bug list
sven.panne@aedion.de [Sun, 23 Sep 2007 12:06:36 +0000 (12:06 +0000)]
Fix bug #1725 (Haddock links between packages)
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
sven.panne@aedion.de [Sun, 23 Sep 2007 09:43:58 +0000 (09:43 +0000)]
Unbreak "dist" target for fresh trees
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
shelarcy [Thu, 20 Sep 2007 13:01:59 +0000 (13:01 +0000)]
make stamp.inplace-gcc-lib copy $(LD) instead of $(GccDir)ld.exe to avoid building problem
v.dijk.bas@gmail.com [Thu, 20 Sep 2007 12:13:06 +0000 (12:13 +0000)]
Removed duplicate entry for derbugging flag -ddump-tc from the user guide
Ian Lynagh [Fri, 21 Sep 2007 16:53:16 +0000 (16:53 +0000)]
Fix building with compilers which don't have an integer for a patch level
Ian Lynagh [Fri, 21 Sep 2007 16:35:52 +0000 (16:35 +0000)]
Move OPTIONS pragmas above comments
Fixes building with -Werror (i.e. validate) and GHC < 6.6
Norman Ramsey [Fri, 21 Sep 2007 13:41:24 +0000 (13:41 +0000)]
massive convulsion in ZipDataflow
After my talk, I got the idea of 'shallow rewriting' for the
dataflow framework. Here it is implemented, along with
some related ideas late making Graph and not LGraph primary.
The only bad thing is that the whole bit is stitched together
out of ill-fitting pieces, kind of like Frankenstein's monster.
A new ZipDataflow will rise out of the ashes.
Norman Ramsey [Fri, 21 Sep 2007 13:40:35 +0000 (13:40 +0000)]
incomplete start on set of intervals for stack model
simonpj@microsoft.com [Wed, 19 Sep 2007 15:05:44 +0000 (15:05 +0000)]
Small changes to mk-ing flow graphs
- ZipCfg: add mkBlockId :: Unique -> BlockId
- MkZipCfg: change sequence --> catAGrpahs
- MkZipCfgCmm: add mkCmmIfThen
Not fully validated, but I don't think anything will break
simonpj@microsoft.com [Wed, 19 Sep 2007 15:03:45 +0000 (15:03 +0000)]
Mostly comments, following NR/SPJ meeting
Ian Lynagh [Thu, 20 Sep 2007 18:39:03 +0000 (18:39 +0000)]
Remove a redundant part of distrib/Makefile's "make install"
simonpj@microsoft.com [Thu, 20 Sep 2007 12:00:49 +0000 (12:00 +0000)]
Fix Trac #1718: interaction of error, unlifted tuples, and casts
See Note [Float coercions (unlifted)] in Simplify
Test is simpl018
Ian Lynagh [Wed, 19 Sep 2007 19:42:22 +0000 (19:42 +0000)]
Hack to get haskell-src's Parser.hs into source distributions
Ian Lynagh [Wed, 19 Sep 2007 14:01:00 +0000 (14:01 +0000)]
Fix --print-docdir for relocatable builds; fixes #1226
Ian Lynagh [Wed, 19 Sep 2007 14:00:37 +0000 (14:00 +0000)]
Use $(RelocatableBuild) rather than $(Windows)
simonpj@microsoft.com [Wed, 19 Sep 2007 17:12:07 +0000 (17:12 +0000)]
Fix exponential-time behaviour with type synonyms; rename -XPartiallyAppliedTypeSynonyms to -XLiberalTypeSynonyms
Fixes exponential behaviour present in GHC 6.6!
I renamed the flag because the old (not very old) name wasn't
describing what it does.
Manuel M T Chakravarty [Wed, 19 Sep 2007 15:07:38 +0000 (15:07 +0000)]
FIX #1688: Givens in checkLoop are not that rigid after all
- This patch re-instates the policy of 6.6.1 to zonk the given constraints
in the simplifier loop.
MERGE TO STABLE
Manuel M T Chakravarty [Wed, 19 Sep 2007 12:20:11 +0000 (12:20 +0000)]
FIX #1713: watch out for type families in splitAppTy functions
MERGE TO STABLE
Ian Lynagh [Tue, 18 Sep 2007 21:57:52 +0000 (21:57 +0000)]
Catch any exceptions thrown by getEnv; fixes #1704
simonpj@microsoft.com [Wed, 19 Sep 2007 14:28:21 +0000 (14:28 +0000)]
Make the error message mentioning -XPatternSignatures spell the flag correctly
Ian Lynagh [Tue, 18 Sep 2007 20:12:49 +0000 (20:12 +0000)]
Give Windows its own definitions of docdir (and htmldir etc)
Ian Lynagh [Tue, 18 Sep 2007 20:06:37 +0000 (20:06 +0000)]
Include build.mk or validate{,-settings}.mk through custom-settings.mk
Ian Lynagh [Tue, 18 Sep 2007 19:48:39 +0000 (19:48 +0000)]
Install extra-gcc-opts in bindists