Ticket #2962: Fix-genericLength-space-leak.dpatch

File Fix-genericLength-space-leak.dpatch, 22.9 KB (added by thorkilnaur, 9 years ago)

Patch to fix space leak in genericLength

1Sat Jan 17 06:46:48 CET 2009  naur@post11.tele.dk
2  * Fix genericLength space leak
4New patches:
6[Fix genericLength space leak
7naur@post11.tele.dk**20090117054648] {
8hunk ./Data/List.hs 568
9 -- particular, instead of returning an 'Int', it returns any type which is
10 -- an instance of 'Num'.  It is, however, less efficient than 'length'.
11 genericLength           :: (Num i) => [b] -> i
13 genericLength []        =  0
14 genericLength (_:l)     =  1 + genericLength l
15hunk ./Data/List.hs 571
17+genericLength l         =  gl l 0
18+                        where
19+                           gl [] a     = a
20+                           gl (_:xs) a = let a' = a + 1 in a' `seq` gl xs a'
23 -- | The 'genericTake' function is an overloaded version of 'take', which
24 -- accepts any 'Integral' value as the number of elements to take.
29[Restore the Haskell 98 behaviour of Show Ratio (#1920)
30Simon Marlow <simonmarhaskell@gmail.com>**20080923134949]
31[Pad version number to
32Ian Lynagh <igloo@earth.li>**20080920155801]
33[TAG 6.10 branch has been forked
34Ian Lynagh <igloo@earth.li>**20080919123437]
35[In nhc98, Word is a type synonym, so class instance is not possible.
37[Fix bugs in Text.Printf (#1548)
38Simon Marlow <marlowsd@gmail.com>**20080916133505]
39[We should be including Rts.h here, not Stg.h
40Simon Marlow <marlowsd@gmail.com>**20080912134901
41 Stg.h is for .hc files only, and it sets up various global register
42 variables.
44[Generic functions that take integral arguments should work the same way as their prelude counterparts
47 The Prelude functions drop, take, and splitAt are unfailing (never call error). This patch changes the Data.List generic versions to behave the same way. At present, they call error on negative arguments.
50[Don't define __hscore_s_issock on Windows
51Ian Lynagh <igloo@earth.li>**20080904202845]
52[Unbreak the GHC build with older versions of gcc
53Ian Lynagh <igloo@earth.li>**20080904185122
54 Patch from kili@outback.escape.de, who says:
55   Stg.h must be included before HsBase.h, because the latter contains
56   function definitions causing older versions of gcc (3.3.5 in my
57   case) to bail out with "error: global register variable follows a
58   function definition" on Regs.h, which is included by Stg.h.
60[Don't make S_ISSOCK use conditional
61Ian Lynagh <igloo@earth.li>**20080904164234
62 We were conditionally defining the C wrapper, but unconditionally using
63 it. So if it didn't exist then things would have broken anyway.
65[Add missing files
66Ian Lynagh <igloo@earth.li>**20080904100951]
67[Remerge concurrent,unique,timeout,st,getopt into base
68Ian Lynagh <igloo@earth.li>**20080903201908]
69[remove 'pure' method from Arrow class (#2517)
70Ross Paterson <ross@soi.city.ac.uk>**20080903144436]
71[make Typeable instances for larger tuples available to non-GHC
72Ross Paterson <ross@soi.city.ac.uk>**20080903113543]
73[Don't look for actual OldException.Exception exceptions
74Ian Lynagh <igloo@earth.li>**20080902224730
75 We don't actually throw them (we throw the new Exception equivalents
76 instead), and looking for them was causing an infinite loop
78[add include/CTypes.h to extra-source-files
79Ross Paterson <ross@soi.city.ac.uk>**20080902153029]
80[avoid relying on the implementation of SomeException
81Ross Paterson <ross@soi.city.ac.uk>**20080902080113
83 This is because Hugs uses a different implementation.
84 No semantic change.
86[non-GHC: leave out Belch functions
87Ross Paterson <ross@soi.city.ac.uk>**20080831180227]
88[non-GHC: add Typeable instance for ForeignPtr
89Ross Paterson <ross@soi.city.ac.uk>**20080831180048]
90[docs: mention that killThread on a completed thread is a no-op
91Simon Marlow <marlowsd@gmail.com>**20080902093126]
92[#2528: reverse the order of args to (==) in nubBy to match nub
93Simon Marlow <marlowsd@gmail.com>**20080902092950
94 This only makes a difference when the (==) definition is not
95 reflexive, but strictly speaking it does violate the report definition
96 of nubBy, so we should fix it.
98[System.Timeout is no longer part of base
100[getopt is no longer part of base
102[Split syb off into its own package
103Ian Lynagh <igloo@earth.li>**20080825214144
104 I've also moved the Data (Complex a) instance into it, and made it
105 portable rather than GHC-only in the process.
107[add extra-source-files field
108Ross Paterson <ross@soi.city.ac.uk>**20080825231317]
109[Fix warnings in PrelIOUtils.c
110Ian Lynagh <igloo@earth.li>**20080825141841]
111[Windows-only fixes for moving concurrent out of base
112Ian Lynagh <igloo@earth.li>**20080824164146]
113[Split off the concurrent hierarchy (concurrent, unique, timeout)
114Ian Lynagh <igloo@earth.li>**20080824123956]
115[Split getopt off into its own package
116Ian Lynagh <igloo@earth.li>**20080824020213]
117[Remove ST stuff that is now in the new st package
118Ian Lynagh <igloo@earth.li>**20080823223014]
119[Fix Windows-only warnings
120Ian Lynagh <igloo@earth.li>**20080823002249]
121[Fix Windows-only warnings in GHC.Conc
122Ian Lynagh <igloo@earth.li>**20080822234837]
123[Suppress some warnings that are hard to fix because of ifdefs
124Ian Lynagh <igloo@earth.li>**20080822233951]
125[Provide blockedOnDeadMVar, blockedIndefinitely for the RTS
126Ian Lynagh <igloo@earth.li>**20080821110723]
127[Fix more warnings
128Ian Lynagh <igloo@earth.li>**20080820233958]
129[Suppress a couple of warnings in GHC.PArr
130Ian Lynagh <igloo@earth.li>**20080820232018
131 The fix isn't immediately obvious to me
133[Fix more warnings
134Ian Lynagh <igloo@earth.li>**20080820231937]
135[Fix warnings in Data.Generics.*
136Ian Lynagh <igloo@earth.li>**20080820230437]
137[Fix some more warnings
138Ian Lynagh <igloo@earth.li>**20080820223252]
139[Ignore some orphan warnings
140Ian Lynagh <igloo@earth.li>**20080820211901]
141[remove some functions that aren't used in base
142Simon Marlow <marlowsd@gmail.com>**20080821142339]
143[remove __hscore_renameFile, it is no longer uesd
144Simon Marlow <marlowsd@gmail.com>**20080818155950
145 System.Directory implements renameFile using unix/Win32 now.
147[Rewrite the documentation for forkOS again
148Simon Marlow <marlowsd@gmail.com>**20080818132856
149 Try to make it clearer that forkOS is only necessary when calling
150 foreing libraries that use thread-local state, and it has nothing to
151 do with scheduling behaviour between Haskell threads.  I also added
152 something about the performance impact of forkOS, and mentioned that
153 the main thread is a bound thread.
155[nhc only: expose Foldable and Traversable instances of Array
156Ross Paterson <ross@soi.city.ac.uk>**20080817002719
158 These were turned off as a side-effect of a previous nhc-only fix for
159 #2176 that is no longer needed.  They should be fine for nhc now.
161[Fix hReady (trac #1063)
162Ian Lynagh <igloo@earth.li>**20080816182715
163 We now throw an EOF exception when appropriate
165[Fix oversight in Control.OldException
166Bertram Felgenhauer <int-e@gmx.de>**20080816132631
167 The NonTermination constructor slipped through in the Exception instance.
169[Eliminate orphan rules and instances in the array package
170Ian Lynagh <igloo@earth.li>**20080816122253]
171[Control.OldException: Map exceptions to old exceptions and back properly.
172Ian Lynagh <igloo@earth.li>**20080814210219
173   * Control.OldException: Map exceptions to old exceptions and back properly.
175 It's really necessary to map them back as well, or the RTS and base library
176 will not recognize exceptions that got caught and rethrown. (See #2508)
178 Patch from Bertram Felgenhauer <int-e@gmx.de>
180[add Traversable generalizations of mapAccumL and mapAccumR (#2461)
181Ross Paterson <ross@soi.city.ac.uk>**20080814162617]
182[simplify definition of Prelude.catch
183Ross Paterson <ross@soi.city.ac.uk>**20080814143650]
184[remove returns from void functions
185Ross Paterson <ross@soi.city.ac.uk>**20080814110841]
186[No reason for Handler and catches to exclude nhc98.
188[Must import ExitCode for its instance to be re-exported.
190 The Cabal library depends on "instance Exception ExitCode", and expects
191 to import it from Control.Exception, not Control.Exception.Base.
193[use New.catch instead of catchException in OldException
194Ross Paterson <ross@soi.city.ac.uk>**20080813071307]
195[use the Haskell 98 module Control.Exception.Base in the Concurrent modules
196Ross Paterson <ross@soi.city.ac.uk>**20080813000219]
197[export Control.Exception.Base
198Ross Paterson <ross@soi.city.ac.uk>**20080812233640]
199[use dummy implementation of timeout for all non-GHCs
200Ross Paterson <ross@soi.city.ac.uk>**20080812151602]
201[Hugs only: fix imports
202Ross Paterson <ross@soi.city.ac.uk>**20080812145654]
203[non-GHC: hide Prelude.catch
204Ross Paterson <ross@soi.city.ac.uk>**20080812145622]
205[add Control.Exception.Base to nhc98 build
207[bump to version 4.0
208Simon Marlow <marlowsd@gmail.com>**20080805153354]
209[Hugs only: don't import exception types -- their instances are now in Control.Exception.Base
210Ross Paterson <ross@soi.city.ac.uk>**20080812140433]
211[split most of Control.Exception into new Control.Exception.Base
212Ross Paterson <ross@soi.city.ac.uk>**20080812124912
214 Move everything but catches/Handler into a new internal module.
215 This was needed to get the new exceptions working with Hugs, because Hugs
216 has the constraint that all Haskell 98 library modules, and everything
217 they include, must be Haskell 98.  This also involves a different
218 representation of SomeException for Hugs, so that type is exported
219 opaquely for Hugs.  Then Control.Exception.Base is Haskell 98 as far as
220 Hugs is concerned, but Control.Exception needs the extensions turned on.
222 Control.Exception re-exports everything from Control.Exception.Base
223 except the functions used by the GHC runtime.
225[remove kludges, now that Control.Exception is imported
226Ross Paterson <ross@soi.city.ac.uk>**20080811180328]
227[threadDelay and friends are GHC-only
228Ross Paterson <ross@soi.city.ac.uk>**20080811175039]
229[fix imports for non-GHC
231[Eq and Ord have moved into GHC.Classes
232Ian Lynagh <igloo@earth.li>**20080807095352]
233[Use the proper CInt type in GHC.Unicode
234Ian Lynagh <igloo@earth.li>**20080806232948]
235[Import wibbles
236Ian Lynagh <igloo@earth.li>**20080806232055]
237[Remove unnecessary Data/Dynamic.hs-boot
238Ian Lynagh <igloo@earth.li>**20080806230623]
239[Remove more redundant GHC.Float imports
240Ian Lynagh <igloo@earth.li>**20080806225411]
241[Remove an unnecessary import
242Ian Lynagh <igloo@earth.li>**20080806224742]
243[Move Int, Float and Double into ghc-prim:GHC.Types
244Ian Lynagh <igloo@earth.li>**20080806191554]
245[Put some explicit import lists in Data.Typeable
246Ian Lynagh <igloo@earth.li>**20080806190353]
247[Fix a couple of imports
248Ian Lynagh <igloo@earth.li>**20080806165549]
249[Remove unused conditional import
250Ian Lynagh <igloo@earth.li>**20080806124930]
251[Swap imports around to get GHC.ForeignPtr out of the base knot
252Ian Lynagh <igloo@earth.li>**20080806121313]
253[Move some bits around to stop Data.Either being in the base import knot
254Ian Lynagh <igloo@earth.li>**20080806120504]
255[Tweak an import
256Ian Lynagh <igloo@earth.li>**20080806000440]
257[Remove the DynIOError constructor of IOErrorType
258Ian Lynagh <igloo@earth.li>**20080805234720
259 As far as I can see it is never used or exported
261[Move some internals around to simplify the import graph a bit
262Ian Lynagh <igloo@earth.li>**20080805221341]
263[Move the Char datatype into ghc-prim
264Ian Lynagh <igloo@earth.li>**20080805204009]
265[Remove an unnecessary import
266Ian Lynagh <igloo@earth.li>**20080805182336]
267[The [] definition has moved to ghc-prim
268Ian Lynagh <igloo@earth.li>**20080805182332]
269[Fix warnings
270Ian Lynagh <igloo@earth.li>**20080805150250]
271[Add a missing case to Show AsyncException
272Ian Lynagh <igloo@earth.li>**20080805142811]
273[Remove GHC.Dotnet
274Ian Lynagh <igloo@earth.li>**20080804215840]
275[Hide standalone deriving clauses from haddock
276Ian Lynagh <igloo@earth.li>**20080804211617]
277[Control.Exception doesn't need to export assertError
278Ian Lynagh <igloo@earth.li>**20080804161838]
279[Generalise the type of mapException; pointed out by Isaac Dupree
280Ian Lynagh <igloo@earth.li>**20080804160941]
281[Remove some unnecessary Data.Tuple imports
282Ian Lynagh <igloo@earth.li>**20080804155956]
283[The tuple datatype definitions have moved to ghc-prim
284Ian Lynagh <igloo@earth.li>**20080804155420]
285[make ExitCode an instance of Exception for nhc98
287[poke and peek come from Foreign.Storable
289[zipWithM_ comes from Control.Monad
291[Fix nhc98 code variations to use the extensible exception API.
293 There is still only one real exception type in nhc98, so it is not truly
294 extensible.  But this is enough to get the base package building again.
296[nhc98 needs the Prelude for this module
298[Change some imports and derive Show (Either a b)
299Ian Lynagh <igloo@earth.li>**20080804004147
300 rather than writing it by hand in GHC.Show
302[Windows fixes
303Ian Lynagh <igloo@earth.li>**20080803180345]
304[Remove the duplicate definition of throwTo in Control.Exception
305Ian Lynagh <igloo@earth.li>**20080803141703
306 It now imports GHC.Conc, so it is no longer necessary
308[Remove the only import of GHC.Exts
309Ian Lynagh <igloo@earth.li>**20080803141944]
310[Move assertError into GHC.IOBase
311Ian Lynagh <igloo@earth.li>**20080803141040]
312[Use onException rather than catchAny
313Ian Lynagh <igloo@earth.li>**20080803114104]
314[Generalise the type of onException
315Ian Lynagh <igloo@earth.li>**20080803003001
316 The type of the thing to do on an exception is now
317     IO b
318 rather than
319     IO ()
320 which better matches functions like bracket.
322[Remove the dangerous Exception functions
323Ian Lynagh <igloo@earth.li>**20080802231358
324 Removed: catchAny, handleAny, ignoreExceptions
325 These make it easy to eat /any/ exception, which is rarely what you want.
326 Normally you either want to:
327 * only catch exceptions in a certain part of the hierarchy, e.g.
328   "file not found", in which case you should only catch exceptions
329   of the appropriate type,
330 or
331 * you want to do some cleanup when an exception happens, and then rethrow
332   the exception, in which case you should use onException, or one of the
333   bracketing functions.
335[Remove an unused import
336Ian Lynagh <igloo@earth.li>**20080801230343]
337[Remove unused imports
338Ian Lynagh <igloo@earth.li>**20080801230059]
339[Remove unused imports in Control.Exception
340Ian Lynagh <igloo@earth.li>**20080801225847]
341[Get rid of some duplicate imports
342Ian Lynagh <igloo@earth.li>**20080801214933]
343[Remove the now-unused GHC/Conc.lhs-boot
344Ian Lynagh <igloo@earth.li>**20080801214707]
345[Make some more imports non-recursive
346Ian Lynagh <igloo@earth.li>**20080801214546]
347[Rejig some code so Control.Exception and GHC.Conc don't need recursive imports
348Ian Lynagh <igloo@earth.li>**20080801214208]
349[Remove the now-unused GHC/TopHandler.lhs-boot
350Ian Lynagh <igloo@earth.li>**20080801212105]
351[Reshuffle GHC.Conc/GHC.TopHandler a bit to remove a recursive import
352Ian Lynagh <igloo@earth.li>**20080801211801]
353[Don't import Control.Concurrent.MVar in GHC.TopHandler
354Ian Lynagh <igloo@earth.li>**20080801200123]
355[Export assertError from Control.Exception to make GHC happy
356Ian Lynagh <igloo@earth.li>**20080801111716
357 It's a wired-in name in GHC. We should possibly move it to another module.
359[TopHandler now uses the new extensible exceptions
360Ian Lynagh <igloo@earth.li>**20080731153553]
361[Comment wibble
362Ian Lynagh <igloo@earth.li>**20080730202127]
363[Make numericEnumFrom more efficient
364Ian Lynagh <igloo@earth.li>**20080730202049]
365[Put in some parens to clarify how things parse
366Ian Lynagh <igloo@earth.li>**20080730201934]
367[applied patches to make enumFrom and friends strict in arguments as per the Report; closes ticket #1997
368Bart Massey <bart@cs.pdx.edu>**20080726080444]
369[Don't use "deriving Typeable" (for portability reasons)
370Ian Lynagh <igloo@earth.li>**20080730194434]
371[Add onException
372Ian Lynagh <igloo@earth.li>**20080730172014]
373[Fix whitespace
374Ian Lynagh <igloo@earth.li>**20080730171951
375 The space after "\begin{code}" was confusing haddock
377[Re-add blocked; it got lost in the extensible exceptions patches
378Ian Lynagh <igloo@earth.li>**20080730145614]
379[Start to actually use extensible exceptions
380Ian Lynagh <igloo@earth.li>**20080730145115]
381[Rejig the extensible exceptions so there is less circular importing
382Ian Lynagh <igloo@earth.li>**20080730122539]
383[Define nonTermination for the RTS to use
384Ian Lynagh <igloo@earth.li>**20080621144420
385 We'll probably need to do the same for some other exceptions too
387[Use extensible exceptions at the lowest level
388Ian Lynagh <igloo@earth.li>**20080621121501
389 Everything above is largely unchanged; just the type of catch and throw.
391[add comment
392Simon Marlow <marlowsd@gmail.com>**20080730114559]
393[add some big warnings to the docs for unsafeIOToSTM (#2401)
394Simon Marlow <marlowsd@gmail.com>**20080730114554]
395[FIX #2376: inline shiftR
396Simon Marlow <marlowsd@gmail.com>**20080730103539
397 Duplicating the default definition for shiftR doesn't seem quite right
398 to me, but it gets the right results when compiling the example
399 program, and I couldn't find a better way to do it.
401[Add instance Show Control.Exception.Exception for nhc98.
403[Extend nhc98's Exception type to resemble ghc's more closely
405[fix dummy async implementations for non-GHC
406Ross Paterson <ross@soi.city.ac.uk>**20080715125521]
407[Fix haddocking with older haddocks
408Ian Lynagh <igloo@earth.li>**20080710190855]
409[Add threadStatus :: ThreadId -> IO ThreadStatus
410Simon Marlow <marlowsd@gmail.com>**20080710151711
412 -- | The current status of a thread
413 data ThreadStatus
414   = ThreadRunning
415         -- ^the thread is currently runnable or running
416   | ThreadFinished
417         -- ^the thread has finished
418   | ThreadBlocked  BlockReason
419         -- ^the thread is blocked on some resource
420   | ThreadDied
421         -- ^the thread received an uncaught exception
422   deriving (Eq,Ord,Show)
424 data BlockReason
425   = BlockedOnMVar
426         -- ^blocked on on 'MVar'
427   | BlockedOnBlackHole
428         -- ^blocked on a computation in progress by another thread
429   | BlockedOnException
430         -- ^blocked in 'throwTo'
431   | BlockedOnSTM
432         -- ^blocked in 'retry' in an STM transaction
433   | BlockedOnForeignCall
434         -- ^currently in a foreign call
435   | BlockedOnOther
436         -- ^blocked on some other resource.  Without @-threaded@,
437         -- I/O and 'threadDelay' show up as 'BlockedOnOther', with @-threaded@
438         -- they show up as 'BlockedOnMVar'.
439   deriving (Eq,Ord,Show)
441 This is useful for concurrency debugging.  I've left threadStatus in
442 GHC.Conc for now, since the ThreadStatus type is somewhat GHC-specific.
444[forkOS: start the new thread in blocked mode iff the parent was (#1048)
445Simon Marlow <marlowsd@gmail.com>**20080709135558
446 This matches the behaviour of forkIO
448[Add Control.Exception.blocked :: IO Bool
449Simon Marlow <marlowsd@gmail.com>**20080709133139
450 Tells you whether async exceptions are currently blocked or not.
452[FIX BUILD (on Windows)
453Simon Marlow <marlowsd@gmail.com>**20080709123110]
454[check CONST_SIGINT
455Simon Marlow <marlowsd@gmail.com>**20080709122527]
456[Make threadWaitRead/threadWaitWrite partially useable on Windows
457Simon Marlow <marlowsd@gmail.com>**20080709111008
459 They work with -threaded by calling fdReady() in a separate thread.
461 "threadWaitRead 0" also works without -threaded (because we happen to
462 know it's virtually equivalent to "hWaitForInput stdin (-1)").
464[FIX #1198: hWaitForInput on Windows
465Simon Marlow <marlowsd@gmail.com>**20080708134254
466 Now we do the appropriate magic in fdReady() to detect when there is
467 real input available, as opposed to uninteresting console events.
469[FIX part of #2301
470Simon Marlow <marlowsd@gmail.com>**20080709094437
472 Control-C now causes the new exception (AsyncException UserInterrupt)
473 to be raised in the main thread.  The signal handler is set up by
474 GHC.TopHandler.runMainIO, and can be overriden in the usual way by
475 installing a new signal handler.  The advantage is that now all
476 programs will get a chance to clean up on ^C.
478 When UserInterrupt is caught by the topmost handler, we now exit the
479 program via kill(getpid(),SIGINT), which tells the parent process that
480 we exited as a result of ^C, so the parent can take appropriate action
481 (it might want to exit too, for example).
483 One subtlety is that we have to use a weak reference to the ThreadId
484 for the main thread, so that the signal handler doesn't prevent the
485 main thread from being subject to deadlock detection.
487[() has moved to ghc-prim:GHC.Unit, and the Eq and Ord instances to Data.Tuple
488Ian Lynagh <igloo@earth.li>**20080624144932]
489[Add GHC.Exts.maxTupleSize :: Int, the size of the largest tuple supported
490Ian Lynagh <igloo@earth.li>**20080622141559]
491[Remove code for older GHC versions
492Ian Lynagh <igloo@earth.li>**20080620194521]
493[Make the macros in Typeable.h add type signatures
494Ian Lynagh <igloo@earth.li>**20080619235808]
495[Fix #2363: getChar cannot be interrupted with -threaded
496Simon Marlow <marlowsd@gmail.com>**20080619141911
497 Now in -threaded mode, instead of just making a blocking call to
498 read(), we call select() first to make sure the read() won't block,
499 and if it would block, then we use threadWaitRead.
501 The idea is that the current thread must be interruptible while it
502 blocks.  This is a little slower than before, but the overhead only
503 applies to blocking Handles (stdin/stdout/stderr, and those created by
504 System.Process).
506[Remove -fglasgow-exts from pragmas and comments
507Ian Lynagh <igloo@earth.li>**20080616230727]
508[Avoid using deprecated flags
509Ian Lynagh <igloo@earth.li>**20080616145207]
510[delete __hscore_{mkstemp,getrlimit,setrlimit} (moved to unix)
511Ross Paterson <ross@soi.city.ac.uk>**20080615224413]
512[Update WCsubst.c for Unicode 5.1.0, and add a README.Unicode
513Ian Lynagh <igloo@earth.li>**20080613201754
514 README.Unicode describes how to do updates in the future.
516[Fix ubconfc
517Ian Lynagh <igloo@earth.li>**20080613201456
518 The current code doesn't seem to be what was used to generate WCsubst.c,
519 so I'm not sure if it never worked, or if my tools work slightly
520 differently to those of the previous user.
522['permutations' is now more lazy and also faster
523Twan van Laarhoven <twanvl@gmail.com>**20080102231712]
524['subsequences' is now more lazy and also faster
525Twan van Laarhoven <twanvl@gmail.com>**20080102231629]
526[Add 'subsequences' and 'permutations' to Data.List
527Twan van Laarhoven <twanvl@gmail.com>**20071218154950]
528[Tweak the definition of (^) again
529Ian Lynagh <igloo@earth.li>**20080601120759
530 This fixes trac #2306 (do the minimum number of (*)s), and also means
531 that we don't use the value of (1 :: a) which causes problems if the
532 Num a definition isn't complete.
534[note about evaluation affecting StableNames
535Simon Marlow <marlowsd@gmail.com>**20080527110549]
536[TAG 2008-05-28
537Ian Lynagh <igloo@earth.li>**20080528003830]
538Patch bundle hash: