Can Samuel please have this email set as allowed for this list?
---------- Forwarded message ----------
From: Samuel Bronson <[EMAIL PROTECTED]>
Date: Sep 24, 2006 7:33 PM
Subject: darcs patch: Add not-yet-working typeOf function for YHC.Dynamic
To: [EMAIL PROTECTED]
Sun Sep 24 14:13:22 EDT 2006 Samuel Bronson <[EMAIL PROTECTED]>
* Add not-yet-working typeOf function for YHC.Dynamic
For instance, print (typeOf "Hello!") gave me this output:
(TyGen "v300")
This reminds me of the impossibility of putting type signatures on
arbitrary subexpressions without scoped type variables. (Which
wouldn't help here anyway, since data isn't scoped...)
New patches:
[Add not-yet-working typeOf function for YHC.Dynamic
Samuel Bronson <[EMAIL PROTECTED]>**20060924181322
For instance, print (typeOf "Hello!") gave me this output:
(TyGen "v300")
This reminds me of the impossibility of putting type signatures on
arbitrary subexpressions without scoped type variables. (Which
wouldn't help here anyway, since data isn't scoped...)
] {
hunk ./src/packages/yhc-base-1.0/YHC/Dynamic.hs 1
-module YHC.Dynamic (PolyTypeRep, TypeRep, typeRep, toTypeRep) where
+module YHC.Dynamic (PolyTypeRep, TypeRep, typeOf, typeRep, toTypeRep) where
hunk ./src/packages/yhc-base-1.0/YHC/Internal.hs 42
- , PolyTypeRep, TypeRep(..), typeRep, _tyCon, _tyGen, toTypeRep
+ , PolyTypeRep, TypeRep(..), typeOf, typeRep, _tyCon, _tyGen, toTypeRep
hunk ./src/packages/yhc-base-1.0/YHC/Internal.hs 150
+
+typeOf :: a -> PolyTypeRep a
+typeOf _ = typeRep
+
}
Context:
[Give better type information - CoreItem is now CoreFunc and CoreData, CoreLet now has a more accurate type (breaks binary compatability, again...)
Neil Mitchell**20060920132218]
[Change the Show instance for Core, now export Show for each of the constructors, not just Core
Neil Mitchell**20060920130130]
[Add coreFunc to the Core API
Neil Mitchell**20060919175519]
[Another dependency for Ix.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060919171058]
[Add dependency for Data.Ix.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060919165550]
[Add a Play instance for Core, and initial Play infrastructure
Neil Mitchell**20060919164102]
[Change the CoreFunc format, to lift the name and arguments up explicitly - much more sensible! (breaks compatability with all external Core tools - but they are all mine)
Neil Mitchell**20060919161741]
[Delete the Read/Show instances for Core, people should use the binary stuff instead
Neil Mitchell**20060919153827]
[Make Core.Core purely generate Core from PosLambda, and the showing/saving to Compile
Neil Mitchell**20060919153344]
[Add a binary read/write for Core
Neil Mitchell**20060919143956]
[Move Core.Pretty to Yhc.Core.Show
Neil Mitchell**20060919142031]
[Move dropModule from Pretty to Yhc.Core.Type
Neil Mitchell**20060919141644]
[Move to Yhc.Core, just move the data structure for now
Neil Mitchell**20060919141044]
[Remove Core.Reduce, wasn't a very good idea, and wasn't used, and wouldn't build
Neil Mitchell**20060919140321]
[Add dependency for Ix on Data.Ratio.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060919082731]
[Copy bootstrap files to the compilation directory rather than installation.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060918190915]
[Minor tweak
Tom Shackell <[EMAIL PROTECTED]>**20060918183056]
[Added YHC.Dynamic support :-)
Tom Shackell <[EMAIL PROTECTED]>**20060918182436]
[Added mod_load as part of the Runtime.API (how did I forget that one?)
Tom Shackell <[EMAIL PROTECTED]>**20060918102314]
[If checking of type sizes failed delete the cache.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060915110303]
[Don't check for svn if we don't need it.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060914152928]
[Allow the user to skip pulling a copy of ctypes by passing skipctypes=1 on the command line. They must provide their own copy if this is going to work.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060914152640]
[Allow commandline options to be stored in a file (options.txt)
Andrew Wilkinson <[EMAIL PROTECTED]>**20060914151911]
[None isn't a valid value, use 0 instead.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060914145032]
[Change failed configure results to None to disable caching.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060914142841]
[Add proper pragma support into Yhc, delete it from Scons
Neil Mitchell**20060914135504]
[Split into processArgs and processMoreArgs, to allow OPTIONS pragma to add more parse information
Neil Mitchell**20060914130423]
[Remove some redundant code
Neil Mitchell**20060914130035]
[Added anna,fluid & prolog to tests
Tom Shackell <[EMAIL PROTECTED]>**20060914132309]
[Add a special core option for Neil. Type scons core=1 to activate.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060914112103]
[Allow the user to override the detected architecture.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060914110551]
[Recalculate dependencies if file modification time has changed.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060914105015]
[Only rebuild files if the .hi files they depend on change. Fixes bug #20.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060914104411]
[Added '-h 40M' flags to paraffins so it works on x64
Tom Shackell <[EMAIL PROTECTED]>**20060913180359]
[Improved the pic and removed the gamteb test
Tom Shackell <[EMAIL PROTECTED]>**20060913175829]
[Use absolute YHC_BASE_PATH
Andrew Wilkinson <[EMAIL PROTECTED]>**20060913164718]
[os.getcwd() doesn't end in a slash.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060913162706]
[Fixed the x2n1 bug
Tom Shackell <[EMAIL PROTECTED]>**20060913155320
The problem was actually a serious issue with regard to the NEEDHEAP analysis.
I'd forgotten that the amount of heap used by an APPLY instruction is not fixed
until runtime - and thus every APPLY needs to be followed by a NEED_HEAP
(providing no memory was allocated). The solution was simply to do that, have the
memory analysis phase introduce NEED_HEAP instructions after APPLY. The most common
case of
APPLY ...
EVAL
still requires no NEED_HEAP (which is correctly determined and removed automatically).
Tom
]
[Use absolute path for YHC_BASE_PATH
Andrew Wilkinson <[EMAIL PROTECTED]>**20060913155844]
[Fix up the test script, assuming that YHC_BASE_PATH is always absolute
Neil Mitchell**20060913154258]
[If the extension is .lhs, then always give the -unlit flag
Neil Mitchell**20060913150026]
[Move the Flags and FileFlags structures into FrontData, where they (hopefully) belong
Neil Mitchell**20060913143135]
[Move the Flags data into the FileFlags information, so each file can have different flags
Neil Mitchell**20060913135506]
[Incorporated nofib tests into the testsuite, modified yhi and tester to do so.
Tom Shackell <[EMAIL PROTECTED]>**20060913120540]
[Fix the Core output so its all with the right name, for Catch
Neil Mitchell**20060911211822]
[Add Eq instances for Core, entirely unneeded for Yhc, but makes Catch a bit easier ;)
Neil Mitchell**20060911175447]
[Make tuples desugar to the same thing everywhere (rather than just the use, not the definition!)
Neil Mitchell**20060911172801]
[Remove empty WheelSieve2 directory, since that test is now in the testsuite
Neil Mitchell**20060911144752]
[Move wheelsieve into the tests, in a manner guaranteed to invoke GC
Neil Mitchell**20060911144617]
[Take less prime numbers, but still enough to ensure GC happens
Neil Mitchell**20060911144159]
[Delete nqueens, is in the test directory as Queens
Neil Mitchell**20060911143951]
[Delete unneeded makefiles in the test directory
Neil Mitchell**20060911143840]
[Remove test/Lit.lhs, has now been moved to parsing/literate
Neil Mitchell**20060911143407]
[Wheelsieve fix (oops)
Tom Shackell <[EMAIL PROTECTED]>**20060911140815]
[Wheelsieve fix
Tom Shackell <[EMAIL PROTECTED]>**20060911140656]
[Make the initial file depend on both its .hi and .hbc file
Neil Mitchell**20060911140036]
[Check if the initial file is dirty or not, make recompilations really quick
Neil Mitchell**20060911133438]
[Add CoreDouble and CoreFloat, to encode floating point numbers in the Core
Neil Mitchell**20060911125853]
[Make lam2core cope with primitives and foreign function calls (which are treated as though they were primitive)
Neil Mitchell**20060911124513]
[Dump the -corep information to a file with the extension .ycr
Neil Mitchell**20060911111539]
[Don't link to libdl on FreeBSD
Andrew Wilkinson <[EMAIL PROTECTED]>**20060907151046]
[Fix an incompatability introduced by ctypes, by making ctypes depend on python
Neil Mitchell**20060907145727]
[Detect when the C compiler doesn't work. And don't require one if we're only building yhc.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060817142541]
[Pass entire environment to GHC. Avoids 'HOME: getEnv: does not exist (no environment variable)' message from GHC.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060815155119]
[Back out change which made Char signed as it's not on Linux PPC.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060815151658]
[Don't make assumptions about char.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060815124721]
[Remove libffi that was in the source code directory
Neil Mitchell**20060815134849]
[Mark packed string as expected to fail
Neil Mitchell**20060815134351]
[Add expected failure concept to tester, so that buildbot reports success
Neil Mitchell**20060815134320]
[Make the tests be executed in the same order regardless of the order the file system finds them in
Neil Mitchell**20060815130142]
[Remove build warnings from yhi.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060815110046]
[Force types to be signed. Should fix bug on ppc linux where char is unsigned by default.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060814105637]
[Don't check /usr/local on Windows
Andrew Wilkinson <[EMAIL PROTECTED]>**20060814100019]
[Add libffi build rule for MacOS X on x86
Andrew Wilkinson <[EMAIL PROTECTED]>**20060814083001]
[Link against libraries on non Darwin operating systems on PPC
Andrew Wilkinson <[EMAIL PROTECTED]>**20060810111259]
[Fix copy and paste error.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060810084957]
[Add support for linux on ppc.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060809105016]
[Actually make my last two patches work.
Andrew Wilkinson <[EMAIL PROTECTED]>**20060809093638]
[Check /usr/local for headers and library files
Andrew Wilkinson <[EMAIL PROTECTED]>**20060808133931]
[Fall back to Python if uname -o fails
Andrew Wilkinson <[EMAIL PROTECTED]>**20060808133902]
[TAG 03_AUG_2006
Neil Mitchell**20060803135817]
Patch bundle hash:
635382a5b33a0b445887fb144169028d48a54672
_______________________________________________
Yhc mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/yhc
