Hi everyone,
Here's a patch that adds (hopefully) correct pre-processing to the Haddock
2 command. It should really be added to the Haddock < 2 path as well
(which AFAIK doesn't do any pre-processing except for CPP and Unlit), but
I've avoided this because of the risk of introducing bugs in the much-used
Haddock < 2 path. In the future this code will be replaced by the new make
system anyway.
The patch also adds CPP include paths and output paths to the commands
sent to GHC (via Haddock). The output paths are needed if certain changes
to the GHC API don't make it into GHC 6.8.2.
With this patch I'm now able to process all boot and core libs of GHC
6.8.1 including base (with a small patch for its Setup.lhs). Hopefully we
can get it into the next version of Cabal bundled with GHC.
David
New patches:
[Do proper pre-processing for Haddock 2
David Waern <[EMAIL PROTECTED]>**20071109162557
Besides pre-processing, this patch adds include paths and output paths
to the ghc flags passed to Hadddock.
] {
hunk ./Distribution/Simple/Build.hs 46
- build, makefile
+ build, makefile, initialBuildSteps
hunk ./Distribution/Simple/Haddock.hs 61
+import Distribution.Simple.Build (initialBuildSteps)
hunk ./Distribution/Simple/Haddock.hs 78
-import Control.Monad (liftM, when, join)
+import Control.Monad (liftM, when, unless, join)
hunk ./Distribution/Simple/Haddock.hs 80
-import Data.List (nub)
hunk ./Distribution/Simple/Haddock.hs 81
+import Data.List (nub)
hunk ./Distribution/Simple/Haddock.hs 120
- let mockAll bi = mapM_ (mockPP ["-D__HADDOCK__"] bi tmpDir)
hunk ./Distribution/Simple/Haddock.hs 127
+
+ let mockFlags
+ | isVersion2 = []
+ | otherwise = ["-D__HADDOCK__"]
+
+ let mockAll bi = mapM_ (mockPP mockFlags bi tmpDir)
+
hunk ./Distribution/Simple/Haddock.hs 196
- let haddock2options bi = if isVersion2
- then ("-B" ++ ghcLibDir) : map ("--optghc=" ++) (ghcSimpleOptions lbi bi)
+ let haddock2options bi preprocessDir = if isVersion2
+ then ("-B" ++ ghcLibDir) : map ("--optghc=" ++) (ghcSimpleOptions lbi bi preprocessDir)
hunk ./Distribution/Simple/Haddock.hs 200
+ when isVersion2 $ initialBuildSteps pkg_descr lbi verbosity suffixes
+
hunk ./Distribution/Simple/Haddock.hs 204
- inFiles <- getModulePaths lbi bi (exposedModules lib ++ otherModules bi)
- mockAll bi inFiles
+ modules = exposedModules lib ++ otherModules bi
+ inFiles <- getModulePaths lbi bi modules
+ unless isVersion2 $ mockAll bi inFiles
hunk ./Distribution/Simple/Haddock.hs 213
- let outFiles = replaceLitExts inFiles
+ let targets
+ | isVersion2 = modules
+ | otherwise = replaceLitExts inFiles
hunk ./Distribution/Simple/Haddock.hs 231
- ++ haddock2options bi
- ++ outFiles
+ ++ haddock2options bi (buildDir lbi)
+ ++ targets
hunk ./Distribution/Simple/Haddock.hs 250
- let outFiles = replaceLitExts inFiles
+ let targets
+ | isVersion2 = srcMainPath : otherModules bi
+ | otherwise = replaceLitExts inFiles
+ let preprocessDir = buildDir lbi </> exeName exe </> exeName exe ++ "-tmp"
hunk ./Distribution/Simple/Haddock.hs 263
- ++ haddock2options bi
- ++ outFiles
+ ++ haddock2options bi preprocessDir
+ ++ targets
hunk ./Distribution/Simple/Haddock.hs 290
-ghcSimpleOptions :: LocalBuildInfo -> BuildInfo -> [String]
-ghcSimpleOptions lbi bi
+ghcSimpleOptions :: LocalBuildInfo -> BuildInfo -> FilePath -> [String]
+ghcSimpleOptions lbi bi mockDir
hunk ./Distribution/Simple/Haddock.hs 293
- ++ ["-i" ++ autogenModulesDir lbi]
- ++ ["-i" ++ l | l <- nub (hsSourceDirs bi)]
hunk ./Distribution/Simple/Haddock.hs 294
+ ++ ["-i"]
hunk ./Distribution/Simple/Haddock.hs 296
+ ++ ["-i" ++ autogenModulesDir lbi]
+ ++ ["-i" ++ l | l <- nub (hsSourceDirs bi)]
+ ++ ["-i" ++ mockDir]
+ ++ ["-I" ++ dir | dir <- includeDirs bi]
+ ++ ["-odir", mockDir]
+ ++ ["-hidir", mockDir]
}
Context:
[fix compilation with GHC 6.2.x
Simon Marlow <[EMAIL PROTECTED]>**20071105113004]
[Install the haddock interface file to the right place
Ian Lynagh <[EMAIL PROTECTED]>**20071030170254]
[Add an interfacedir configure flag, for where to put haddock insterface files
Ian Lynagh <[EMAIL PROTECTED]>**20071029174908]
[put the binary-dist copy in the right place
Simon Marlow <[EMAIL PROTECTED]>**20071026132515]
[fix $(TOP)
Simon Marlow <[EMAIL PROTECTED]>**20071026101915]
[fix XML (at least, makes it work here)
Simon Marlow <[EMAIL PROTECTED]>**20071026101907]
[no longer need to pass --allow-missing-html to haddock
Ross Paterson <[EMAIL PROTECTED]>**20071026084124
This option only affects Haddock if it is invoked with --use-package,
and Cabal no longer uses that option, as it now gets the arguments for
--read-interface from ghc-pkg directly (cf patch "rejig location of
package interfaces for haddock").
]
[no longer need to pass --ghc-pkg to haddock
Ross Paterson <[EMAIL PROTECTED]>**20071026073126
Haddock only runs ghc-pkg if invoked with --use-package, and Cabal no
longer uses that option, as it now gets the arguments for --read-interface
from ghc-pkg directly (cf patch "rejig location of package interfaces
for haddock").
]
[fix help text (--PROG-arg is now --PROG-option)
Ross Paterson <[EMAIL PROTECTED]>**20071026001045]
[Adjust verbosity of a step in configure
Duncan Coutts <[EMAIL PROTECTED]>**20071024160758]
[Compile a .c rather than a .hs file to make a .o file in ld -x configure test
Duncan Coutts <[EMAIL PROTECTED]>**20071024150505
Since when bootstrapping ghc we're not in any position to compile .hs files
that easily.
]
[Create the dist/ dir early in the configuration process
Duncan Coutts <[EMAIL PROTECTED]>**20071024133554]
[Create temp files in dist, since it is now guaranteed to exist
Duncan Coutts <[EMAIL PROTECTED]>**20071024102850]
[Regenerate GHC/Makefile.hs from GHC/Makefile.in
Duncan Coutts <[EMAIL PROTECTED]>**20071023183804]
[Only use ld -x on systems where ld supports that.
Duncan Coutts <[EMAIL PROTECTED]>**20071023183738
Hopefully this fixes it for both ordinary builds and via ghc makefiles.
]
[pass cpp-options to cpphs
Ross Paterson <[EMAIL PROTECTED]>**20071023184210]
[Refuse to run any commands if the .cabal has been modified
Simon Marlow <[EMAIL PROTECTED]>**20071023093111
See GHC bug #1372
This is a consistency check, intended to prevent this class of build
failures:
* Package P is updated, its version number is bumped, the
new version is compiled and installed.
* Package Q depends on P. Q is modified to use the new P, and Q's
.cabal file is updated with changes to the build-depends field to
depend on the new version of P.
* The user has an old build of Q. They pull the changes to Q and
'setup build' without cleaning or re-configuring. Build errors
ensue, because the code of Q depends on changes to P's API, and
we're still building against the old P.
Note that you can't get segfaults this way, only build errors.
This also relies on some new consistency checking in GHC 6.8 to work
properly. If the user re-configures their Q build and then issues
'setup build' without cleaning, GHC must now realise that the package
flags have changed, and re-compile all the affected Q modules. GHC
6.6 would not do this, but 6.8 does.
]
[Be explicit about the base version dependency
Duncan Coutts <[EMAIL PROTECTED]>**20071023122417
don't just leave it up to the default search order
]
[typo in comment
Ross Paterson <[EMAIL PROTECTED]>**20071022091235]
[Fix a bug in the Unlit pre-processor
David Waern <[EMAIL PROTECTED]>**20071024181908
With this patch, unlit can handle line pragmas with filenames that contain
spaces.
]
[Correct the spelling of mingw32 in os alias list
Duncan Coutts <[EMAIL PROTECTED]>**20071018195641
Doh!
]
[refinement of fix for #1785: don't use xargs' -s option at all
Simon Marlow <[EMAIL PROTECTED]>**20071019124522]
[FIX GHC bug #1785: use 2048 as the maximum command-line size
Simon Marlow <[EMAIL PROTECTED]>**20071018140500
Apparently Solaris sets a limit of 2048 here
]
[don't fail if xxx_hsc_make.c is gone
Ross Paterson <[EMAIL PROTECTED]>**20071018164245
The non-GHC hsc2hs deletes it even if the compilation fails.
]
[Use cpp-options rather than abusing ghc-options
Duncan Coutts <[EMAIL PROTECTED]>**20071017164914]
[Add a "cpp-options:" field and use it for pre-processing .hs .hsc .chs files
Duncan Coutts <[EMAIL PROTECTED]>**20071017164747
This is for pre-processing Haskell modules, not for C code. We already have
cc-options for that purpose. Up 'til now people have been abusing ghc-options
for this purpose. Even Cabal itself was guilty of doing that.
]
[Figure out if hsc2hs is using gcc or ghc as it's C compiler
Duncan Coutts <[EMAIL PROTECTED]>**20071017143108
and pass the appropriate flags on the basis of knowing that.
This is a hack.
What we should do longer term is make hsc2hs always use gcc as it's C compiler
and have Cabal figure out the right flags to pass it, rather than using ghc
to pass on the appropriate flags to gcc.
]
[Change the handling of cpp & ghc flags for hsc2hs
Duncan Coutts <[EMAIL PROTECTED]>**20071016231652
The hsc2hs that comes with ghc uses ghc as the C compiler. This means we must
escape the real cc flags. It also means we can ask ghc to add extra include
dirs that we might need to find includes of dependent packages. This is a bit
of a hack. In the longer term it'd be better for Cabal to collect the include
dirs and cc options of dependent packages and to pass them explicitly itself.
]
[Improve error messages for failed sanity checks.
Thomas Schilling <[EMAIL PROTECTED]>**20070925144955]
[Translate flat files into sectionized files, by duplicating global
Thomas Schilling <[EMAIL PROTECTED]>**20070925144921
dependecies into each non-empty section. The previous solution dumped
them into the library section, even if it would have been empty
otherwise.
]
[Move ghcVerbosity function into GHC module to share code
Duncan Coutts <[EMAIL PROTECTED]>**20071014165730]
[Note current development version number in release notes
Duncan Coutts <[EMAIL PROTECTED]>**20071012121835]
[Update the TODO list, mostly removing things.
Duncan Coutts <[EMAIL PROTECTED]>**20071012121702
Most of these are duplicated by Trac bugs.
We should try and get rid of this TODO file completely and just use Trac.
]
[Add logging functions notice, info, debug functions and use them consistently
Duncan Coutts <[EMAIL PROTECTED]>**20071012113237
We previously had this kind of code all over the place:
> when (verbosity >= verbose)
> (putStrLn "some message")
We now replace that with:
> info verbosity "some message"
Much nicer.
]
[When parsing ghc-pkg output, only look at the first package.conf file for GlobalPackageDB and SpecificPackageDB, and all package.conf files for UserPackageDB.
[EMAIL PROTECTED]
Before, we would consider user packages when fulfilling dependencies
for global installs. ghc-pkg will refuse to install packages globally if they
use user packages. Thus, without this patch, global installs can fail when you have user packages installed.
]
[Pass -c (silent create) to ar when verbosity < normal.
[EMAIL PROTECTED]
[Change --verbosity= option of SetupWrapper (cabal-setup) to --verbose=, since that is what the rest of Cabal uses.
[EMAIL PROTECTED]
[Pass -w -v0 to ghc when compiling Setup.{lhs,hs} in SetupWrapper when verbosity == silent.
[EMAIL PROTECTED]
[Update documentation on configurations
Duncan Coutts <[EMAIL PROTECTED]>**20071012015338
Describe the new syntax and make variuous changes to the
description of the meaning.
]
[Use -O2 for compiling .c files when we configure --enable-optimization
Duncan Coutts <[EMAIL PROTECTED]>**20071011223607
Seems a reasonable default behaviour.
]
[Add support for Haddock 2.0
David Waern <[EMAIL PROTECTED]>**20071021231415]
[parameterise InstalledPackageInfo over the type of module names
Simon Marlow <[EMAIL PROTECTED]>**20071003114947
This is useful in GHC, we can instantiate InstalledPackageInfo to
ModuleName and avoid lots of String<->FastString conversions.
]
[Remove commented-out code.
[EMAIL PROTECTED]
Push to 1.2.
]
[Rename parseDescription to parsePackageDescription.
[EMAIL PROTECTED]
This should be pushed to cabal-1.2 to make cabal-install work with 1.2.
]
[Bump version number to 1.3
[EMAIL PROTECTED]
[Expose parseDescription.
[EMAIL PROTECTED]
[Fix GenericPrackageDescription pretty printing to make it parsable. It still does not include all information.
[EMAIL PROTECTED]
[Haddock comment for the available dependencies argument to finalizePackageDescription.
[EMAIL PROTECTED]
[Add extra-libs to shared library linking
Clemens Fruhwirth <[EMAIL PROTECTED]>**20070927123923]
[Actually -stubdir only works well in ghc-6.8 due to -I search dir mess
Duncan Coutts <[EMAIL PROTECTED]>**20070926131843
In ghc-6.6 the Foo/Bar.hc files only #include "Bar_stub.h" rather than
#include "Foo/Bar_stub.h". This means when we set the stubdir so that the
_stub.h files don't sit next to the .hs file then the include search path
is not right to find the _stub.h file. In ghc-6.8 this is fixed so that
it adds the -stubdir path to the include search path when calling gcc.
]
[The -stubdir flag is supported in ghc-6.6 not just 6.6.1
Duncan Coutts <[EMAIL PROTECTED]>**20070926114223]
[Add -package-name to GHC invocation when linking DSOs
Clemens Fruhwirth <[EMAIL PROTECTED]>**20070926090025]
[Be more fuzzy with os, arch and impl matching in conditions. Fixes bug #158.
Duncan Coutts <[EMAIL PROTECTED]>**20070925132608
Do all the comparisons case insensitively and add some OS aliases so that
if os(windows) works if the System.Info.os is actually "mingw32".
]
[Don't generate links when dependent docs not installed. Fixes bug #157.
Duncan Coutts <[EMAIL PROTECTED]>**20070924142520
Generates a warning rather than failing as before.
]
[Fix warnings
Ian Lynagh <[EMAIL PROTECTED]>**20070923122921]
[Fix pre-processing for executables
Duncan Coutts <[EMAIL PROTECTED]>**20070922105414
The pre-processed files should go into the exe's build dir, not the
lib's build dir. Also pre-process main modules, fixing bug #14.
]
[Add a boring file
Ian Lynagh <[EMAIL PROTECTED]>**20070913203550]
[Remove some now-unecessary cleaning that causes problems for haskell-src
Ian Lynagh <[EMAIL PROTECTED]>**20070920203257
The code that removes .hs files in the source tree generated from .y files
had a comment:
XXX: This is actually no longer necessary, but we keep it, so that
clean works correctly on packages built with an older version of Cabal
This was causing problems for source distributions that include such
generated files (including haskell-src in extralibs).
]
[Move expensive conviguration actions inside branch where they're used
Duncan Coutts <[EMAIL PROTECTED]>**20070917085834
We only need to configure the compiler and resolve any "Cabal-Version:"
dependency when we're actually going to compile the Setup.hs program.
We can otherwise save time by not calling ghc and ghc-pkg unnecessarily.
(I only noticed this because I've got >150 registered packages which
causes ghc and ghc-pkg to be very slow to start.)
]
[Note in .cabal file that we're using a custom build-type
Duncan Coutts <[EMAIL PROTECTED]>**20070917033959]
[Pass on all the cabal-setup options to cabal, not just unrecognised ones
Duncan Coutts <[EMAIL PROTECTED]>**20070917025121
So for example -v and -w get passed on rather than swallowed by cabal-setup
itself.
]
[Put setup program and setup .o/.hi files into dist/setup
Duncan Coutts <[EMAIL PROTECTED]>**20070917024907
That way they get cleaned automatically and do not clutter the top dir.
]
[Correct the verbosity range in the "--help" text
Duncan Coutts <[EMAIL PROTECTED]>**20070917024416
The correct range in 0--3, not 0--2 or 0--5
]
[look in odir for source and .hs files, not buildDir lbi
Duncan Coutts <[EMAIL PROTECTED]>**20070917145030
odir = buildDir lbi for the lib case, which is why it worked most of the time
in testing but in the exe case it's a different dir.
]
[Put _stub.{c|h} files under dist/ so they'll get cleaned. Fixes bug #154.
Duncan Coutts <[EMAIL PROTECTED]>**20070917123042
Only used with ghc-6.6 and later which supports the -stubdir flag.
With earlier ghc versions the files still end up in the src dirs and so do not
get cleaned.
]
[Fix haddockDir
Ian Lynagh <[EMAIL PROTECTED]>**20070912133051]
[Add htmlDirTemplate to inplaceDirs
Ian Lynagh <[EMAIL PROTECTED]>**20070912125749]
[Add a --htmldir flag
Ian Lynagh <[EMAIL PROTECTED]>**20070912122145]
[Don't forcibly append "pkgName (package pkg_descr)" to htmldir
Ian Lynagh <[EMAIL PROTECTED]>**20070911192814]
[TAG 2007-09-06
Ian Lynagh <[EMAIL PROTECTED]>**20070906212150]
[use OPTIONS instead of OPTIONS_GHC for now, the latter doesn't work with GHC 6.2.x
Simon Marlow <[EMAIL PROTECTED]>**20070912105243]
[Don't create empty data dirs. Fixes bug #153.
Duncan Coutts <[EMAIL PROTECTED]>**20070912113803
Patch contributed by Sven Panne.
]
[Use confgurations to help build Cabal for ghc-6.2.x
Duncan Coutts <[EMAIL PROTECTED]>**20070912112934
Replacing a long-standing comment telling people how to do it manually.
]
[warning police
Simon Marlow <[EMAIL PROTECTED]>**20070907140731]
[Update D.S.GHC/Makefile.hs
Clemens Fruhwirth <[EMAIL PROTECTED]>**20070906213507]
[Add shared library building to GHC module (also via Makefile)
Clemens Fruhwirth <[EMAIL PROTECTED]>**20070906213132]
[fix type clash: Data.Version -> Distribution.Version
[EMAIL PROTECTED]
[fix broken #ifdefs for nhc98
[EMAIL PROTECTED]
[spell nhc98-options correctly
[EMAIL PROTECTED]
[Fix various mispellings of nhc98.
[EMAIL PROTECTED]
[nhc-options should be spelled nhc98-options
[EMAIL PROTECTED]
[TAG 1.2.0
Duncan Coutts <[EMAIL PROTECTED]>**20070906121830]
Patch bundle hash:
55299b171bec53d8df9673c8c9c307db1a6e320a_______________________________________________
cabal-devel mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/cabal-devel
