haskell-directory.git
19 years ago[project @ 2005-03-10 17:23:06 by malcolm] nhc98-1-18-release
malcolm [Thu, 10 Mar 2005 17:23:06 +0000 (17:23 +0000)]
[project @ 2005-03-10 17:23:06 by malcolm]
Change configuration for nhc98 on Cygwin only.

19 years ago[project @ 2005-03-10 10:00:39 by simonpj]
simonpj [Thu, 10 Mar 2005 10:00:39 +0000 (10:00 +0000)]
[project @ 2005-03-10 10:00:39 by simonpj]
Read instances for tuples

19 years ago[project @ 2005-03-09 17:47:50 by simonpj]
simonpj [Wed, 9 Mar 2005 17:47:50 +0000 (17:47 +0000)]
[project @ 2005-03-09 17:47:50 by simonpj]
Add instances for Bounded and Show up to 15-tuples

19 years ago[project @ 2005-03-07 13:02:37 by simonmar]
simonmar [Mon, 7 Mar 2005 13:02:37 +0000 (13:02 +0000)]
[project @ 2005-03-07 13:02:37 by simonmar]
Add dynTypeRep, from John Meacham.

19 years ago[project @ 2005-03-07 10:40:44 by simonmar]
simonmar [Mon, 7 Mar 2005 10:40:44 +0000 (10:40 +0000)]
[project @ 2005-03-07 10:40:44 by simonmar]
merge rev. 1.4.2.1 to HEAD

19 years ago[project @ 2005-03-05 15:13:01 by panne]
panne [Sat, 5 Mar 2005 15:13:01 +0000 (15:13 +0000)]
[project @ 2005-03-05 15:13:01 by panne]
Warning police again: Use the "official" hs_free_stable_ptr from
HsFFI.h instead of the internal freeStablePtr, avoiding

   implicit declaration of function `freeStablePtr'

warnings from gcc when compiling via C.

19 years ago[project @ 2005-03-04 18:26:48 by sof]
sof [Fri, 4 Mar 2005 18:26:48 +0000 (18:26 +0000)]
[project @ 2005-03-04 18:26:48 by sof]
Temper 'libm' testing -- if 'atan' is available straight from libc,
no need to include libm.

Merge to STABLE

19 years ago[project @ 2005-03-03 05:11:41 by chak]
chak [Thu, 3 Mar 2005 05:11:41 +0000 (05:11 +0000)]
[project @ 2005-03-03 05:11:41 by chak]
Merge to STABLE

* Fixed two bugs reported on glasgow-haskell-users

19 years ago[project @ 2005-03-02 16:39:56 by ross]
ross [Wed, 2 Mar 2005 16:39:57 +0000 (16:39 +0000)]
[project @ 2005-03-02 16:39:56 by ross]
*Config.h files are in include/ (MERGE to STABLE)

19 years ago[project @ 2005-03-02 14:46:14 by simonmar]
simonmar [Wed, 2 Mar 2005 14:46:14 +0000 (14:46 +0000)]
[project @ 2005-03-02 14:46:14 by simonmar]
distcleaning of things generated by configure

19 years ago[project @ 2005-03-02 13:11:00 by simonmar]
simonmar [Wed, 2 Mar 2005 13:11:00 +0000 (13:11 +0000)]
[project @ 2005-03-02 13:11:00 by simonmar]
We should not assume that the timeout parameter to select() is updated
with the time left over after select() returns.  Linux does this, but
FreeBSD does not.

Fixes -threaded hangs on FreeBSD.

19 years ago[project @ 2005-02-26 12:14:54 by panne]
panne [Sat, 26 Feb 2005 12:14:56 +0000 (12:14 +0000)]
[project @ 2005-02-26 12:14:54 by panne]
Moved Monoid instances of collection types to Data.Monoid, concentrating non-H98
stuff to a single place.

19 years ago[project @ 2005-02-25 10:42:24 by simonmar]
simonmar [Fri, 25 Feb 2005 10:42:24 +0000 (10:42 +0000)]
[project @ 2005-02-25 10:42:24 by simonmar]
Add instance Typeable Queue

19 years ago[project @ 2005-02-24 09:58:23 by simonmar]
simonmar [Thu, 24 Feb 2005 09:58:24 +0000 (09:58 +0000)]
[project @ 2005-02-24 09:58:23 by simonmar]
nDoc fixes from Sven Panne.  Generally fixing of Haddock links, adding
some signatures, and in some cases exporting type constructors that
are mentioned in the types of exported identifiers.

19 years ago[project @ 2005-02-23 06:31:22 by dons]
dons [Wed, 23 Feb 2005 06:31:22 +0000 (06:31 +0000)]
[project @ 2005-02-23 06:31:22 by dons]
Typo in comment only. Spotted by sjanssen on #haskell.

19 years ago[project @ 2005-02-21 11:36:07 by simonmar]
simonmar [Mon, 21 Feb 2005 11:36:07 +0000 (11:36 +0000)]
[project @ 2005-02-21 11:36:07 by simonmar]
docs only: clarify language in a couple of places.

From: Paul Steckler [paul.steckler.ctr@metnet.navy.mil]

19 years ago[project @ 2005-02-18 18:30:40 by ross]
ross [Fri, 18 Feb 2005 18:30:40 +0000 (18:30 +0000)]
[project @ 2005-02-18 18:30:40 by ross]
Rename package description fields as in InstalledPackageInfo:

options-ghc -> ghc-options
options-hugs -> hugs-options
options-nhc -> nhc-options
extra-libs -> extra-libraries

19 years ago[project @ 2005-02-18 15:06:45 by simonmar]
simonmar [Fri, 18 Feb 2005 15:06:45 +0000 (15:06 +0000)]
[project @ 2005-02-18 15:06:45 by simonmar]
Rename fields in InstalledPackageInfo for consistency with
PackageDescription & buildInfo:

 extra-libs (extraLibraries) --> extra-libraries (extraLibraries)
 extra-cc-opts (extraCcOpts) --> cc-options (ccOptions)
 extra-ld-opts (extraLdOpts) --> ld-options (ldOptions)
 extra-hugs-opts (extraHugsOpts) --> hugs-options (hugsOptions)
 extra-frameworks (extraFrameworks) --> frameworks (frameworks)

19 years ago[project @ 2005-02-15 08:09:43 by ross]
ross [Tue, 15 Feb 2005 08:09:43 +0000 (08:09 +0000)]
[project @ 2005-02-15 08:09:43 by ross]
Hugs only: use binary handles for copyFile

19 years ago[project @ 2005-02-13 10:53:13 by malcolm]
malcolm [Sun, 13 Feb 2005 10:53:13 +0000 (10:53 +0000)]
[project @ 2005-02-13 10:53:13 by malcolm]
Eliminate more explicit dependencies in Makefiles: use hmake to
create the bootstrapping .hc files, as well as the ordinary .o files.

19 years ago[project @ 2005-02-11 11:36:23 by simonmar]
simonmar [Fri, 11 Feb 2005 11:36:23 +0000 (11:36 +0000)]
[project @ 2005-02-11 11:36:23 by simonmar]
Add bracketOnError

19 years ago[project @ 2005-02-11 01:55:56 by ross]
ross [Fri, 11 Feb 2005 01:55:56 +0000 (01:55 +0000)]
[project @ 2005-02-11 01:55:56 by ross]
track syntax changes:

* add License-File and Synopsis fields

* rename Hidden-Fields as Other-Fields

These files are used only by Hugs, but are also useful as examples.

19 years ago[project @ 2005-02-07 15:26:10 by malcolm]
malcolm [Mon, 7 Feb 2005 15:26:10 +0000 (15:26 +0000)]
[project @ 2005-02-07 15:26:10 by malcolm]
Place imports before #includes, just to reduce the number of 'file
not found' warnings from hmake.

19 years ago[project @ 2005-02-07 12:21:29 by simonmar]
simonmar [Mon, 7 Feb 2005 12:21:29 +0000 (12:21 +0000)]
[project @ 2005-02-07 12:21:29 by simonmar]
After no response on libraries@haskell.org... John Meacham's
Data.Graph patch, which returns an extra component from
graphFromEdges.  The old version of graphFromEdges is available as
graphFromEdges'.

19 years ago[project @ 2005-02-07 09:56:42 by ross]
ross [Mon, 7 Feb 2005 09:56:43 +0000 (09:56 +0000)]
[project @ 2005-02-07 09:56:42 by ross]
a few docs (for STABLE)

19 years ago[project @ 2005-02-05 00:41:35 by ross]
ross [Sat, 5 Feb 2005 00:41:37 +0000 (00:41 +0000)]
[project @ 2005-02-05 00:41:35 by ross]
more Haddock fixes.

Now the only dangling links are in System.Directory, referring to the
hidden module GHC.IOBase for constructors of IOException.

19 years ago[project @ 2005-02-04 14:36:52 by simonmar]
simonmar [Fri, 4 Feb 2005 14:36:52 +0000 (14:36 +0000)]
[project @ 2005-02-04 14:36:52 by simonmar]
Add a comment to Ross's previous commit (sorry, forgot to commit my
version of that change earlier).

19 years ago[project @ 2005-02-04 14:20:57 by ross]
ross [Fri, 4 Feb 2005 14:20:57 +0000 (14:20 +0000)]
[project @ 2005-02-04 14:20:57 by ross]
if this can't be hidden, at least make it not-home

19 years ago[project @ 2005-02-03 10:38:44 by simonmar]
simonmar [Thu, 3 Feb 2005 10:38:45 +0000 (10:38 +0000)]
[project @ 2005-02-03 10:38:44 by simonmar]
unhide a few modules

19 years ago[project @ 2005-02-03 10:32:11 by ross]
ross [Thu, 3 Feb 2005 10:32:20 +0000 (10:32 +0000)]
[project @ 2005-02-03 10:32:11 by ross]
hide GHC internals from Haddock

19 years ago[project @ 2005-02-02 15:28:49 by simonmar]
simonmar [Wed, 2 Feb 2005 15:30:09 +0000 (15:30 +0000)]
[project @ 2005-02-02 15:28:49 by simonmar]
doc fix

19 years ago[project @ 2005-02-02 15:23:59 by simonmar]
simonmar [Wed, 2 Feb 2005 15:23:59 +0000 (15:23 +0000)]
[project @ 2005-02-02 15:23:59 by simonmar]
doc fixes

19 years ago[project @ 2005-02-02 15:22:54 by simonmar]
simonmar [Wed, 2 Feb 2005 15:22:54 +0000 (15:22 +0000)]
[project @ 2005-02-02 15:22:54 by simonmar]
Doc fix

19 years ago[project @ 2005-02-02 15:22:19 by simonmar]
simonmar [Wed, 2 Feb 2005 15:22:19 +0000 (15:22 +0000)]
[project @ 2005-02-02 15:22:19 by simonmar]
Doc fixes

19 years ago[project @ 2005-02-02 15:21:02 by simonmar]
simonmar [Wed, 2 Feb 2005 15:21:02 +0000 (15:21 +0000)]
[project @ 2005-02-02 15:21:02 by simonmar]
doc fixes

19 years ago[project @ 2005-02-02 15:20:11 by simonmar]
simonmar [Wed, 2 Feb 2005 15:20:11 +0000 (15:20 +0000)]
[project @ 2005-02-02 15:20:11 by simonmar]
tiny doc fix

19 years ago[project @ 2005-02-02 14:54:18 by ross]
ross [Wed, 2 Feb 2005 14:54:19 +0000 (14:54 +0000)]
[project @ 2005-02-02 14:54:18 by ross]
an instance for FunPtr, and minor Hugs fixes:

* for Hugs only, move the instances for ST, STRef and STArray back.
  Having them in Data.Typeable, which is imported by Data.Dynamic,
  would mean that every invocation of runhugs or ffihugs would need
  the -98 option.

* Hugs also has MVar and the exception types.

and NHC has ForeignPtr too.

19 years ago[project @ 2005-02-02 13:47:24 by simonpj]
simonpj [Wed, 2 Feb 2005 13:47:24 +0000 (13:47 +0000)]
[project @ 2005-02-02 13:47:24 by simonpj]
Generalise gfindtype to

gfindtype :: (Data x, Typeable y) => x -> Maybe y

(It was (Data x, Data y) => ..., but there's no reason to
require Data y.)

Pointed out by Jim Apple

19 years ago[project @ 2005-02-02 13:45:05 by malcolm]
malcolm [Wed, 2 Feb 2005 13:45:05 +0000 (13:45 +0000)]
[project @ 2005-02-02 13:45:05 by malcolm]
nhc98 has Data.Typeable.

19 years ago[project @ 2005-02-02 13:26:13 by simonpj]
simonpj [Wed, 2 Feb 2005 13:26:20 +0000 (13:26 +0000)]
[project @ 2005-02-02 13:26:13 by simonpj]
I've moved Typeable instances so that they are
either in the module that defines the type
or in the Typeable module (which defines the class)

GHC dislikes "orphan" instances, and even for humans
this makes it easier to find.

I have continued to use the INSTANCE_TYPEABLE macros,
rather than GHC's deriving( Typeable ) mechanism, so
that it'll still work for Hugs and NHC.  Nevertheless,
I may well have missed some Hugs- or NHC-specific imports,
for which I apologise.  Malcolm, Ross you may want to try
a fresh build.

19 years ago[project @ 2005-02-02 10:59:16 by malcolm]
malcolm [Wed, 2 Feb 2005 10:59:16 +0000 (10:59 +0000)]
[project @ 2005-02-02 10:59:16 by malcolm]
Build all package sources using hmake.  The inaccurate and ever-changing
Makefile dependencies can now all be thrown away, hurrah!

19 years ago[project @ 2005-02-01 17:32:19 by ross]
ross [Tue, 1 Feb 2005 17:32:19 +0000 (17:32 +0000)]
[project @ 2005-02-01 17:32:19 by ross]
docs

19 years ago[project @ 2005-02-01 16:47:27 by malcolm]
malcolm [Tue, 1 Feb 2005 16:47:28 +0000 (16:47 +0000)]
[project @ 2005-02-01 16:47:27 by malcolm]
Make the cpp directives in NHC.SizedTypes directly usable by
nhc98 with internal cpphs, avoiding ANSI-only string-pasting.

19 years ago[project @ 2005-02-01 13:41:41 by simonmar]
simonmar [Tue, 1 Feb 2005 13:41:41 +0000 (13:41 +0000)]
[project @ 2005-02-01 13:41:41 by simonmar]
Move #include of HsBaseConfig.h up

19 years ago[project @ 2005-02-01 13:02:37 by simonmar]
simonmar [Tue, 1 Feb 2005 13:02:37 +0000 (13:02 +0000)]
[project @ 2005-02-01 13:02:37 by simonmar]
Make hDuplicateTo actually use dup2() rather than dup().  The
difference is noticeable if you want to eg. redirect stdout and then
use executeFile or spawn sub-processes.

19 years ago[project @ 2005-02-01 11:52:08 by malcolm]
malcolm [Tue, 1 Feb 2005 11:52:08 +0000 (11:52 +0000)]
[project @ 2005-02-01 11:52:08 by malcolm]
Record more dependencies.

19 years ago[project @ 2005-02-01 10:12:16 by krasimir]
krasimir [Tue, 1 Feb 2005 10:12:17 +0000 (10:12 +0000)]
[project @ 2005-02-01 10:12:16 by krasimir]
Fixes for Windows

19 years ago[project @ 2005-02-01 00:52:20 by ross]
ross [Tue, 1 Feb 2005 00:52:22 +0000 (00:52 +0000)]
[project @ 2005-02-01 00:52:20 by ross]
more regex test down to libraries/base

19 years ago[project @ 2005-01-31 21:07:15 by panne]
panne [Mon, 31 Jan 2005 21:07:15 +0000 (21:07 +0000)]
[project @ 2005-01-31 21:07:15 by panne]
Documentation fix only ("in" is a keyword)

19 years ago[project @ 2005-01-31 19:54:22 by panne]
panne [Mon, 31 Jan 2005 19:54:22 +0000 (19:54 +0000)]
[project @ 2005-01-31 19:54:22 by panne]
Ooops, my evil XEmacs inserted an extraneous character... :-}

19 years ago[project @ 2005-01-31 19:28:42 by panne]
panne [Mon, 31 Jan 2005 19:28:42 +0000 (19:28 +0000)]
[project @ 2005-01-31 19:28:42 by panne]
* Unbreak System.Info for Hugs, rearranging the whole module a bit.

* Platform-specific exports suck, so even Hugs has a compilerVersion
  now, although it is currently always version 0.0 :-)

Tested with Hugs only...

19 years ago[project @ 2005-01-31 18:33:48 by ross]
ross [Mon, 31 Jan 2005 18:33:48 +0000 (18:33 +0000)]
[project @ 2005-01-31 18:33:48 by ross]
fix for non-GHC

19 years ago[project @ 2005-01-31 13:52:26 by malcolm]
malcolm [Mon, 31 Jan 2005 13:52:26 +0000 (13:52 +0000)]
[project @ 2005-01-31 13:52:26 by malcolm]
Make it compile for non-GHC.

19 years ago[project @ 2005-01-31 13:51:22 by simonmar]
simonmar [Mon, 31 Jan 2005 13:51:24 +0000 (13:51 +0000)]
[project @ 2005-01-31 13:51:22 by simonmar]
Some improvements to System.Cmd.{system,rawSystem} on Un*x systems:
these commands now do the appropriate signal handling, namely ignoring
SIGINT/SIGQUIT in the parent but allowing these signals in the child.
This behaviour matches the Un*x system().

What this means is that when System.Cmd.system is executing, ^C only
kills the child process, the parent will see an exception.

19 years ago[project @ 2005-01-31 13:46:24 by simonmar]
simonmar [Mon, 31 Jan 2005 13:46:24 +0000 (13:46 +0000)]
[project @ 2005-01-31 13:46:24 by simonmar]
oops, remove debugging modification

19 years ago[project @ 2005-01-31 13:45:50 by malcolm]
malcolm [Mon, 31 Jan 2005 13:45:50 +0000 (13:45 +0000)]
[project @ 2005-01-31 13:45:50 by malcolm]
Work around type-system bug (cxt in lhs pattern) in nhc98.

19 years ago[project @ 2005-01-31 12:57:26 by simonmar]
simonmar [Mon, 31 Jan 2005 12:57:27 +0000 (12:57 +0000)]
[project @ 2005-01-31 12:57:26 by simonmar]
Cleanup: convert System/Posix/Signals.hsc into a plain .hs file, and use the
configure script to get the appropriate constants.

19 years ago[project @ 2005-01-29 16:10:27 by wolfgang]
wolfgang [Sat, 29 Jan 2005 16:10:27 +0000 (16:10 +0000)]
[project @ 2005-01-29 16:10:27 by wolfgang]
import CString on non-windows platforms.

19 years ago[project @ 2005-01-28 23:33:57 by krasimir]
krasimir [Fri, 28 Jan 2005 23:33:58 +0000 (23:33 +0000)]
[project @ 2005-01-28 23:33:57 by krasimir]
- The output from uncaught exceptions handler is redirected to RTS's errorBelch.
- The output from Debug.Trace is redirected to RTS's debugBelch
- Usually errorBelch and debugBelch messages go to stderr except for
Windows GUI applications. For GUI applications the Debug.Trace output is
redirected to debug console and the exceptions message is displayed in message box.

19 years ago[project @ 2005-01-28 16:09:06 by malcolm]
malcolm [Fri, 28 Jan 2005 16:09:06 +0000 (16:09 +0000)]
[project @ 2005-01-28 16:09:06 by malcolm]
Add compilerVersion for nhc98.

19 years ago[project @ 2005-01-28 15:03:06 by simonmar]
simonmar [Fri, 28 Jan 2005 15:03:08 +0000 (15:03 +0000)]
[project @ 2005-01-28 15:03:06 by simonmar]
Remove some unnecessary #includes of ghcconfig.h

19 years ago[project @ 2005-01-28 14:55:05 by simonmar]
simonmar [Fri, 28 Jan 2005 14:55:05 +0000 (14:55 +0000)]
[project @ 2005-01-28 14:55:05 by simonmar]
Remove unnecessary ghcconfig.h include

19 years ago[project @ 2005-01-28 13:54:56 by simonmar]
simonmar [Fri, 28 Jan 2005 13:54:56 +0000 (13:54 +0000)]
[project @ 2005-01-28 13:54:56 by simonmar]
System.Info now exports:

  compilerName :: String
  compilerVersion :: Data.Version.Version

as requested ages ago by Template Haskell users.  This allows
"conditional compilation" based on the GHC version in TH code.

I've implemented compilerName for all compilers (I hope).  Other
compiler vendors are encouraged to provide their own implementations
of compilerVersion.

19 years ago[project @ 2005-01-28 13:36:25 by simonmar]
simonmar [Fri, 28 Jan 2005 13:36:34 +0000 (13:36 +0000)]
[project @ 2005-01-28 13:36:25 by simonmar]
Catch up with updates to platform #defines.

Generally: use _HOST_ rather than _TARGET_ (except in Cabal where we
have to retain compatibility with previous GHC versions).

19 years ago[project @ 2005-01-28 10:15:44 by ross]
ross [Fri, 28 Jan 2005 10:15:44 +0000 (10:15 +0000)]
[project @ 2005-01-28 10:15:44 by ross]
fix foreign imports for non-GHC

19 years ago[project @ 2005-01-27 14:38:05 by ross]
ross [Thu, 27 Jan 2005 14:38:05 +0000 (14:38 +0000)]
[project @ 2005-01-27 14:38:05 by ross]
doc tweaks

19 years ago[project @ 2005-01-27 14:19:50 by simonpj]
simonpj [Thu, 27 Jan 2005 14:19:50 +0000 (14:19 +0000)]
[project @ 2005-01-27 14:19:50 by simonpj]
Fix regex.h include stuff; Ross and Malcolm might want to look at the new comment

19 years ago[project @ 2005-01-27 10:46:19 by malcolm]
malcolm [Thu, 27 Jan 2005 10:46:19 +0000 (10:46 +0000)]
[project @ 2005-01-27 10:46:19 by malcolm]
Hook up System.Directory.Internals.

19 years ago[project @ 2005-01-27 10:45:47 by simonpj]
simonpj [Thu, 27 Jan 2005 10:45:48 +0000 (10:45 +0000)]
[project @ 2005-01-27 10:45:47 by simonpj]
--------------------------------------------
          Replace hi-boot files with hs-boot files
   --------------------------------------------

This major commit completely re-organises the way that recursive modules
are dealt with.

  * It should have NO EFFECT if you do not use recursive modules

  * It is a BREAKING CHANGE if you do

====== Warning: .hi-file format has changed, so if you are
====== updating into an existing HEAD build, you'll
====== need to make clean and re-make

The details:  [documentation still to be done]

* Recursive loops are now broken with Foo.hs-boot (or Foo.lhs-boot),
  not Foo.hi-boot

* An hs-boot files is a proper source file.  It is compiled just like
  a regular Haskell source file:
ghc Foo.hs generates Foo.hi, Foo.o
ghc Foo.hs-boot generates Foo.hi-boot, Foo.o-boot

* hs-boot files are precisely a subset of Haskell. In particular:
- they have the same import, export, and scoping rules
- errors (such as kind errors) in hs-boot files are checked
  You do *not* need to mention the "original" name of something in
  an hs-boot file, any more than you do in any other Haskell module.

* The Foo.hi-boot file generated by compiling Foo.hs-boot is a machine-
  generated interface file, in precisely the same format as Foo.hi

* When compiling Foo.hs, its exports are checked for compatibility with
  Foo.hi-boot (previously generated by compiling Foo.hs-boot)

* The dependency analyser (ghc -M) knows about Foo.hs-boot files, and
  generates appropriate dependencies.  For regular source files it
  generates
Foo.o : Foo.hs
Foo.o : Baz.hi -- Foo.hs imports Baz
Foo.o : Bog.hi-boot -- Foo.hs source-imports Bog

  For a hs-boot file it generates similar dependencies
Bog.o-boot : Bog.hs-boot
Bog.o-boot : Nib.hi -- Bog.hs-boto imports Nib

* ghc -M is also enhanced to use the compilation manager dependency
  chasing, so that
ghc -M Main
  will usually do the job.  No need to enumerate all the source files.

* The -c flag is no longer a "compiler mode". It simply means "omit the
  link step", and synonymous with -no-link.

19 years ago[project @ 2005-01-26 15:41:01 by simonmar]
simonmar [Wed, 26 Jan 2005 15:41:01 +0000 (15:41 +0000)]
[project @ 2005-01-26 15:41:01 by simonmar]
Add System.Directory to $(ALL_DIRS)

19 years ago[project @ 2005-01-26 15:40:37 by simonmar]
simonmar [Wed, 26 Jan 2005 15:40:37 +0000 (15:40 +0000)]
[project @ 2005-01-26 15:40:37 by simonmar]
remove System.FilePath; add System.Directory.Internals.

19 years ago[project @ 2005-01-26 15:19:19 by malcolm]
malcolm [Wed, 26 Jan 2005 15:19:19 +0000 (15:19 +0000)]
[project @ 2005-01-26 15:19:19 by malcolm]
Remove System.FilePath.

19 years ago[project @ 2005-01-26 14:55:41 by simonmar]
simonmar [Wed, 26 Jan 2005 14:55:44 +0000 (14:55 +0000)]
[project @ 2005-01-26 14:55:41 by simonmar]
Remove System.FilePath pending a redesign of the interface.
Temporarily introduce System.Directory.Internals as a home for some of
the bits of System.FilePath we were already using elsewhere.

19 years ago[project @ 2005-01-26 14:36:42 by malcolm]
malcolm [Wed, 26 Jan 2005 14:36:42 +0000 (14:36 +0000)]
[project @ 2005-01-26 14:36:42 by malcolm]
Patch from John Meacham:
Generalise a couple functions in Data.Map to be usable in an arbitrary
monad (rather than being restricted to Maybe).

19 years ago[project @ 2005-01-26 14:33:29 by malcolm]
malcolm [Wed, 26 Jan 2005 14:33:29 +0000 (14:33 +0000)]
[project @ 2005-01-26 14:33:29 by malcolm]
Fix module name.

19 years ago[project @ 2005-01-26 13:27:44 by malcolm]
malcolm [Wed, 26 Jan 2005 13:27:44 +0000 (13:27 +0000)]
[project @ 2005-01-26 13:27:44 by malcolm]
Change stability notation from 'stable' to 'experimental', since many
people seem to agree there are plenty of bugs here.

19 years ago[project @ 2005-01-26 12:23:33 by simonmar]
simonmar [Wed, 26 Jan 2005 12:23:33 +0000 (12:23 +0000)]
[project @ 2005-01-26 12:23:33 by simonmar]
I think InvalidArgument is more useful than OtherError for EBADF.

19 years ago[project @ 2005-01-25 22:17:37 by krasimir]
krasimir [Tue, 25 Jan 2005 22:17:37 +0000 (22:17 +0000)]
[project @ 2005-01-25 22:17:37 by krasimir]
Fixed bug with splitFileExt "foo.bar." and splitFileName "foo:bar"

19 years ago[project @ 2005-01-25 17:06:39 by ross]
ross [Tue, 25 Jan 2005 17:06:39 +0000 (17:06 +0000)]
[project @ 2005-01-25 17:06:39 by ross]
add Cabal package descriptions

19 years ago[project @ 2005-01-24 04:26:17 by wolfgang]
wolfgang [Mon, 24 Jan 2005 04:26:17 +0000 (04:26 +0000)]
[project @ 2005-01-24 04:26:17 by wolfgang]
Always flush stdout & stderr at exit, even when terminating abnormally
due to an exception.

19 years ago[project @ 2005-01-23 20:20:55 by wolfgang]
wolfgang [Sun, 23 Jan 2005 20:20:55 +0000 (20:20 +0000)]
[project @ 2005-01-23 20:20:55 by wolfgang]
Flush stdout and stderr also when exiting due to an ExitException.

19 years ago[project @ 2005-01-23 13:48:19 by panne]
panne [Sun, 23 Jan 2005 13:48:19 +0000 (13:48 +0000)]
[project @ 2005-01-23 13:48:19 by panne]
Added Isaac's getOpt variant (named getOpt' now), so System.Console.GetOpt
and Distribution.GetOpt are identical now.

19 years ago[project @ 2005-01-23 13:12:52 by panne]
panne [Sun, 23 Jan 2005 13:12:52 +0000 (13:12 +0000)]
[project @ 2005-01-23 13:12:52 by panne]
Unify the base and Cabal versions of GetOpt a bit.

19 years ago[project @ 2005-01-23 11:30:39 by panne]
panne [Sun, 23 Jan 2005 11:30:39 +0000 (11:30 +0000)]
[project @ 2005-01-23 11:30:39 by panne]
* Nuked one #ifdef in isPathSeparator, the generated code is the same
* Added a ToDo for searchPathSeparator

19 years ago[project @ 2005-01-22 12:18:49 by panne]
panne [Sat, 22 Jan 2005 12:18:49 +0000 (12:18 +0000)]
[project @ 2005-01-22 12:18:49 by panne]
Doc change only: I don't think that toList should give the guarantee
that the resulting list is in ascending order...

19 years ago[project @ 2005-01-21 21:52:56 by panne]
panne [Fri, 21 Jan 2005 21:52:56 +0000 (21:52 +0000)]
[project @ 2005-01-21 21:52:56 by panne]
*sigh* Once again: Fixed a Haddock comment which broke "make html"... Why
on earth do we have "make validate"?? :ยด-( I think we should introduce some
kind of disciplinary punishment for checking in broken Haddock comments,
like porting the mangler to a new platform or porting Adjustor.c to an ABI
more irregular than the one on PowerPC...

19 years ago[project @ 2005-01-21 19:59:01 by sof]
sof [Fri, 21 Jan 2005 19:59:01 +0000 (19:59 +0000)]
[project @ 2005-01-21 19:59:01 by sof]
win32 only:
  Tidy up delivery and handling of console events by having the low-level
  console event handler signal the RTS thread blocked waiting for I/O.

19 years ago[project @ 2005-01-21 16:02:47 by simonmar]
simonmar [Fri, 21 Jan 2005 16:02:48 +0000 (16:02 +0000)]
[project @ 2005-01-21 16:02:47 by simonmar]
Don't try to run finalizers at program exit.  This turned out to be
hard if not impossible to do in general, so now we don't attempt it at
all.

The Main.main wrapper, previously called runIO and now called
runMainIO, flushes stdout and stderr before exiting.  This should
catch most cases where programs rely on Handles being flushed at
program exit, but note that now if you simply drop a Handle in your
program, there's no guarantee it'll be flushed on exit.  If the
punters complain enough, I suppose we could implement a global
Handle table and flush them all at exit... I'd rather not do this if
possible, though.  Better to teach people to close their Handles
properly.

19 years ago[project @ 2005-01-21 15:12:21 by simonmar]
simonmar [Fri, 21 Jan 2005 15:12:21 +0000 (15:12 +0000)]
[project @ 2005-01-21 15:12:21 by simonmar]
hClose on stdin,stdout,stderr now actually closes the file
descriptor.  Before, there was no way to actually close these file
descriptors, which might be necessary in some cases - especially when
stdin/stdout are pipes and you need to indicate to the other end of
the pipe that you've finished I/O, but without exiting the program.

19 years ago[project @ 2005-01-21 11:44:08 by ross]
ross [Fri, 21 Jan 2005 11:44:09 +0000 (11:44 +0000)]
[project @ 2005-01-21 11:44:08 by ross]
alter the interface of splitLookup and splitMember, placing the match
between the trees of smaller and larger elements in the returned triple.

19 years ago[project @ 2005-01-21 10:52:47 by ross]
ross [Fri, 21 Jan 2005 10:52:47 +0000 (10:52 +0000)]
[project @ 2005-01-21 10:52:47 by ross]
refine docs a bit

19 years ago[project @ 2005-01-21 09:55:52 by malcolm]
malcolm [Fri, 21 Jan 2005 09:55:52 +0000 (09:55 +0000)]
[project @ 2005-01-21 09:55:52 by malcolm]
Typo

19 years ago[project @ 2005-01-20 19:00:26 by ross]
ross [Thu, 20 Jan 2005 19:00:28 +0000 (19:00 +0000)]
[project @ 2005-01-20 19:00:26 by ross]
cross-references and minor doc fixes

19 years ago[project @ 2005-01-20 14:39:09 by malcolm]
malcolm [Thu, 20 Jan 2005 14:39:09 +0000 (14:39 +0000)]
[project @ 2005-01-20 14:39:09 by malcolm]
Lots more new dependencies.

19 years ago[project @ 2005-01-20 14:22:26 by simonmar]
simonmar [Thu, 20 Jan 2005 14:22:26 +0000 (14:22 +0000)]
[project @ 2005-01-20 14:22:26 by simonmar]
Fill in the haddock-interfaces and haddock-html fields in the
package.conf files.

To do this I had to make some changes:

  - haddock-interfaces requires the value of $(datadir).  We can't
    just plug this in, because $(datadir) might change at install-time
    (eg. a Windows installer can be placed anywhere, as can a Unix
    binary .tar.gz distribution).  The current trick is for the
    compiler to splice in the value of $libdir in package.conf at
    runtime.  So we could extend this mechanism and tell the compiler
    the value of $datadir via a command-line option, but that seems
    ugly.

    On Windows, $datadir==$libdir, so we don't need any changes:
    package.conf still uses $libdir, and a Windows installation is
    independent of its absolute location.  Even 'make install' on
    Windows should have this property.

    On Unix:
- for 'make install' and in-place execution, we just use
          absolute paths in package.conf

- for a binary dist, we generate a package.conf that refers
  to $libdir and $datadir, and splice in the values at
  install-time (distrib/Makefile-bin.in).

  - Also, I renamed $libdir to $topdir to more closely reflect its
    actual meaning.  This is somewhat malicious in that it will flush
    out all those clients using $libdir when they really shouldn't
    be :-)

19 years ago[project @ 2005-01-20 11:09:54 by malcolm]
malcolm [Thu, 20 Jan 2005 11:09:54 +0000 (11:09 +0000)]
[project @ 2005-01-20 11:09:54 by malcolm]
Hook up Data.FunctorM, and update dependencies from Data.Set, Data.IntMap,
Data.Map etc, on Data.Typeable.

19 years ago[project @ 2005-01-20 10:36:43 by malcolm]
malcolm [Thu, 20 Jan 2005 10:36:43 +0000 (10:36 +0000)]
[project @ 2005-01-20 10:36:43 by malcolm]
Include Typeable instance for nhc98.

19 years ago[project @ 2005-01-19 23:33:25 by ross]
ross [Wed, 19 Jan 2005 23:33:26 +0000 (23:33 +0000)]
[project @ 2005-01-19 23:33:25 by ross]
Move comments inside #if's they refer to -- makes the cpp output, e.g. as
seen by Hugs users, more sensible.

19 years ago[project @ 2005-01-19 23:32:04 by ross]
ross [Wed, 19 Jan 2005 23:32:04 +0000 (23:32 +0000)]
[project @ 2005-01-19 23:32:04 by ross]
fix name in Typeable instance

19 years ago[project @ 2005-01-19 22:33:32 by ralf]
ralf [Wed, 19 Jan 2005 22:33:36 +0000 (22:33 +0000)]
[project @ 2005-01-19 22:33:32 by ralf]
Added quite a few more Data instances.

For these datatypes:

- Data.Array
- Data.FiniteMap
- Data.IntMap
- Data.IntSet
- Data.Map
- Data.Set
- GHC.ST
- GHC.ForeignPtr
- Control.Concurrent.MVar and friends

(In some cases, this also required adding Typeable instances.)

Most or all additions have been tagged by
#ifdef __GLASGOW_HASKELL__ ...

For the _abstract_ datatypes in the above list,
we use the from... and to... projections and injections
so that we can process the containers or whatever as lists.
We don't provide the reflection API for abstract datatypes.

If anyone sees Data-wanting datatypes in the base or elsewhere
in the module space, please let me know, or please go ahead,
and add coverage for SYB. As far as I understand, the base
code cannot use "deriving (Typeable, Data)", whereas this
is very well an option for non-base modules. Indeed it is
readily done like this in:
- network/Network/URI
- haskell-src