ghc-hetmet.git
20 years ago[project @ 2004-03-23 10:48:41 by simonmar]
simonmar [Tue, 23 Mar 2004 10:48:41 +0000 (10:48 +0000)]
[project @ 2004-03-23 10:48:41 by simonmar]
Update docs for --enable-threaded-rts (now deprecated).

20 years ago[project @ 2004-03-23 10:04:18 by simonmar]
simonmar [Tue, 23 Mar 2004 10:04:18 +0000 (10:04 +0000)]
[project @ 2004-03-23 10:04:18 by simonmar]
Give the malloc functiosn the __malloc__ attribute, if available.
Helps with aliasing optimisations, apparently.

Submitted by: Duncan Coutts <duncan.coutts@worcester.oxford.ac.uk>

20 years ago[project @ 2004-03-23 10:03:18 by simonmar]
simonmar [Tue, 23 Mar 2004 10:03:18 +0000 (10:03 +0000)]
[project @ 2004-03-23 10:03:18 by simonmar]
Add GNUC3_ATTRIBUTE() macro

Submitted by: Duncan Coutts <duncan.coutts@worcester.oxford.ac.uk>

20 years ago[project @ 2004-03-23 09:15:54 by ralf]
ralf [Tue, 23 Mar 2004 09:15:54 +0000 (09:15 +0000)]
[project @ 2004-03-23 09:15:54 by ralf]
Version 6.3 is encoded as 603 not 630.
GHC builds fine again through both stages on my system.
Ralf

20 years ago[project @ 2004-03-22 11:55:39 by simonmar]
simonmar [Mon, 22 Mar 2004 11:55:39 +0000 (11:55 +0000)]
[project @ 2004-03-22 11:55:39 by simonmar]
Typo

20 years ago[project @ 2004-03-22 11:48:30 by simonmar]
simonmar [Mon, 22 Mar 2004 11:48:30 +0000 (11:48 +0000)]
[project @ 2004-03-22 11:48:30 by simonmar]
change this message:

   <prog>: GHC not built for -prof

to

   <prog>: not built for -prof

Suggested by: Sven Panne.

20 years ago[project @ 2004-03-22 11:06:18 by simonmar]
simonmar [Mon, 22 Mar 2004 11:06:18 +0000 (11:06 +0000)]
[project @ 2004-03-22 11:06:18 by simonmar]
Fix uninitialised ref.

Submitted by: Julian Seward (who else!)

20 years ago[project @ 2004-03-22 09:00:28 by ralf]
ralf [Mon, 22 Mar 2004 09:00:28 +0000 (09:00 +0000)]
[project @ 2004-03-22 09:00:28 by ralf]
Boot-strapping ghc revealed an issue related to yesterday's renaming
of Data.Typeable. I can now build a stage1 compiler again. Let me try
a stage2 one as well.

ralf        2004/03/21 11:06:04 PST

  Modified files:
    ghc/compiler/prelude PrelNames.lhs
    ghc/compiler/typecheck TcGenDeriv.lhs
  Log:
  Implemented renaming for Data.Typeable
  as far as deriving mechanism is involved.
  See http://www.haskell.org//pipermail/libraries/2004-March/001846.html

  Revision  Changes    Path
  1.88      +1 -1      fptools/ghc/compiler/prelude/PrelNames.lhs
  1.107     +1 -1      fptools/ghc/compiler/typecheck/TcGenDeriv.lhs

20 years ago[project @ 2004-03-21 19:06:01 by ralf]
ralf [Sun, 21 Mar 2004 19:06:03 +0000 (19:06 +0000)]
[project @ 2004-03-21 19:06:01 by ralf]
Implemented renaming for Data.Typeable
as far as deriving mechanism is involved.
See http://www.haskell.org//pipermail/libraries/2004-March/001846.html

20 years ago[project @ 2004-03-19 23:20:20 by panne]
panne [Fri, 19 Mar 2004 23:20:20 +0000 (23:20 +0000)]
[project @ 2004-03-19 23:20:20 by panne]
Fixed  the JOB line in heap  profiles, it contained superfluous spaces
and an evil line break.

Merge to STABLE
(This fix looks quite right, but again, I leave this to the Master of
Releases (tm), because there might already be tools depending on the
slightly wrong old format.)

20 years ago[project @ 2004-03-19 23:17:06 by panne]
panne [Fri, 19 Mar 2004 23:17:06 +0000 (23:17 +0000)]
[project @ 2004-03-19 23:17:06 by panne]
Restrict the setlocale() to LC_CTYPE (was LC_ALL), because printf and friends
are LC_NUMERIC-sensitive. This fixes the generation of heap profiles for e.g.
de_DE locales where a comma separates the fractional part of a number, not a
dot. (There was a similar bug in Hugs regarding atof, BTW, which was fixed last
December.)

Merge to STABLE
(I leave this to the Grand Marshal of Releases, Mr. SimonM, because I'm not 100%
sure of all the consequences.)

20 years ago[project @ 2004-03-19 13:25:24 by simonpj]
simonpj [Fri, 19 Mar 2004 13:25:24 +0000 (13:25 +0000)]
[project @ 2004-03-19 13:25:24 by simonpj]
Add comments

20 years ago[project @ 2004-03-19 13:24:52 by simonpj]
simonpj [Fri, 19 Mar 2004 13:24:52 +0000 (13:24 +0000)]
[project @ 2004-03-19 13:24:52 by simonpj]
Fix naming bogon for Typeable

20 years ago[project @ 2004-03-18 14:06:18 by simonpj]
simonpj [Thu, 18 Mar 2004 14:06:19 +0000 (14:06 +0000)]
[project @ 2004-03-18 14:06:18 by simonpj]
Arrange that deriving(Typeable) works for higher kinds

20 years ago[project @ 2004-03-18 13:52:32 by simonmar]
simonmar [Thu, 18 Mar 2004 13:52:32 +0000 (13:52 +0000)]
[project @ 2004-03-18 13:52:32 by simonmar]
Default max stack size is now 8M.

20 years ago[project @ 2004-03-18 13:52:09 by simonmar]
simonmar [Thu, 18 Mar 2004 13:52:09 +0000 (13:52 +0000)]
[project @ 2004-03-18 13:52:09 by simonmar]
Change the default max stack size (+RTS -K<n>) to 8M.

20 years ago[project @ 2004-03-17 15:38:13 by simonpj]
simonpj [Wed, 17 Mar 2004 15:38:13 +0000 (15:38 +0000)]
[project @ 2004-03-17 15:38:13 by simonpj]
Wibble

20 years ago[project @ 2004-03-17 13:59:06 by simonpj]
simonpj [Wed, 17 Mar 2004 13:59:19 +0000 (13:59 +0000)]
[project @ 2004-03-17 13:59:06 by simonpj]
------------------------
More newtype clearing up
------------------------

* Change the representation of TyCons so that it accurately reflects
* data     (0 or more constrs)
* newtype  (1 constr)
* abstract (unknown)
  Replaces DataConDetails and AlgTyConFlavour with AlgTyConRhs

* Add IfaceSyn.IfaceConDecls, a kind of stripped-down analogue
  of AlgTyConRhs

* Move NewOrData from BasicTypes to HsDecl (it's now an HsSyn thing)

* Arrange that Type.newTypeRep and splitRecNewType_maybe unwrap just
  one layer of new-type-ness, leaving the caller to recurse.

  This still leaves typeRep and repType in Type.lhs; these functions
  are still vaguely disturbing and probably should get some attention.

Lots of knock-on changes.  Fixes bug in ds054.

20 years ago[project @ 2004-03-17 10:44:06 by simonpj]
simonpj [Wed, 17 Mar 2004 10:44:06 +0000 (10:44 +0000)]
[project @ 2004-03-17 10:44:06 by simonpj]
More Windows building notes

20 years ago[project @ 2004-03-17 10:12:31 by simonpj]
simonpj [Wed, 17 Mar 2004 10:12:33 +0000 (10:12 +0000)]
[project @ 2004-03-17 10:12:31 by simonpj]
Fix debug-printing for Insts

20 years ago[project @ 2004-03-17 08:54:19 by simonpj]
simonpj [Wed, 17 Mar 2004 08:54:19 +0000 (08:54 +0000)]
[project @ 2004-03-17 08:54:19 by simonpj]
Use sub-kind rather than kind-equality check in matchTys,
just as we do in TcUnify.  This is someting I failed to
do when shifting to the funky new kind representation.

20 years ago[project @ 2004-03-16 17:32:05 by simonpj]
simonpj [Tue, 16 Mar 2004 17:32:05 +0000 (17:32 +0000)]
[project @ 2004-03-16 17:32:05 by simonpj]
Print slightly more informative types in debug mode

20 years ago[project @ 2004-03-16 15:35:54 by simonmar]
simonmar [Tue, 16 Mar 2004 15:35:54 +0000 (15:35 +0000)]
[project @ 2004-03-16 15:35:54 by simonmar]
Don't make configure read-only.

20 years ago[project @ 2004-03-16 13:46:54 by ralf]
ralf [Tue, 16 Mar 2004 13:46:54 +0000 (13:46 +0000)]
[project @ 2004-03-16 13:46:54 by ralf]
Adapted deriving Data a bit.
So everything should be now as shown in the Boilerplate II paper.

20 years ago[project @ 2004-03-16 11:39:50 by simonmar]
simonmar [Tue, 16 Mar 2004 11:39:50 +0000 (11:39 +0000)]
[project @ 2004-03-16 11:39:50 by simonmar]
Remove all known hacks in rawSystem:

  - no splitting of the program name using toArgs

  - no avoiding translate when the string already appears to be quoted

  - no avoiding translate for the command name

We now keep separate program name & args for various SysTools
programs: gcc, as, ld, mkdll.

MERGE TO STABLE

20 years ago[project @ 2004-03-16 05:05:31 by mthomas]
mthomas [Tue, 16 Mar 2004 05:05:31 +0000 (05:05 +0000)]
[project @ 2004-03-16 05:05:31 by mthomas]
Retrofit Simon's DOCBOOK_CATALOG fix for platforms without DOCBOOK.

20 years ago[project @ 2004-03-15 12:32:11 by simonmar]
simonmar [Mon, 15 Mar 2004 12:32:11 +0000 (12:32 +0000)]
[project @ 2004-03-15 12:32:11 by simonmar]
Sync up with the stable branch (small change to the way worker threads
exit that sneaked onto the branch by mistake).

20 years ago[project @ 2004-03-13 00:56:45 by sof] Approx_11550_changesets_converted
sof [Sat, 13 Mar 2004 00:56:45 +0000 (00:56 +0000)]
[project @ 2004-03-13 00:56:45 by sof]
tidy up conditional forkProcess() support

20 years ago[project @ 2004-03-12 21:37:27 by ross]
ross [Fri, 12 Mar 2004 21:37:27 +0000 (21:37 +0000)]
[project @ 2004-03-12 21:37:27 by ross]
Deal gracefully with arrow commands where an expression is expected,
instead of panicking.  (Bug reported by Isaac Jones)

merge to STABLE

20 years ago[project @ 2004-03-12 10:16:34 by simonmar]
simonmar [Fri, 12 Mar 2004 10:16:34 +0000 (10:16 +0000)]
[project @ 2004-03-12 10:16:34 by simonmar]
clean up pretty printing slightly

20 years ago[project @ 2004-03-11 14:34:22 by simonpj]
simonpj [Thu, 11 Mar 2004 14:34:24 +0000 (14:34 +0000)]
[project @ 2004-03-11 14:34:22 by simonpj]
Fix a nasty and long-standing bug in the handling of functional dependencies.

The story is told in comments with TcSimplify.tcSimplifyRestricted.  We were
simpifying a group of constraints *twice*: once to establish the type vars to
quantify over, and once "for real" but less brutally.  Unfortunately, the
less-brutally part meant that we did less improvement, which in turn meant
that an invariant (no captured constraints) was violated.  Consequential
bizarre results.

The test is typecheck/should_compile/tc177

20 years ago[project @ 2004-03-11 14:31:56 by simonpj]
simonpj [Thu, 11 Mar 2004 14:31:56 +0000 (14:31 +0000)]
[project @ 2004-03-11 14:31:56 by simonpj]
Better printing for LHsBinds

20 years ago[project @ 2004-03-11 14:31:35 by simonpj]
simonpj [Thu, 11 Mar 2004 14:31:35 +0000 (14:31 +0000)]
[project @ 2004-03-11 14:31:35 by simonpj]
Sanity checking

20 years ago[project @ 2004-03-11 14:31:13 by simonpj]
simonpj [Thu, 11 Mar 2004 14:31:13 +0000 (14:31 +0000)]
[project @ 2004-03-11 14:31:13 by simonpj]
Comments

20 years ago[project @ 2004-03-11 14:29:59 by simonpj]
simonpj [Thu, 11 Mar 2004 14:29:59 +0000 (14:29 +0000)]
[project @ 2004-03-11 14:29:59 by simonpj]
Add sanity check

20 years ago[project @ 2004-03-11 14:29:29 by simonpj]
simonpj [Thu, 11 Mar 2004 14:29:29 +0000 (14:29 +0000)]
[project @ 2004-03-11 14:29:29 by simonpj]
Better layout in interppSP

20 years ago[project @ 2004-03-11 10:52:55 by simonpj]
simonpj [Thu, 11 Mar 2004 10:52:55 +0000 (10:52 +0000)]
[project @ 2004-03-11 10:52:55 by simonpj]
Fix a long-standing bug; the type checker should recover from a failure
in the default method of a class declaration, else the default method
doesn't get added to the type envt, and we get an opaque
    tcLookupGlobal: `FunnyError.$dmb' is not in scope
message.

20 years ago[project @ 2004-03-11 10:00:23 by simonmar]
simonmar [Thu, 11 Mar 2004 10:00:23 +0000 (10:00 +0000)]
[project @ 2004-03-11 10:00:23 by simonmar]
Don't call hGetArray with a size of zero (fixes read021).

20 years ago[project @ 2004-03-11 09:53:45 by simonmar]
simonmar [Thu, 11 Mar 2004 09:53:45 +0000 (09:53 +0000)]
[project @ 2004-03-11 09:53:45 by simonmar]
Add dependency on GHC.TopHandler if this is the main module.

20 years ago[project @ 2004-03-10 00:18:21 by wolfgang]
wolfgang [Wed, 10 Mar 2004 00:18:21 +0000 (00:18 +0000)]
[project @ 2004-03-10 00:18:21 by wolfgang]
Remove -static flag from non-darwin PowerPC builds
(this was a relic from the long-dead AIX support)

Merge to STABLE.

20 years ago[project @ 2004-03-08 16:08:52 by simonmar]
simonmar [Mon, 8 Mar 2004 16:08:52 +0000 (16:08 +0000)]
[project @ 2004-03-08 16:08:52 by simonmar]
Apply $libdir hack to GHCi libs in the same way as ordinary .a libs.

20 years ago[project @ 2004-03-08 12:47:41 by simonmar]
simonmar [Mon, 8 Mar 2004 12:47:41 +0000 (12:47 +0000)]
[project @ 2004-03-08 12:47:41 by simonmar]
Add small embellishment to one of the filename rules.

20 years ago[project @ 2004-03-08 11:20:53 by simonmar]
simonmar [Mon, 8 Mar 2004 11:20:53 +0000 (11:20 +0000)]
[project @ 2004-03-08 11:20:53 by simonmar]
Ignore #! lines for scripts.

20 years ago[project @ 2004-03-08 10:31:00 by stolz]
stolz [Mon, 8 Mar 2004 10:31:00 +0000 (10:31 +0000)]
[project @ 2004-03-08 10:31:00 by stolz]
-Wall police

20 years ago[project @ 2004-03-08 10:27:22 by stolz]
stolz [Mon, 8 Mar 2004 10:27:22 +0000 (10:27 +0000)]
[project @ 2004-03-08 10:27:22 by stolz]
INLINEs for Intel's icc

20 years ago[project @ 2004-03-07 17:56:35 by ross]
ross [Sun, 7 Mar 2004 17:56:36 +0000 (17:56 +0000)]
[project @ 2004-03-07 17:56:35 by ross]
if X is available, build X11 and HGL packages.

20 years ago[project @ 2004-03-05 16:04:52 by simonmar]
simonmar [Fri, 5 Mar 2004 16:04:52 +0000 (16:04 +0000)]
[project @ 2004-03-05 16:04:52 by simonmar]
Be clearer about when source files must match module names.

20 years ago[project @ 2004-03-05 16:04:05 by simonmar]
simonmar [Fri, 5 Mar 2004 16:04:05 +0000 (16:04 +0000)]
[project @ 2004-03-05 16:04:05 by simonmar]
Fail with an error if the interface file we're reading doesn't contain
the module we were expecting.

20 years ago[project @ 2004-03-05 15:09:37 by stolz]
stolz [Fri, 5 Mar 2004 15:09:37 +0000 (15:09 +0000)]
[project @ 2004-03-05 15:09:37 by stolz]
Fix #ifdef'ed flags for FreeBSD.

20 years ago[project @ 2004-03-05 15:05:28 by simonmar]
simonmar [Fri, 5 Mar 2004 15:05:28 +0000 (15:05 +0000)]
[project @ 2004-03-05 15:05:28 by simonmar]
comments

20 years ago[project @ 2004-03-05 14:12:13 by stolz]
stolz [Fri, 5 Mar 2004 14:12:13 +0000 (14:12 +0000)]
[project @ 2004-03-05 14:12:13 by stolz]
Reduce noise if there's no in-tree version of Happy.

20 years ago[project @ 2004-03-05 13:58:20 by simonmar]
simonmar [Fri, 5 Mar 2004 13:58:20 +0000 (13:58 +0000)]
[project @ 2004-03-05 13:58:20 by simonmar]
Fix a URL link

20 years ago[project @ 2004-03-03 17:11:51 by simonpj]
simonpj [Wed, 3 Mar 2004 17:11:51 +0000 (17:11 +0000)]
[project @ 2004-03-03 17:11:51 by simonpj]
Comments only

20 years ago[project @ 2004-03-03 17:10:33 by simonpj]
simonpj [Wed, 3 Mar 2004 17:10:33 +0000 (17:10 +0000)]
[project @ 2004-03-03 17:10:33 by simonpj]
Patches to configuration script for MSYS.  It's not clear that this'll
always work, because we generate paths like /c/mingw/bin, which MinGW
binaries won't understand -- but all of them seem to be fed to the
MSYS shell, which translates them to c:/mingw/bin etc.

Perhaps the Right Thing is to implement msyspath, by analogy with
cygpath, to convert /c/xxx to c:/xxx?

Anyway, this patch seems to work for now.

20 years ago[project @ 2004-03-02 22:22:48 by ralf]
ralf [Tue, 2 Mar 2004 22:22:48 +0000 (22:22 +0000)]
[project @ 2004-03-02 22:22:48 by ralf]
Once more revised some details of the Data class.
Comitting the Data.Generics* library in a second.

20 years ago[project @ 2004-03-02 13:57:55 by simonpj]
simonpj [Tue, 2 Mar 2004 13:57:55 +0000 (13:57 +0000)]
[project @ 2004-03-02 13:57:55 by simonpj]
New chapter on MSYS, filenames etc

20 years ago[project @ 2004-03-01 14:18:35 by simonmar]
simonmar [Mon, 1 Mar 2004 14:18:36 +0000 (14:18 +0000)]
[project @ 2004-03-01 14:18:35 by simonmar]
Threaded RTS improvements:

  - Make the main_threads list doubly linked.  Have threads
    remove themselves from this list when they complete, rather
    than searching for completed main threads each time around
    the scheduler loop.  This removes an O(n) loop from the
    scheduler, but adds some new constraints (basically completed
    threads must remain on the run queue until dealt with, including
    threads which have been killed by an async exception).

  - Add a pointer from the TSO to the StgMainThread struct, for
    main threads.  This avoids a number of places where we had
    to traverse the list of main threads to find the right one,
    including one place in the scheduler loop.  Adding a field to
    a TSO is cheap.

  - taskStart: we should be resetting the startingWorkerThread flag
    in here.  Not sure why we aren't; maybe this got lost at some point.

  - Use the BlockedOnCCall flags in the non-threaded RTS too.  Q: what
    should happen if a thread does a foreign call which re-enters the
    RTS, and then sends an async exception to the original thread?
    Answer: it should deadlock, which it does in the threaded RTS, and
    this commit makes it do so in the non-threaded RTS too (see
    testsuite/tests/concurrent/should_run/conc040.hs).

20 years ago[project @ 2004-03-01 14:11:01 by simonmar]
simonmar [Mon, 1 Mar 2004 14:11:01 +0000 (14:11 +0000)]
[project @ 2004-03-01 14:11:01 by simonmar]
Remove Deadlock_closure (not used any more).

20 years ago[project @ 2004-03-01 13:44:10 by simonmar]
simonmar [Mon, 1 Mar 2004 13:44:10 +0000 (13:44 +0000)]
[project @ 2004-03-01 13:44:10 by simonmar]
add a few RTS ways (combinations) I missed out

20 years ago[project @ 2004-03-01 13:34:54 by simonmar]
simonmar [Mon, 1 Mar 2004 13:34:54 +0000 (13:34 +0000)]
[project @ 2004-03-01 13:34:54 by simonmar]
Warning police, and omit a useless debugging msg in !RTS_SUPPORTS_THREADS.

20 years ago[project @ 2004-03-01 12:33:17 by simonmar]
simonmar [Mon, 1 Mar 2004 12:33:17 +0000 (12:33 +0000)]
[project @ 2004-03-01 12:33:17 by simonmar]
wibble

20 years ago[project @ 2004-03-01 10:17:53 by simonmar]
simonmar [Mon, 1 Mar 2004 10:17:53 +0000 (10:17 +0000)]
[project @ 2004-03-01 10:17:53 by simonmar]
Update version numbering policy: we now allow snapshots along the
stable branch (eg. 6.2.20040229).

20 years ago[project @ 2004-03-01 10:16:28 by simonmar]
simonmar [Mon, 1 Mar 2004 10:16:28 +0000 (10:16 +0000)]
[project @ 2004-03-01 10:16:28 by simonmar]
Update docs for RTS options and GhcRtsThreaded

20 years ago[project @ 2004-03-01 10:14:54 by simonmar]
simonmar [Mon, 1 Mar 2004 10:14:54 +0000 (10:14 +0000)]
[project @ 2004-03-01 10:14:54 by simonmar]
Add -fno-generics to SRC_HC_OPTS

20 years ago[project @ 2004-03-01 10:13:01 by simonmar]
simonmar [Mon, 1 Mar 2004 10:13:01 +0000 (10:13 +0000)]
[project @ 2004-03-01 10:13:01 by simonmar]
build target should use '::' not ':'

20 years ago[project @ 2004-03-01 09:49:48 by simonmar]
simonmar [Mon, 1 Mar 2004 09:49:48 +0000 (09:49 +0000)]
[project @ 2004-03-01 09:49:48 by simonmar]
make it compile (win32)

20 years ago[project @ 2004-02-28 15:37:09 by ralf]
ralf [Sat, 28 Feb 2004 15:37:09 +0000 (15:37 +0000)]
[project @ 2004-02-28 15:37:09 by ralf]
Revised dataTypeOf member in "deriving ( ... Data ...)".
This revised schemes relies on the updated modules
Data.Typeable and Data.Generics*.

20 years ago[project @ 2004-02-27 16:16:31 by simonmar]
simonmar [Fri, 27 Feb 2004 16:16:31 +0000 (16:16 +0000)]
[project @ 2004-02-27 16:16:31 by simonmar]
cosmetic only

20 years ago[project @ 2004-02-27 15:58:54 by simonmar]
simonmar [Fri, 27 Feb 2004 15:58:54 +0000 (15:58 +0000)]
[project @ 2004-02-27 15:58:54 by simonmar]
redo the condition-cache hack

20 years ago[project @ 2004-02-27 15:56:25 by simonmar]
simonmar [Fri, 27 Feb 2004 15:56:25 +0000 (15:56 +0000)]
[project @ 2004-02-27 15:56:25 by simonmar]
Remove an unnecessary #ifdef

20 years ago[project @ 2004-02-27 13:35:16 by simonmar]
simonmar [Fri, 27 Feb 2004 13:35:16 +0000 (13:35 +0000)]
[project @ 2004-02-27 13:35:16 by simonmar]
hs_exit(): call exitScheduler() *after* finalizeWeakPointersNow(),
because the latter needs to create/run threads.

20 years ago[project @ 2004-02-27 12:39:16 by simonmar]
simonmar [Fri, 27 Feb 2004 12:39:16 +0000 (12:39 +0000)]
[project @ 2004-02-27 12:39:16 by simonmar]
scheduleWaitThread(): don't call THREAD_RUNNABLE because the new
thread is bound, and therefore only runnable by the current OS
thread.  Calling THREAD_RUNNABLE just wakes up workers and slows
things down.

20 years ago[project @ 2004-02-26 16:49:34 by simonmar]
simonmar [Thu, 26 Feb 2004 16:49:34 +0000 (16:49 +0000)]
[project @ 2004-02-26 16:49:34 by simonmar]
waitCondition: use RELEASE_LOCK/ACQUIRE_LOCK because these check the
return value for errors.

20 years ago[project @ 2004-02-26 16:42:57 by simonmar]
simonmar [Thu, 26 Feb 2004 16:42:57 +0000 (16:42 +0000)]
[project @ 2004-02-26 16:42:57 by simonmar]
Check result of SetEvent() for errors.

20 years ago[project @ 2004-02-26 16:39:40 by simonmar]
simonmar [Thu, 26 Feb 2004 16:39:40 +0000 (16:39 +0000)]
[project @ 2004-02-26 16:39:40 by simonmar]
#include "RtsUtils.h" for barf prototype

20 years ago[project @ 2004-02-26 16:36:21 by simonmar]
simonmar [Thu, 26 Feb 2004 16:36:21 +0000 (16:36 +0000)]
[project @ 2004-02-26 16:36:21 by simonmar]
Add the usual multi-slurp protection

20 years ago[project @ 2004-02-26 16:31:44 by simonmar]
simonmar [Thu, 26 Feb 2004 16:31:44 +0000 (16:31 +0000)]
[project @ 2004-02-26 16:31:44 by simonmar]
Avoid signalling thread_ready_cond if rts_n_waiting_tasks==0.

20 years ago[project @ 2004-02-26 16:21:36 by simonmar]
simonmar [Thu, 26 Feb 2004 16:21:36 +0000 (16:21 +0000)]
[project @ 2004-02-26 16:21:36 by simonmar]
merge rev. 1.6.2.3 from ghc-6-2-branch

20 years ago[project @ 2004-02-26 16:19:32 by simonmar]
simonmar [Thu, 26 Feb 2004 16:19:32 +0000 (16:19 +0000)]
[project @ 2004-02-26 16:19:32 by simonmar]
thread_id_mutex isn't required (at least for non-SMP), since
createThread is under the Capability.

20 years ago[project @ 2004-02-26 16:14:21 by simonmar]
simonmar [Thu, 26 Feb 2004 16:14:21 +0000 (16:14 +0000)]
[project @ 2004-02-26 16:14:21 by simonmar]
no need to declare thread_ready_cond

20 years ago[project @ 2004-02-26 16:13:46 by simonmar]
simonmar [Thu, 26 Feb 2004 16:13:46 +0000 (16:13 +0000)]
[project @ 2004-02-26 16:13:46 by simonmar]
thread_ready_cond is initialised in two places

20 years ago[project @ 2004-02-26 14:49:05 by simonpj]
simonpj [Thu, 26 Feb 2004 14:49:05 +0000 (14:49 +0000)]
[project @ 2004-02-26 14:49:05 by simonpj]
Back out hiding change -- hiding should hide the qualified name too

20 years ago[project @ 2004-02-26 12:42:07 by simonpj]
simonpj [Thu, 26 Feb 2004 12:42:07 +0000 (12:42 +0000)]
[project @ 2004-02-26 12:42:07 by simonpj]
Fix a long-standing bug, whereby 'hiding' hid the qualified name as
well as the unqualified one.

This bug is not as easy to fix in the 6.2 branch, unfortunately; don't merge.

20 years ago[project @ 2004-02-26 11:41:22 by simonmar]
simonmar [Thu, 26 Feb 2004 11:41:22 +0000 (11:41 +0000)]
[project @ 2004-02-26 11:41:22 by simonmar]
oops, fix the build

20 years ago[project @ 2004-02-25 17:35:44 by simonmar]
simonmar [Wed, 25 Feb 2004 17:35:44 +0000 (17:35 +0000)]
[project @ 2004-02-25 17:35:44 by simonmar]
Feeble performance hack for the threaded RTS: instead of
allocating/releasing a new condition variable for each new call-in, we
just cache one in the RTS and re-use it for the next call.

On a little test I have here which does lots of call-ins on Windows,
this reduces the slowdown for using the threaded RTS from a factor of
7-8 down to a factor of 4-5.  I'm aiming for a factor of 2 or better...

20 years ago[project @ 2004-02-25 16:18:00 by simonmar]
simonmar [Wed, 25 Feb 2004 16:18:00 +0000 (16:18 +0000)]
[project @ 2004-02-25 16:18:00 by simonmar]
ShutdownIOManager: don't free the IOManagerState, because the worker
threads haven't necessarily died yet and might reference it again.

Apparently fixes a strange bug: I have a Haskell program which on
Windows causes the shell to exit after I run it.  I don't know what
the real cause of this is, but this is a clear bug and fixing it stops
the shell from exiting.

MERGE TO STABLE

20 years ago[project @ 2004-02-25 14:51:37 by simonpj]
simonpj [Wed, 25 Feb 2004 14:51:38 +0000 (14:51 +0000)]
[project @ 2004-02-25 14:51:37 by simonpj]
Add a -fno-full-laziness flag

20 years ago[project @ 2004-02-25 13:57:38 by simonmar]
simonmar [Wed, 25 Feb 2004 13:57:38 +0000 (13:57 +0000)]
[project @ 2004-02-25 13:57:38 by simonmar]
- For each language option, document exactly the syntax that is
  stolen from Haskell 98.

- Fix up docrot at the beginning of the chapter.

20 years ago[project @ 2004-02-25 13:54:30 by simonpj]
simonpj [Wed, 25 Feb 2004 13:54:31 +0000 (13:54 +0000)]
[project @ 2004-02-25 13:54:30 by simonpj]
Yet another fix to the -Onot optimisation that hides data type
representations in .hi files.

1.  Expose the representation if any fields are exposed

2.  Don't expose newtypes whose data-cons are abstract, unless the
    rep type is a FFI type.  (Previously we were more conservative
    and always exposed newtypes, just in case of a foreign decl.)

20 years ago[project @ 2004-02-25 11:49:14 by simonmar]
simonmar [Wed, 25 Feb 2004 11:49:15 +0000 (11:49 +0000)]
[project @ 2004-02-25 11:49:14 by simonmar]
Document -threaded and -debug options.

20 years ago[project @ 2004-02-25 11:34:29 by simonmar]
simonmar [Wed, 25 Feb 2004 11:34:30 +0000 (11:34 +0000)]
[project @ 2004-02-25 11:34:29 by simonmar]
Allow configuration of RTS ways by tweaking the GhcRTSWays variable.
Add default settings to config.mk.in, and add a comment to explain
which values make sense here.

Also, I disabled the debug_p way by default; it's of limited
usefulness to the average punter.

20 years ago[project @ 2004-02-25 11:31:24 by simonmar]
simonmar [Wed, 25 Feb 2004 11:31:24 +0000 (11:31 +0000)]
[project @ 2004-02-25 11:31:24 by simonmar]
Combined build tags get '_' between them

20 years ago[project @ 2004-02-25 11:24:34 by simonmar]
simonmar [Wed, 25 Feb 2004 11:24:34 +0000 (11:24 +0000)]
[project @ 2004-02-25 11:24:34 by simonmar]
Add missing case to allowedWith

20 years ago[project @ 2004-02-25 11:17:47 by simonmar]
simonmar [Wed, 25 Feb 2004 11:17:47 +0000 (11:17 +0000)]
[project @ 2004-02-25 11:17:47 by simonmar]
Delete options for building the debugging RTS; it is now built by default.

20 years ago[project @ 2004-02-25 11:16:20 by simonmar]
simonmar [Wed, 25 Feb 2004 11:16:20 +0000 (11:16 +0000)]
[project @ 2004-02-25 11:16:20 by simonmar]
Fix a test

20 years ago[project @ 2004-02-25 10:34:44 by simonmar]
simonmar [Wed, 25 Feb 2004 10:34:44 +0000 (10:34 +0000)]
[project @ 2004-02-25 10:34:44 by simonmar]
Add an entry to the FAQ about buffering of stdout

20 years ago[project @ 2004-02-24 17:56:00 by simonpj]
simonpj [Tue, 24 Feb 2004 17:56:00 +0000 (17:56 +0000)]
[project @ 2004-02-24 17:56:00 by simonpj]
more hi-boot documentation

20 years ago[project @ 2004-02-24 17:37:11 by simonmar]
simonmar [Tue, 24 Feb 2004 17:37:11 +0000 (17:37 +0000)]
[project @ 2004-02-24 17:37:11 by simonmar]
wibble to previous commit; forgot to save my buffer

20 years ago[project @ 2004-02-24 17:33:32 by simonmar]
simonmar [Tue, 24 Feb 2004 17:33:34 +0000 (17:33 +0000)]
[project @ 2004-02-24 17:33:32 by simonmar]
Experimental support for RTS-only "ways"

HEADS UP!  This changes the way that the threaded RTS is used, and
also the use of debugging RTSs:

  - We always build threaded and debugging variants of the RTS now.
    The --enable-threaded-rts configure option is ignored (and will
    be removed at some point).

  - New option:  -debug     enables the debugging RTS

  - New option:  -threaded  enables the threaded RTS.  When the threaded
    RTS is stable enough, we might make it the default.

The new options just cause a different variant of the RTS to be linked
in, and they cause one or two extra options to be enabled too.  The
implementation is via the usual ways machinery in the compiler, except
that these ways are labelled as RTS-only, and so don't require
rebuilding all the libraries too.

All of this means we can ship threaded and debugging RTSs with GHC, so
that users don't need to fetch and build a GHC source tree to use
them.

I'd like to get this functionality into 6.2.1 if possible, so please
test (I'm willing to stretch the definition of "interface change" to
accomodate this, since having a threaded RTS available without having
to build GHC will be a big win for the Visual Studio project).

20 years ago[project @ 2004-02-24 17:13:56 by simonpj]
simonpj [Tue, 24 Feb 2004 17:13:56 +0000 (17:13 +0000)]
[project @ 2004-02-24 17:13:56 by simonpj]
Better error message for no-instance in deriving clause