Commit Graph

6887 Commits

Author SHA1 Message Date
kr.angelov
b7798f9348 add evaluator.c in Android.mk 2014-09-30 12:37:18 +00:00
kr.angelov
f25b45e13d fix in reader.h for compiling jpgf.c 2014-09-30 11:25:27 +00:00
kr.angelov
6be25aa872 fix in pgf_print_expr for lambda abstractions with more than one argument 2014-09-30 09:42:18 +00:00
aarne
7c5cb356ca AppChi needs more stack space now 2014-09-30 09:35:51 +00:00
kr.angelov
70455b88a7 bugfix in the gate evaluate_value_lambda 2014-09-30 08:42:06 +00:00
kr.angelov
f70eecb63c minor stylistic change in GenerateBC 2014-09-29 15:08:00 +00:00
kr.angelov
80acad4447 bugfix in the pattern matching compiler and a number of other fixes that I somehow did not push before 2014-09-29 15:00:04 +00:00
kr.angelov
1ecd61f7f5 bugfix in the RET instruction 2014-09-29 13:04:49 +00:00
kr.angelov
87c5c4b0cc enlarge the code window in pgf_jit_gates to ensure enough space on 64-bit machines 2014-09-29 07:34:32 +00:00
aarne
aef787aeb1 checked that course of value tables (table P [...]) have the correct number of values w.r.t the type P. This was previously not checked, and caused hard-to-find run-time errors. 2014-09-27 17:14:35 +00:00
aarne
ca1cad68d0 command for AppEngFin 2014-09-26 14:48:09 +00:00
kr.angelov
7036d56d29 the literals API is now exposed in Python 2014-09-25 11:59:40 +00:00
kr.angelov
8e64fc6e7e switch off the debugging of the JIT compiler which I had turned on accidentally. silence two harmless warnings 2014-09-25 11:44:45 +00:00
kr.angelov
14e6eec5ec now a complete JIT compiler and ByteCode compiler for the def rules in the abstract syntax. there might be some bugs yet to be found, meta variables and computation under lambda is only partially supported 2014-09-25 10:35:06 +00:00
hallgren
928217f9d9 examples/app/Makefile2: add missing -probs flag 2014-09-23 12:20:44 +00:00
hallgren
13fe719a26 doc/gf-developers.t2t: add a note about how to create OS X Installer packages 2014-09-23 12:18:17 +00:00
hallgren
db9aeb2fa2 haskell-bind/PGF2.hsc: unexport functions that break referential transparency
loadConcr, unloadConcr and addLiteral modify the Concr structure as a side
effect. This means that other functions with a Concr argument (e.g. parse
and linearize) are no longer pure.

Possible solutions:

1. Don't try to hide the imperative nature of the C run-time system: remove
   all uses of unsafePerformIO and let all functions operate in the IO monad.

2. Don't export functions with side effects. Perhaps the desired functionality
   of loadConcr, unloadConcr and addLiteral can be folded into readPGF.
   The Concr structures can then treaded as immutable after after the
   readPGF function returns...
2014-09-22 14:05:17 +00:00
kr.angelov
3ced115807 forgot to type check the type of a typed let expression 2014-09-17 22:25:08 +00:00
kr.angelov
e34d5b0409 appForm now skips over Typed expressions 2014-09-17 22:02:36 +00:00
kr.angelov
a5f81ff81c the type checker for the abstract syntax now allows let expressions in def rules, since they are easily compilable to byte code. This fails in the Haskell runtime since let expressions are not allowed as abstract syntax expressions. 2014-09-17 21:59:45 +00:00
kr.angelov
818896dec2 fix a portability problem with the Python binding (found by Prasanth) 2014-09-17 16:12:40 +00:00
kr.angelov
3e3178abfa dummy jit_base_tail_finishr for x86_64. 2014-09-16 12:29:35 +00:00
aarne
8d18828008 enabled VP-valued rules for V2V, V3, etc in App. Improve quality with less loss of speed than VPSlash-valued original rules. 2014-09-15 16:10:18 +00:00
aarne
2d91097f49 enabled many more functions, in particular verb patterns, in App, to increase translation quality. This comes with 30% longer translation time with my benchmark but should still be OK. 2014-09-14 08:43:21 +00:00
aarne
d19b38657f GF index: shortcut links to Android app, Best practices, REMU, and new Android development instructions 2014-09-12 09:01:43 +00:00
kr.angelov
621d748bac a major revision of the bytecode generator and JIT compiler. the effect is that now we can compute with lambda functions and with true tail recursion 2014-09-11 15:39:39 +00:00
hallgren
18ee232497 gf.cabal: add missing c-sources for the Haskell binding to the C run-time library 2014-09-11 11:53:07 +00:00
hallgren
4f9af46ed0 Setup.hs: don't ask darcs for the version history if _darcs is not present
This is to avoid a (harmless) error message from darcs when compiling sources
obtained e.g. from github.
2014-09-11 11:47:17 +00:00
kr.angelov
df0602723d now release the FunPtr:s that are allocated for each literal callback 2014-09-10 15:41:53 +00:00
kr.angelov
80725e872b added an API for custom literals in the Haskell binding 2014-09-10 14:56:18 +00:00
kr.angelov
b553729f37 added loadConcr/unloadConcr to the Haskell binding. This exposes an API for loading grammars compiled with -split-pgf 2014-09-10 14:35:54 +00:00
hallgren
566aeb93f5 Wide Coverage Demo web app: use the App grammar instead of the Translate grammar
This gives faster but slightly lower quality translations.
2014-09-09 13:53:10 +00:00
hallgren
4eb6b55e98 (1) Refactor concurrency, (2) write to .gfo.tmp then rename to .gfo
(1) introduces the module GF.Infra.Concurreny with lifted concurrency
    operators (to reduce uses of liftIO) and some additional concurrency
    utilities, e.g. a function for sequential logging that is used in
    both GF.CompileInParallel and GFServer.
(2) avoids leaving broken .gfo files behind if compilation is aborted.
2014-09-08 15:43:20 +00:00
kr.angelov
4d28c7632e the code for def rules now uses proper graph update to preserve lazyness 2014-09-05 11:53:02 +00:00
kr.angelov
86b5f78c57 full support for recursive def rules in the C runtime 2014-09-05 10:09:43 +00:00
hallgren
a21ffc1941 gf.cabal, gf-server.cabal: add version bounds on network & httpd-shed
* httpd-shed-0.4 does not specify an upper bound on network, but it fails
  to build against network>=2.6. This is fixed in httpd-shed-0.4.0.2.
* With network-2.6, the Network.URI modules is moved to a separate package,
  so for the time being GF requires network>=2.3 && <2.6. This is compatible
  with the four most recent versions of the Haskell Platform.
2014-09-03 15:40:26 +00:00
hallgren
e0e6079c91 src/server: refactoring to isolate dependencies on the cgi/fastcgi packages
* Introducing the module CGI, re-exporting a subset of the cgi package. It
  might complete replace the cgi package in the future.
* Introducing the module CGIUtils, containing functions from FastCGIUtils that
  have nothing to do with fastcgi.

Some low level hackery with unsafePerformIO and global variables was left
in FastCGIUtils, but it is actually not used, neither for gf -server nor
exec/pgf-fcgi.hs.
2014-09-02 12:27:47 +00:00
kr.angelov
bfd414554d partial implementation for recursive def rules 2014-09-01 14:51:20 +00:00
kr.angelov
342f6e3797 bug fixes in the JIT compiler 2014-09-01 12:53:14 +00:00
aarne
4e8cd9119e by_Prep causing conflict in AppChi, therefore excluded from the import of PhrasebookChi 2014-08-30 23:13:39 +00:00
hallgren
d3d1d20a8a examples/app/Makefile2: simplifed Makefile for the App grammar (doesn't work yet)
This makefile just calls GF once and lets GF figure out in which order to
compile things. It uses the -j flag to enable parallel compilation and
specifies an explicit -path, overriding the -path flags in the source files.
This allows all needed modules to be found automatically and ensures that
that alltenses is consistently used everywhere. But for some reason, this
doesn't work...
2014-08-30 17:00:14 +00:00
hallgren
d0d470620c Add OS X Installed package to the GF 3.6 download page 2014-08-29 11:48:53 +00:00
hallgren
bcf0aa42cd bin/build-binary-dist.sh: just a small fix 2014-08-29 11:48:10 +00:00
kr.angelov
cee5f590ca added evaluator.h which I had forgoten 2014-08-29 08:36:53 +00:00
hallgren
658b9adabe You can now use 'make pkg' to build OS X Installer packages for GF
The script bin/build-binary-dist.sh has been updated to build either a plain
.tar.gz package or OS X Installer package (.pkg).

Note that bin/build-binary-dist.sh is designed to build and include the
C run-time system in the binary package.  If the C run-time system fails to
build, no binary package will be created.
2014-08-28 13:30:22 +00:00
hallgren
883d1ae21c gf.cabal ghc-options: don't force -j, -auto-all has been renamed to -fprof-auto 2014-08-25 23:30:22 +00:00
hallgren
fdf7d06f9a Fix GHC 7.4 compatibility issue caused by previous patch 2014-08-25 11:26:58 +00:00
hallgren
a494c713e1 Setup.hs: disable parallel RGL build support for now, since it is incompatible with Cabal<1.20 2014-08-25 11:22:20 +00:00
hallgren
d84c5ef171 Experimental: parallel batch compilation of grammars
On my laptop these changes speed up the full build of the RGL and example
grammars with 'cabal build' from ~95s to ~43s and the zero build from ~18s
to ~5s.

The main change is the introduction of the module GF.CompileInParallel that
replaces GF.Compile and the function GF.Compile.ReadFiles.getAllFiles. At
present, it is activated with the new -j flag, and it is only used when
combined with --make or --batch. In addition, to get parallel computations,
you need to add GHC run-time flags, e.g., +RTS -N -A20M -RTS, to the command
line.

The Setup.hs script has been modified to pass the appropriate flags to GF
for parallel compilation when compiling the RGL and example grammars, but you
need a recent version of Cabal for this to work (probably >=1.20).

Some additonal refactoring were made during this work. A new monad is used to
avoid warnings/error messages from different modules to be intertwined when
compiling in parallel, so some functios that were hardiwred to the IO or IOE
monads have been lifted to work in arbitrary monads that are instances in
the appropriate classes.
2014-08-25 09:56:00 +00:00
aarne
9253d54b7e made the "how old" construction in Chinese more efficient by moving variants to the top level; the resulting rule is the same. This kind of inefficiencies are best found if you compile with the -v option and check for constructors creating unexpectedly many PGF rules. 2014-08-24 14:28:48 +00:00