Vitor Santos Costa
81635b48fb
use isattvar
2010-03-08 09:17:40 +00:00
Vitor Santos Costa
3070e4ce9d
fix locking issues with dynamic predicates and threads (obs from Paulo Moura).
2010-03-03 22:11:50 +00:00
Vítor Santos Costa
692a482adf
fix gc call while in meta-call.
2010-02-19 15:37:36 +00:00
Vitor Santos Costa
3b4bb45c62
fix error handling in builtin arithmetic (obs from Jose Santos).
2010-02-18 10:56:59 +00:00
Vítor Manuel de Morais Santos Costa
4d341f2aab
fix bad call to gc (obs from Jose Santos).
2010-01-29 15:21:00 +00:00
Vítor Manuel de Morais Santos Costa
bcc4a27206
cut_succeed and cut_fail should recover pool of try_c and should
...
do trim_trail (not that the latter really matters).
2010-01-26 18:03:09 +00:00
Vítor Manuel de Morais Santos Costa
b0ea945ebc
Merge branch 'master' of gitosis@yap.dcc.fc.up.pt:yap-6
2010-01-26 11:40:47 +00:00
Vítor Manuel de Morais Santos Costa
4d54ea4c92
improve PP on threaded version
2010-01-26 11:40:32 +00:00
Vitor Santos Costa
ce3a193ae7
fix forgotten unlocks.
2010-01-25 09:01:04 +00:00
Vítor Manuel de Morais Santos Costa
5ef31fbfd2
experiment in or-parallel threads.
2010-01-14 15:58:19 +00:00
Vitor Santos Costa
df2b11a6ec
define Optype for xxc, for instructions whose constant argument is a term, like functor(a,X,T).
2009-12-06 00:37:48 +00:00
Vítor Manuel de Morais Santos Costa
d746afaa32
UserCode may changed YENV, so be careful, be very careful! (obs from Nuno Fonseca).
2009-11-27 13:14:30 +00:00
Vítor Manuel de Morais Santos Costa
b5ad71c9c1
improve functor tracing
2009-11-10 11:25:39 +00:00
Vítor Manuel de Morais Santos Costa
2fdf8e9bda
fix bad ! register when calling continuation of meta-predicate (obs from A Kimmig).
2009-11-03 23:17:06 +00:00
Costa Vitor
418fa1a396
64 bit machines may be using S while calling ExpandIndex.
2009-08-19 20:22:04 -05:00
Ricardo Rocha
fc31e7066e
improve the support for deterministic tabling (still experimental,
...
problems mainly with the garbage collector)
2009-08-07 17:29:53 +01:00
Costa Vitor
66683aeef9
slots were not inited in tabling code, due to bad ifdef (obs from Chris Mungall).
2009-07-17 18:39:50 -05:00
Ricardo Rocha
5e4d7078ab
support for trie compact lists (version 2): this new version uses
...
different marks to distinguish the lists ending with the empty list
from the lists ending with a non-empty list term.
2009-07-05 05:14:08 +01:00
Costa Vitor
026a666055
trail check on C entry: make sure we don't have unhandled trail overflows.
2009-06-26 14:08:15 -05:00
Costa Vitor
772d091a70
fix bad call to retry_counter
2009-06-26 13:16:42 -05:00
Costa Vitor
246d32551a
do not set PREG until being sure we have to jump
2009-06-23 06:50:53 -05:00
Costa Vitor
d47560342c
more fixes to call_counter: try to keep on counting system builtins.
2009-06-22 12:40:55 -05:00
Costa Vitor
50f7a80546
transfer call and profiler to spy.
2009-06-22 11:12:56 -05:00
Costa Vitor
53fa330d90
always use YREG, not YENV, in case we have enough registers.
2009-06-16 21:20:35 -05:00
Vitor Santos Costa
95f28228f6
- separate user_try and user_retry
...
- make sure we have slots inited in retrying user C code.
2009-06-01 15:30:22 -05:00
Vitor Santos Costa
e84ecf2eaf
fix #37 : give resource error if trying to create a functor with bignum arity
...
also, allow meta-call to create large functors > Int.
2009-05-29 22:34:50 -05:00
Costa Vitor
c00facdd17
exception at fail could loop with bad env. #59
2009-05-22 19:43:17 -05:00
Costa Vitor
8fcc534a19
fix bad error propagation in term comparison #53
2009-05-22 18:44:05 -05:00
Vitor Santos Costa
8e7695a42c
make Yap_Eval only on entry.
2009-05-16 22:41:44 -07:00
Vitor Santos Costa
61973a2be6
be careful: error handling may have changed PC in call_c_with_fail!
2009-05-09 23:40:17 -04:00
Vitor Santos Costa
f53c741d96
improve low-level tracer.
2009-05-09 23:30:20 -04:00
Vitor Santos Costa
35c1cb6338
put future environment when waking up goals before failure.
2009-05-04 18:11:43 -05:00
Vitor Santos Costa
2016e118b3
Ulrich Neumerkel's patch so that variables are declared in front of a block.
2009-05-02 10:54:09 -05:00
Vitor Santos Costa
8275f2232c
fix handling of multiassignment variables with tabling;
...
fix bad overflow handling when copying terms from tries.
2009-05-01 11:53:59 -05:00
Vitor Santos Costa
c92765d962
SREG and S could be different; also S might be in use if there is an
...
overflow and the index code is jumping to the middle.
2009-04-18 15:22:51 -05:00
Vitor Santos Costa
7e98e5acdb
fix another internal deadlock.
2009-04-17 14:17:43 -05:00
Vitor Santos Costa
9624026beb
locking fixes.
2009-04-16 11:19:26 -05:00
Vitor Santos Costa
b2ad3bd000
first cut at Anderson's JIT code.
2009-04-07 23:55:16 +01:00
Vitor Santos Costa
5c2e5c524b
recognise exceptions within comparisons.
2009-03-26 00:37:57 +00:00
Vitor Santos Costa
828964c3c1
remove optimised arithmetic instructions. They made the emulator even bigger than
...
usual, and they were not performing.
2009-03-09 11:52:01 +00:00
Vitor Santos Costa
afd979a246
improve substantially handling of soft disjunctions.
2009-03-05 16:12:21 +00:00
Vitor Santos Costa
15b86cded4
new merged instructions and small changes to the emulator.
2009-03-03 10:04:13 +00:00
Vitor Santos Costa
a000af113b
OSX portability updates
...
start of support for commons prolog initiative
2009-02-16 12:25:03 +00:00
Vítor Santos Costa
8a6e0acedc
compiler support for udi including user_index instruction
2009-02-12 21:35:31 +00:00
Vitor Santos Costa
8a3978e3e1
rename BinaryTestPredFlag to BinaryPredFlag
...
get rid of small annoying arithmetic bugs
2009-02-09 21:56:40 +00:00
Vítor Santos Costa
a8ac9bfb43
generate known atoms and functors automatically
2008-12-23 01:53:52 +00:00
Vítor Santos Costa
a36be5bf20
massive changes to support new scheme for arithmetic:
...
- compilation and assembly support
2008-12-09 12:54:27 +00:00
Vítor Santos Costa
e737599dc4
Change to simpler Eval mechanism
...
- avoid duplicate code
- implement different optimised code.
2008-12-04 23:33:32 +00:00
Vítor Santos de Costa
681d0f186d
fix bad type for p_number_y
2008-10-07 22:00:38 +01:00
Vítor Santos Costa
e252ef0b2f
more debugger fixes!
...
- make sure donotrace and oncenotrace are known to the signal mechanism:
they should disable creeping
- avoid creep at deallocate: it is a very bad idea :(
2008-09-23 23:43:01 +01:00
