rishi/scryer-prolog
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
3,876 Commits 2 Branches 18 Tags
08aed04dd2aeed354d1f407158e3359437e8071a
Commit Graph

786 Commits

Author SHA1 Message Date
Aleksy Grabowski
2b73107631 Explain non_counted_backtracking a little bit more 2024-09-06 08:15:45 +02:00
Markus Triska
0005dcaacb ADDED: Preliminary support for statistics/2. 
This is needed for benchmarking library(reif) and its newly provided
goal expansion (#2433).

It partly addresses #321.
 2024-07-06 11:33:02 +02:00
Mark Thom
f3eea6302d delay instantiation errors on lists in phrase/3 until runtime (#1665, #2396) 2024-04-29 15:28:17 -06:00
Markus Triska
795a7c690d ADDED: meta_predicate/1 declarations for (',')//2 and (;)//2 
This addresses #2366.
 2024-03-21 22:14:40 +01:00
Markus Triska
767d90d5e7 ENHANCED: further partial evaluation for the common control sequence ~d 2024-03-20 21:10:14 +01:00
Markus Triska
bc04790fd5 extend partial evaluation to computing lengths when statically possible 2024-03-20 21:10:14 +01:00
Markus Triska
58f7c94e74 goal expansion for format/2 and format/3 
In this way, we benefit from partial evaluation of format strings.
 2024-03-20 21:10:14 +01:00
Markus Triska
8af29c200d strengthen type check 2024-03-20 21:10:10 +01:00
Markus Triska
1db6bff890 ENHANCED: Partial evaluation of format_//2. 
This moves analysis of the format string to compilation time when
possible, so that parsing it at run time is no longer necessary.
 2024-03-20 20:59:43 +01:00
Markus Triska
ee60c463c8 indicate deprecated functionality 2024-03-16 21:23:12 +01:00
Markus Triska
e1a6b5eb03 DOC: Un-document the ability to use bytes with encoding(octet). 
This usage is now considered deprecated. Use strings instead for
greater efficiency and type consistency in your applications.

This functionality is a candidate for removal in a future commit.
 2024-03-16 20:37:06 +01:00
Markus Triska
4ebdd97ad3 DOC: Improve descriptions to make clear what encoding(octet) means. 
In particular, octet does *not* mean that the input is a list of
Prolog bytes in the sense of "integers between 0..255", but that the
character codes are used directly as byte values.
 2024-03-16 20:36:22 +01:00
Markus Triska
6b8a679a51 explain potential side-channel attacks due to compact string representation 
This legitimate concern was already raised by @infogulch in:

https://github.com/mthom/scryer-prolog/issues/1309#issuecomment-1080028854

Thank you a lot!
 2024-03-01 22:03:57 +01:00
Markus Triska
ec251b254c ENHANCED: Safe HMAC verification, using constant time string comparison. 
Without this provision, the expected HMAC can be gathered from timing
differences depending on the position where the strings first diverge,
and hence an attacker can forge an authenticated message by supplying
the gathered HMAC.

Test case, using exp(E) to succeed exactly 2^E times:

    exp(E) :-
        N is 2^E,
        between(1, N, _).

yielding:

    ?- Options = [algorithm(sha512),hmac([1,2,3])],
       Ds = "test",
       crypto_data_hash(Ds, H, Options),
       phrase((seq(As),seq(Bs)), H),
       same_length(Bs, Cs),
       maplist(=(a), Cs),
       append(As, Cs, H1),
       time((exp(10),crypto_data_hash(Ds, H1, Options),false)).
    %@    % CPU time: 0.710s, 7_942_187 inferences
    %@    % CPU time: 0.713s, 7_942_187 inferences
    %@    % CPU time: 0.712s, 7_942_187 inferences
    %@    % CPU time: 0.711s, 7_942_187 inferences
    %@    % CPU time: 0.710s, 7_942_187 inferences
    %@    % CPU time: 0.711s, 7_942_187 inferences
    %@    % CPU time: 0.710s, 7_942_187 inferences

    ?- length(_, L), time((exp(10),crypto_data_hash("test", "3caebd1a0a2647930319a660b7d3642eb380fbd43202f9f6d08aabaa9ba50c39522a12ead10f0423f0af613cbc6fea74ad682ee11f563cc2e735722004fda2ba", [algorithm(sha512),hmac([0,L])]),false)).
    %@    % CPU time: 0.733s, 7_878_699 inferences
    %@    % CPU time: 0.734s, 7_878_699 inferences
    %@    % CPU time: 0.732s, 7_878_699 inferences
    %@    % CPU time: 0.733s, 7_878_699 inferences
    %@    % CPU time: 0.733s, 7_878_699 inferences
    %@    % CPU time: 0.733s, 7_878_699 inferences
    %@    % CPU time: 0.733s, 7_878_699 inferences
    %@    % CPU time: 0.733s, 7_878_699 inferences
    %@    % CPU time: 0.733s, 7_878_699 inferences
    %@    % CPU time: 0.733s, 7_878_699 inferences
    %@    % CPU time: 0.733s, 7_878_699 inferences
    %@    % CPU time: 0.515s, 5_525_404 inferences
    %@    error('$interrupt_thrown',repl/0).
 2024-03-01 18:59:48 +01:00
Mark
ba362e2fe0 issue callable type error from dynamic_module_resolution if module is not an atom (#2345) 2024-02-28 20:35:38 -07:00
Markus Triska
1dd0c599c6 DOC: Add DocLog comments for reasoning about elliptic curves. 2024-02-24 20:43:57 +01:00
Markus Triska
27852eafd7 ADDED: Hash-based message authentication code (HMAC), using hmac(Key). 2024-02-22 20:53:27 +01:00
Markus Triska
8f2e9c6b94 FIXED: char_type/2 for unbound first argument. 
Surrogate pairs form a gap in valid character codes, see:

    https://github.com/mthom/scryer-prolog/issues/2326#issuecomment-1937864665

Many thanks to @Skgland for the pointer, and to @librarianmage for the
question that spawned this!

This addresses #2326.
 2024-02-11 22:19:19 +01:00
Mark
539a1aee2c fix tests broken by singleton 2024-02-07 21:46:32 -07:00
Markus Triska
53b7d9eec9 ENHANCED: Bidirectional char_type/2, addressing #2321. 
Suggested by @librarianmage, many thanks!
 2024-02-07 20:57:28 +01:00
Markus Triska
aa98a7e7d6 ADDED: clpz_t/2, generalizing support for use with library(reif) 
This was suggested and contributed by @librarianmage in:

    https://github.com/mthom/scryer-prolog/issues/2225#issuecomment-1890801923

Many thanks!

If anyone can find a better predicate name, please let us know any time!
 2024-02-06 19:05:29 +01:00
Adrián Arroyo Calle
ec4a8745e7 Add all_mdoules test and fix library(csv) 2024-02-04 21:34:01 +01:00
Markus Triska
3505cc3ba0 update answers using latest toplevel output 2024-01-25 21:33:27 +01:00
Markus Triska
05ba5f4358 rely on recently improved argument indexing for determinism 
This is possible as of 4fd37335f5.
 2024-01-25 21:31:08 +01:00
Mark
bc616ca7d8 move call_residue_vars/2 from atts.pl to iso_ext.pl 2024-01-25 12:40:24 -07:00
Markus Triska
a6ccf95076 improve error context 
Source: https://github.com/mthom/scryer-prolog/issues/2304#issuecomment-1906434756
 2024-01-23 21:43:34 +01:00
Markus Triska
665f319a0e no longer use (->)//2, since it is implementation defined 
This addresses #2304.
 2024-01-23 21:42:36 +01:00
Markus Triska
81dba11ab1 use newly available false//0 2024-01-21 22:53:22 +01:00
Markus Triska
388fa5baa9 no longer use (->)//2, since it is implementation dependent 2024-01-21 22:50:36 +01:00
Markus Triska
fe3241c07c remove checks that are now no longer needed due to representation errors 2024-01-21 10:28:19 +01:00
Markus Triska
0559ddca2a ENHANCED: Throw representation errors for DCG bodies that use unsupported constructs. 
This addresses #2285.
 2024-01-21 10:28:19 +01:00
Markus Triska
98a046500f MODIFIED: (->)/2 in DCGs is now only supported in the context of if-then-else. 
That is, ( If -> Then ; Else ) is still supported, but ( If -> Then ) not.
 2024-01-21 10:28:19 +01:00
Mark Thom
58cd0d1669 Merge pull request #2301 from aarroyoc/docs-copy-term-3 
Move copy_term/3 to library(iso_ext)
 2024-01-19 10:18:59 -07:00
Adrián Arroyo Calle
99c85459a7 Move copy_term/3 to library(iso_ext) 2024-01-19 17:33:45 +01:00
Adrián Arroyo Calle
958bf51648 Throw error in call_residue_vars/2 2024-01-19 16:58:26 +01:00
Mark Thom
6fb3b61441 Merge pull request #2276 from triska/hophrase 
ADDED: Support for higher-order non-terminals phrase//[2,3]
 2024-01-11 11:56:10 -07:00
notoria
902cd5c3ea Special case for (^)/2 
The only solutions for `Z #= X^Z` are `X = Z, X in -1\/1`. Also helps
avoid big numbers.
 2024-01-08 21:50:02 +01:00
Mark Thom
f9eadc8e6a Merge pull request #2278 from notoria/clpz 
Small fixes for `CLP(ℤ)`
 2024-01-07 18:59:38 -07:00
notoria
1f3de74cbd Fix bound computation in (^)/2 
?- A in 0..1, B in-1..0, A^2 #= B.
   false. % unexpected.
?- A^2 #= B, A in 0..1, B in-1..0.
   A = 0, B = 0.
 2024-01-07 20:19:40 +01:00
notoria
770ead9c05 Remove min/2 and max/2 when possible 
?- A #= min(A,B).
   clpz:(B#>=A), clpz:(A#=min(A,B)). % unexpected.
?- A #= min(A,0).
   clpz:(A in inf..0), clpz:(A#=min(A,0)). % unexpected.
 2024-01-07 20:15:28 +01:00
notoria
fe371ff1d1 Fix min/2 and max/2 
?- X in 0..1, Y in 2..4, Z #= max(X,Y).
   clpz:(X in 0..1), clpz:(Z#=max(X,Y)), clpz:(Z in 2..4), clpz:(Z#>=Y), clpz:(Y in 2..4). % unexpected.
?- X in 0..1, Y in 2..4, Z #= max(Y,X).
   Y = Z, clpz:(X in 0..1), clpz:(Y in 2..4).
 2024-01-07 20:14:32 +01:00
Markus Triska
2f783f0aef Correct arithmetic inequalities accidentally broken in f5d9a67f36. 
Example:

    ?- X #>= Y.
    %@    clpz:(X#>=Y).
 2024-01-07 09:33:54 +01:00
Mark Thom
92b85d4ba6 Merge pull request #2264 from Skgland/fix-os-argv-1 
Fix os:argv/1 in case the raw args contain multiple `--`
 2024-01-06 15:52:53 -07:00
Markus Triska
e702fe5c68 ADDED: Support for higher-order non-terminals phrase//[2,3] 
These non-terminals take a grammar rule body and additional arguments
as arguments. These arguments are appended to the first argument.

A key motivation for the introduction of these non-terminals is found
in the discussion and sample code provided by @bakaq in:

    https://github.com/mthom/scryer-prolog/discussions/2260

In this way, portable higher-order DCG programming is possible while
keeping the logical grammar rule expansion implementation dependent.

Example:

    ?- phrase(phrase('.', a, []), Cs).
       Cs = "a".
 2024-01-06 14:02:13 +01:00
Markus Triska
f2b63d1689 Prevent assertion of clauses for (-->)/2 when DCGs are enabled. 
This addresses #1508.
 2024-01-05 19:04:53 +01:00
Skgland
78b83ca1e1 simplify os:argv/1 fix 2024-01-04 19:26:10 +01:00
Skgland
ea0130d114 use once to limit solutions instead of cut 2024-01-04 18:14:32 +01:00
Mark Thom
5abc72cc8b Merge pull request #2268 from aarroyoc/docs-toplevel 
Small fixes in documentation and toplevel.pl
 2024-01-03 14:52:34 -07:00
Adrián Arroyo Calle
d58e91303b Small fixes in documentation and toplevel.pl 2024-01-03 19:52:50 +01:00
Skgland
8bd0317e9c fix os:argv(V) not completing 2024-01-02 21:43:38 +01:00