summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
...
* ffi: almost bug: 64 bit signed big/little endian type.Kaz Kylheku2019-01-231-4/+4
| | | | | | | | | | | | | | * ffi.c (ffi_be_i64_get, ffi_le_i64_get): The code here for 32 bit platforms is fishy: it is using the signed cnum type for the low 32 bits, when that should be ucnum. We are actually okay because ths code would only be executed on a platform where cnum is 32 bits. We are saved by the fact that we are doing left shifts (no sign extension), that shifting a 1 into a sign bit is harmless on two's complement machines, and that the lo32 value is ultimately passed to unum whereby it is coerced to a ucnum argument type (which preserves the bit representation) and prevents logior from seeing a negative value.
* Fix some instances of 4 bytes = 32 bits assumption.Kaz Kylheku2019-01-237-42/+48
| | | | | | | | | | | | | | | | | * hash.c (equal_hash, eql_hash, cobj_eq_hash_op, hash_hash_op): Multiply object size by CHAR_BIT and switch on number of bits, rather than bytes. * sysif.c (off_t_num): Likewise. * arith.c, ffi.c, itypes.c, rand.c: In numerous #if directive, fix size tests from bytes to bits. * configure: in the test that detects integer types, and in the test for enabling large file offsets, detect one more variable from the system: the value of CHAR_BIT. This turns into SIZEOF_BYTE. We use that value instead of a hard-coded 8.
* sysif: use double-intptr function from arith.Kaz Kylheku2019-01-233-5/+6
| | | | | | | | | | | * arith.c (bignum_dbl_ipt): Change internal function to external linkage. * arith.h (bignum_dbl_ipt): Conditionally declared. * sysif.c (num_off_t): Use bignum_dbl_ipt instead of open-coding exactly the same thing that bignum_dbl_ipt does. Abort is changed to same internal_error as in off_t_num.
* sysif: remove low-level MPI code for off_t conversion.Kaz Kylheku2019-01-232-50/+12
| | | | | | | | | | | * sysif.c (off_t_num): Retarget to just use c_i32 or c_i64 from ctypes.c, depending on which of these types is the same width as off_t. (stdio_fseek): Pass identifying "self" string to off_t_num. Don't use off_t_num for fseek's offset argument which is of type long; for that we can use c_long from ctypes.c. * sysif.h (off_t_num): Declaration updated.
* mpi: put access macros into mp_ namespaceKaz Kylheku2019-01-225-26/+34
| | | | | | | | | | | | | | | | | | * mpi/mpi.h (mp_sign, mp_isneg, mp_used, mp_alloc, mp_digits, mp_digit): New macros, based on renaming SIGN, ISNEG, USED, ALLOC, DIGITS and DIGIT. (mp_set_prec): Use mp_digits instead of DIGITS. * mpi/mpi.c (SIGN, ISNEG, USED, ALLOC, DIGITS, DIGIT): Macros defined here now, as local aliases for their namespaced counterparts. * arith.c (signum, floordiv): Replace ISNEG with mp_isneg. * rand.c (random): Replace ISNEG with mp_isneg. * sysif.c (off_t_num): Replace USED, DIGITS and ISNEG with mp_used, mp_digits and mp_isneg.
* doc: review of Lisp CompilationKaz Kylheku2019-01-201-3/+14
| | | | | | | | * txr.1: eval mistakenly referred to instead of compile-file under Compile File. Spurious pluralization of literals fixed. New "Bound symbols in dwim" heading to separate text from discussion of unbound symbols in dwim. Clarifying text added about compile-toplevel's treatment of form.
* Version 206.txr-206Kaz Kylheku2019-01-186-186/+206
| | | | | | | | | | * RELNOTES: Updated. * configure, txr.1: Bumped version and date. * share/txr/stdlib/ver.tl: Likewise. * txr.vim, tl.vim: Regenerated.
* mpi: some params should be mp_size, not mp_digit.Kaz Kylheku2019-01-182-9/+9
| | | | | | | | | * mpi/mpi.c (s_mp_2expt, mp_trunc_comp, mp_trunc, mp_bit): Arguments that give a number of bits or bit position should be mp_size; they are not related to digit values. * mpi/mpi.h (mp_trunc_comp, mp_trunc, mp_bit): Declaration updated.
* mpi: remove unused mplogic module.Kaz Kylheku2019-01-181-1/+1
| | | | * Makefile (MPI_OBJ_BASE): remove mplogic.o.
* mpi: hardening; bug found.Kaz Kylheku2019-01-184-10/+15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The flaw in the rand function caused by mp_count_ones is serious. For instance calls to (rand 3000000000) are expected to produce evenly distributed values in the range zero to one less than 3 billion. Due to this flaw, only values in the range [0, 2147483648) are produced. The function is wrongly informed that 3000000000 is a power of two, and so it subtracts one from the number of bits needed for the clamping bit mask, causing raw random values to be clamped to 31 bits instead of 32. * arith.c (logcount): Use mp_err to capture return value from mp_count_ones, rather than mp_size. If the result is less than zero, throw an internal error. (The only non-internal error indication that could come from this function is a memory allocation, and we already use chk_malloc under MPI). * mpi/mpi.c (mp_count_ones): Switch return type to mp_err, because this function returns error codes, which are negative. (mp_is_pow_two): Fix broken test that is always true. This affects our random number module; it's used in the random function for bignum moduli. (mp_toradix_case): Use unsigned char temporary for exchanging two unsigned chars, rather than a char temporary. (s_mp_div_d): Assert that the partial quotient t and remainder w, of mp_word_type, have values that fit into the mp_digit variables to which they are being assigned. (s_mp_ispow2d): The result of s_highest_bit(d) is unsigned, so if we subtract 1 from 0, we create a large value which likely converts to -1. Let's ensure that clearly with a cast to (int) of the return value. * mpi/mpi.h (mp_count_ones): Declaration updated. * mpi/mplogic.c (mpl_rsh, mpl_lsh): Fix two places where we shift the constant 1 (of type int) left by a number of bits that can exceed the type int. Note, that these functions are currently not called anywhere in TXR.
* mpi: use wchar_t string for text-to-bignum.Kaz Kylheku2019-01-183-8/+5
| | | | | | | | | | | | | * mpi/mpi.c (mp_read_radix): Take const wchar_t * string rather than unsigned char *. (s_mp_tovalue): Take character argument as wchar_t rather than int. * mpi/mpi.h (mp_read_radix): Declaration updated. * lib.c (int_str): Avoid a malloc/free and UTF-8 conversion by passing the original wide string to mp_read_radix. This removes a TODO dating back to December 2011.
* int-flo: take advantage of unsigned range.Kaz Kylheku2019-01-181-0/+3
| | | | | | | | * arith.c (int_flo): If the floating point value is in the ucnum range, we can convert to integer by way of that type; we gain a little bit of range. For instance on 32 bits, floating values in the range 2147483648.0 to 4294967295.0 can be converted via the fast path.
* configure: improvements related to int_ptr_t.Kaz Kylheku2019-01-181-9/+3
| | | | | | | | | | | | | | * configure: We remove the uintptr variable and do not generate the HAVE_UINTPTR_T constant in config.h. This is always 1, and not actually tested anywhere. In situations when uintptr is not established, intptr is also not established and the configure script fails. We simply assume that for whatever type we detect as inptr_t, we can derive the unsigned type. (INT_PTR_MAX): We define this a bit differently; instead of interpolating into the expression the underlying C type, we use the int_ptr_t typedef that the previous lines of config.h establish. (UINT_PTR_MAX): New constant introduced in config.h.
* int-flo: fix always-false test.Kaz Kylheku2019-01-181-1/+1
| | | | | | | | * arith.c (int_flo): Fix reversed inequality tests which cause the test to be always false. This always false test prevents use of the faster code path for converting floating-point point values that are in range of the cptr type.
* Fix bug in bignum single-digit addition.Kaz Kylheku2019-01-181-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Sigh; this is a continuation of the same topic that was addressed in November 14, 2016: "Fix bug in bignum addition". Commit SHA: c30a96120f53b960db56bc05a7ce6310bb2528f5. Though I fixed s_mp_add, s_mp_add_d has the same problem. Two digit-sized quantities are added together in the digit-sized type, and so the CARRYOUT(w) from the result is always zero. This bug causes the following consequences (using behavior under 32 bit as an example): 1. Wrong arithmetic: (succ (pred (expt 2 32)) -> 0 2. Wrong conversion from decimal: 4294967296 -> 0 4294967297 -> 1 4294967298 -> 2 4294967299 -> 3 As well as all values that begin with the above digit sequences. 4. Wrong conversion from floating-point. (toint 4294967296.0) -> 0 * mpi/mpi.c (s_mp_add_d): Add missing cast to the initial addition that adds the incoming digit d to the least significant limb of the bignum, so that the addition is done in the wider mp_word type, allowing CARRYOUT(w) to calculate a nonzero k value.
* carray: fix vec/list conversion bug.Kaz Kylheku2019-01-172-13/+31
| | | | | | | | | | | | | | | | | | | | If a zero-length carray is converted with vec-carray or list-carray and the null-term-p argument is t, there is an exception about a negative index. An empty vector or list should be returned in this case, and the documentation says exactly that. Also, if a carray of unknown length is converted, there is an exception from vec-carray, as documented, but it's an uninformative one that is incidentally produced when -1 is passed to the vec function. The list-carray just returns nil, contravening the documentation. * ffi.c (vec_carray, list_carray): Fix the problems described above. * txr.1: Reviewing the documentation for these functions, an improperly terminated sentence was found.
* Copyright year bump 2019.Kaz Kylheku2019-01-16107-108/+108
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * LICENSE, LICENSE-CYG, METALICENSE, Makefile, args.c, args.h, arith.c, arith.h, buf.c, buf.h, cadr.c, cadr.h, combi.c, combi.h, configure, debug.c, debug.h, eval.c, eval.h, ffi.c, ffi.h, filter.c, filter.h, ftw.h, gc.c, gc.h, glob.c, glob.h, hash.c, hash.h, itypes.c, itypes.h, jmp.S, lib.c, lib.h, lisplib.c, lisplib.h, match.c, match.h, parser.c, parser.h, parser.l, parser.y, protsym.c, rand.c, rand.h, regex.c, regex.h, share/txr/stdlib/asm.tl, share/txr/stdlib/awk.tl, share/txr/stdlib/build.tl, share/txr/stdlib/cadr.tl, share/txr/stdlib/compiler.tl, share/txr/stdlib/conv.tl, share/txr/stdlib/doloop.tl, share/txr/stdlib/error.tl, share/txr/stdlib/except.tl, share/txr/stdlib/ffi.tl, share/txr/stdlib/getopts.tl, share/txr/stdlib/getput.tl, share/txr/stdlib/hash.tl, share/txr/stdlib/ifa.tl, share/txr/stdlib/keyparams.tl, share/txr/stdlib/op.tl, share/txr/stdlib/package.tl, share/txr/stdlib/path-test.tl, share/txr/stdlib/place.tl, share/txr/stdlib/pmac.tl, share/txr/stdlib/socket.tl, share/txr/stdlib/stream-wrap.tl, share/txr/stdlib/struct.tl, share/txr/stdlib/tagbody.tl, share/txr/stdlib/termios.tl, share/txr/stdlib/trace.tl, share/txr/stdlib/txr-case.tl, share/txr/stdlib/type.tl, share/txr/stdlib/vm-param.tl, share/txr/stdlib/with-resources.tl, share/txr/stdlib/with-stream.tl, share/txr/stdlib/yield.tl, signal.c, signal.h, socket.c, socket.h, stream.c, stream.h, struct.c, struct.h, strudel.c, strudel.h, sysif.c, sysif.h, syslog.c, syslog.h, termios.c, termios.h, txr.1, txr.c, txr.h, unwind.c, unwind.h, utf8.c, utf8.h, vm.c, vm.h, vmop.h, win/cleansvg.txr: Extended Copyright line to 2018.
* doc: with-dyn-lib mention other deffi-* macros.Kaz Kylheku2019-01-161-4/+18
| | | | | * txr.1: with-dyn-lib relates not only to deffi but also deffi-var and deffi-sym.
* ffi: diagnose missing with-dyn-lib.Kaz Kylheku2019-01-161-24/+23
| | | | | | | | | | | | | | | If the programmer writes a deffi, deffi-sym or deffi-var that makes simple references, and that macro is not wrapped in with-dyn-lib, an unhelpful error message results about sys:ffi-lib being unbound. We can detect this situation and provide a warning. * share/txr/stdlib/ffi.tl (sys:with-dyn-lib-check, sys:expand-sym-ref): New functions. (deffi, deffi-sym, deffi-var): Capture environment parameter. Common code replaced by call to sys:expand-sym-ref, where the missing sys:ffi-lib situation is diagnosed. This is only done in cases when the simple reference syntax occurs.
* Version 205.txr-205Kaz Kylheku2019-01-156-274/+300
| | | | | | | | | | * RELNOTES: Updated. * configure, txr.1: Bumped version and date. * share/txr/stdlib/ver.tl: Likewise. * txr.vim, tl.vim: Regenerated.
* ffi: arrays: be more forgiving of length mismatches.Kaz Kylheku2019-01-152-9/+38
| | | | | | | | | | * ffi.c (min): New macro. (ffi_array_put_common): Tolerate sequences which are shorter than the array. Use seq_info to classify the sequence and use separate code for the vector and list case, avoiding taking the length of the list. * txr.1: Documented.
* ffi: bugfix: char array shouldn't null terminate.Kaz Kylheku2019-01-151-1/+3
| | | | | | | | | | | * ffi.c (ffi_char_array_put): The char array put operation should only null terminate when the null_term flag is set; i.e. it's a zarray type. The bug here is that when a Lisp string of length > N is put into an (array N char), the C array gets null terminated, which is wrong. Only in the case when the string is exactly of length N is there no null termination. In all cases when the length >= N, we want truncation without null termination.
* ffi: remove useless loop.Kaz Kylheku2019-01-151-12/+6
| | | | | | * ffi.c (make_ffi_type_array): Remove a useless loop that was left behind when the essential part of its body was removed.
* asm: fix wrong level check in operand parsing.Kaz Kylheku2019-01-091-1/+1
| | | | | | | | * share/txr/stdlib/asm.tl (parse-compound-operand): The index of a (v <lev> <index>) operand must be compard to the maximum index constant, not to the maximum level constant. This bug will prevent compiling frames that have more than 64 variables, which is a serious limitation from 1024.
* asm: fix incorrect level check in frame opcode.Kaz Kylheku2019-01-091-2/+2
| | | | | | * share/txr/stdlib/asm.tl (op-frame): Check the level of the operand against %max-v-lev%, not %max-lev-idx%, which gives the maximum index within a level.
* asm/compiler: fix incorrect frame-related constant.Kaz Kylheku2019-01-091-1/+1
| | | | | | * share/txr/stdlib/vm-param.tl (%max-lev-idx%): The maximum index within a level is one less than the maximum level size, not two less. Use pred rather than ppred to derive it.
* asm/compiler: rename small level/index constants.Kaz Kylheku2019-01-092-3/+4
| | | | | | | | | | | | | | | | * share/txr/stdlib/vm-param.tl (%max-sm-lev-idx%): This constant is named inconsistently relative to %max-lev-idx%. It is providing the maximum level (encodable in a small operand), whereas %max-lev-idx% provides the maximum index within a level. It is hereby renamed to %max-sm-lev%. The %max-sm-lev-idx% name is re-used to denote the quantity which it suggests: the maximum index within a level (encodable in a small operand), which is 63. * share/txr/stdlib/asm.tl (small-op-p): Use the new %max-sm-lev-idx% in place of %sm-lev-size%, getting rid of the funny range starting with -1. Replace the original %max-sm-lev-idx% with its new name, %max-sm-lev%.
* New function: square.Kaz Kylheku2019-01-054-0/+75
| | | | | | | | | | | | | The square function calulates (* x x) but is faster for bignum integers by taking advantage of mp_sqr. * arith.c (square): New function. * eval.c (eval_init): Register square as intrinsic. * lib.h (square): Declared. * txr.1: Documented.
* Eliminate ALLOCA_H.Kaz Kylheku2018-12-3119-22/+22
| | | | | | | | | | | | | * configure: Instead of generating a definition of ALLOCA_H, generate the variable HAVE_ALLOCA_<name> with a value of 1, where <name> is one of stdlib, alloca or malloc. * alloca.h: New header. * args.c, eval.c, ffi.c ffi.c, ftw.c, hash.c, lib.c, match.c, parser.c, parser.y, regex.c, socket.c, stream.c, struct.c, sysif.c, syslog.c, termios.c, unwind.c, vm.c: Include "alloca.h" instead of ALLOCA_H.
* Get alloca from stdlib.h, if possible.Kaz Kylheku2018-12-311-1/+1
| | | | | | | | | | * configure: try <stdlib.h> first for alloca. This should fix a build issue which happens on the Musl library and perhaps elsewhere. The problem on Musl is that #include <stdlib.h> already includes <alloca.h>. That header contains an alloca macro which interfers with our subsequent #include ALLOCA_H directive: ALLOCA_H expands to <alloca.h> and the alloca token gets further expanded.
* Remove HAVE_ALLOCA.Kaz Kylheku2018-12-311-1/+0
| | | | | * configure: don't define HAVE_ALLOCA in config.h. It is not used anywhere. Moreover, alloca isn't optional.
* doc: formatting under lop.Kaz Kylheku2018-12-241-2/+2
| | | | | | * txr.1: Closing quote didn't come out in .codn; we must use the \(dq code in this context. Let's make the opening quote \(dq also for consistency.
* doc: grammar fixes under registeer-tentative-def.Kaz Kylheku2018-12-241-4/+2
| | | | | * txr.1: Wrong tense/person of "to expect"; spurious occurrence of foo identifier deleted.
* doc: formatting under caseql*.Kaz Kylheku2018-12-241-1/+1
| | | | * txr.1: .meti should be used to typeset code inline.
* Version 204.txr-204Kaz Kylheku2018-12-176-464/+491
| | | | | | | | | | * RELNOTES: Updated. * configure, txr.1: Bumped version and date. * share/txr/stdlib/ver.tl: Likewise. * txr.vim, tl.vim: Regenerated.
* UTF-8: fix incorrect decoding of four-byte sequences.Kaz Kylheku2018-12-171-1/+1
| | | | | | | | | utf8.c (utf8_decode): The wch_min value is set incorrectly for the four byte case due to an extra zero; it should be only 0x10000. Code points encoded to four utf8 bytes start at this value. The consequence of this error is that utf8-encoded characters in this range are treated as invalid bytes after being decoded due to failing the range test.
* defvar: bugfix: bad state caused by exception.Kaz Kylheku2018-12-141-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | The following situation can occur: (defvar v expr) ;; expr throws! What happens is that the internal hash table of global variables ends up with an entry that has a nil value, instead of the expected (v . <value>) cons cell. The hash table entry is created when the table is probed for the existence of v, but then is never populated with a binding because expr throws. The consequence then is that (boundp v) returns nil, yet subsequent (defvar v ...) expressions think that v exists, and refuse to define it. (defparm v ...) also fails; it relies on (defvar ...) which fails to define the variable, and then tries to assign to it, which throws. * eval.c (rt_defvarl): If the hash table exists, but has a nil value, treat that as an undefined variable. Thus, define the variable not only if the hash cell is newly made, but also if it already exists, with a null cdr.
* compiler: move variable down in file.Kaz Kylheku2018-12-131-2/+2
| | | | | | * share/txr/stdlib/compiler.tl (assumed-fun): This variable shouldn't be the first item in the compiler. It is moved after the definitions of structs and important constants.
* nzerop: new function.Kaz Kylheku2018-12-134-1/+48
| | | | | | | | | | * arith.c (nzerop): New function. * eval.c (eval_init): Register nzerop intrinsic. * lib.h (nzerop): Declared. * txr.1: Documented.
* Use tnil instead of if2 in arithmetic predicates.Kaz Kylheku2018-12-131-9/+9
| | | | | * arith.c (zerop, plusp, minusp): style: if2(expr, t) should be using tnil(expr).
* plusp: wrong name in error message.Kaz Kylheku2018-12-131-1/+1
| | | | * arith.c (plusp): plusp wrongly reports itself as zerop.
* configure: extend some config variable descriptions.Kaz Kylheku2018-12-121-2/+11
| | | | | | | * configure: Add some help text for variables controllinig the compiler command. Mention that CFLAGS and LDFLAGS are honored from the environment or make command line on top of any of the flags settable here.
* Turn off stack protector gcc option.Kaz Kylheku2018-12-121-1/+1
| | | | | | | * configure (opt_flags): Add -fno-stack-protector to disable this feature that is on by default in some toolchains/distros. As a result, I'm seeing an over 5% VM speedup on a loop benchmark.
* bugfix: LDFLAGS not pulled in.Kaz Kylheku2018-12-121-1/+1
| | | | | * Makefile (TXR_LDFLAGS): Fix use of wrong assignment operator which clobbers the previous assignment of $(LDFLAGS).
* Eliminate various unneeded header inclusions.Kaz Kylheku2018-12-124-7/+0
| | | | | | | | | | * hash.c: Don't include "cadr.h". * stream.c: Don't include "struct.h". * strudel.c: No <syslog.h>, ALLOCA_H, "args.h" or "utf8.h". * vm.c: No "hash.h".
* Drastically reduce inclusion of <dirent.h>.Kaz Kylheku2018-12-1125-25/+1
| | | | | | | | | | | | | | | | | | | The <dirent.h> header is included all over the place because it is needed by a single declaration in stream.h. That declaration is for a function that is only called within stream.c, so we make it internal. Now only stream.c has to include <dirent.h>. * buf.c, debug.c, eval.c, ffi.c, filter.c, gc.c, gencadr.txr, hash.c, lib.c, lisplib.c, match.c, parser.c, regex.c, socket.c, struct.c, strudel.c, sysif.c, syslog.c, termios.c, txr.c, unwind.c, vm.c: Remove #include <dirent.h>. * cadr.c: Regenerated. * stream.c (make_dir_stream): Make external function static. * stream.h (make_dir_stream): Declaration updated.
* Version 203.txr-203Kaz Kylheku2018-11-296-593/+611
| | | | | | | | | | * RELNOTES: Updated. * configure, txr.1: Bumped version and date. * share/txr/stdlib/ver.tl: Likewise. * txr.vim, tl.vim: Regenerated.
* doc: bad formatting under --noninteractiveKaz Kylheku2018-11-281-1/+3
| | | | * txr.1: Fix .code macro occurring in middle of line.
* New range testing functions.Kaz Kylheku2018-11-274-0/+76
| | | | | | | | | | | * eval.c (eval_init): Register in-range and in-range* intrinsics. * lib.c (in_range, in_range_star): New functions. * lib.h (in_range, in_range_star): Declared. * txr.1: Documented.
* case macros: bugfixes for evaluated keys.Kaz Kylheku2018-11-271-5/+5
| | | | | | | | | * eval.c: Fix incorrect treatment of cases like (caseql* ((a b) ...)). The check for evaluation must come before we do the keys = car(keys) transformation. Also, hash_keys must be updated to the evaluated keys, otherwise if the hash table optimization is used, the table will contain the original expressions as the keys, not their values.
generated by cgit v1.2.3 (git 2.25.1) at 2025-03-20 18:38:24 +0000