summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
...
* signals: disable stack overflow in handler.Kaz Kylheku2021-06-242-0/+15
| | | | | | | | | | | * signal.c (sig_handler): For a is_cpu_exception signal, we temporarily disable the stack limit. It might be executing on the sigaltstack buffer, which is almost certainly below the stack limit. * tests/012/stack.tl: New test case. We raise a SIGSEGV and check that in the handler, the stack limit is disabled, and that we can executed code.
* txr: stack protection in pattern language.Kaz Kylheku2021-06-243-0/+13
| | | | | | | | | * txr.c (do_match_line, match_files): call gc_stack_check on entry. * tests/012/stack2.txr: New file. * tests/012/stack2.expected: New file.
* limit print depth/width when diagnosing oveflow.Kaz Kylheku2021-06-241-0/+11
| | | | | | | | | * eval.c (error_trace): If the error is a stack overflow, then save the printing depth and width, and set them to stringent values, to minimize recursion in the printer. This minimizes the chances of a segfault or runaway iteration under some conditions. A repro test case is (print '#1=(#1#)) entered into the listener.
* doc: standardize on "user-defined".Kaz Kylheku2021-06-241-7/+7
| | | | | * txr.1: Replace a few occurrences of "application-defined" with the more prevalent "user-defined", including in one heading.
* install-tests: use relative path in run.sh.Kaz Kylheku2021-06-241-1/+1
| | | | | | | | * Makefile (install-tests): In the generated run.sh, let's allow flexibility in the installation location of the tests by avoiding a cd to an absolute path where the tests are assumed to be installed. Let's assume that they are installed relative to where the run.sh script is, and cd there.
* parser: no string allocation when scanning floats.Kaz Kylheku2021-06-244-195/+211
| | | | | | | | | | | | | | | | | | | Each time the scanner processes a floating-point token, it allocates a string object, just so it can call flo_str. The object is then garbage. Let's stop doing that. * lib.c (flo_str_utf8): New function, closely based on flo_str. Takes a char * string. * lib.h (flo_str_utf8): Declared. * parser.l (out_of_range_float): Take the token as a const char * string instead of a Lisp string, so we can just pass yytext to this function. (grammar): Use flo_str_utf8 instead flo_str, and pass yytext to out_of_range_float. * lex.yy.c.shipped: Updated.
* Test for stack overflow protection.Kaz Kylheku2021-06-242-0/+34
| | | | | | * tests/012/stack.tl: New file. * tets/common.tl (mvtest): New macro.
* file layout: moving share/txr/stdlib to stdlib.Kaz Kylheku2021-06-2453-13/+32
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This affects run-time also. Txr installations where the executable is not in directory ending in ${bindir} will look for stdlib rather than share/txr/stdlib, relative to the determined installation directory. * txr.c (sysroot_init): If we detect relative to the short name, or fall back on the program directory, use stdlib rather than share/txr/stdlib as the stdlib_path. * INSTALL: Update some installation notes not to refer to share/txr/stdlib but stdlib. * Makefile (STDLIB_SRCS): Refer to stdlib, not share/txr/stdlib. (clean): In unconfigured mode, remove the old share/txr/stdlib entirely. Remove .tlo files from stdlib. (install): Install lib materials from stdlib. * txr.1: Updated documentation under Deployment Directory Structure. * share/txr/stdlib/{asm,awk,build,cadr}.tl: Renamed to stdlib/{asm,awk,build,cadr}.tl. * share/txr/stdlib/{compiler,conv,copy-file,debugger}.tl: Renamed to stdlib/{compiler,conv,copy-file,debugger}.tl. * share/txr/stdlib/{defset,doc-lookup,doc-syms,doloop}.tl: Renamed to stdlib/{defset,doc-lookup,doc-syms,doloop}.tl. * share/txr/stdlib/{each-prod,error,except,ffi}.tl: Renamed to stdlib/{each-prod,error,except,ffi}.tl. * share/txr/stdlib/{getopts,getput,hash,ifa}.tl: Renamed to stdlib/{getopts,getput,hash,ifa}.tl. * share/txr/stdlib/{keyparams,match,op,optimize}.tl: Renamed to stdlib/{keyparams,match,op,optimize}.tl. * share/txr/stdlib/{package,param,path-test,pic}.tl: Renamed to stdlib/{package,param,path-test,pic}.tl. * share/txr/stdlib/{place,pmac,quips,save-exe}.tl: Renamed to stdlib/{place,pmac,quips,save-exe}.tl. * share/txr/stdlib/{socket,stream-wrap,struct,tagbody}.tl: Renamed to stdlib/{socket,stream-wrap,struct,tagbody}.tl. * share/txr/stdlib/{termios,trace,txr-case,type}.tl: Renamed to stdlib/{termios,trace,txr-case,type}.tl. * share/txr/stdlib/{ver,vm-param,with-resources,with-stream}.tl: Renamed to stdlib/{ver,vm-param,with-resources,with-stream}.tl. * share/txr/stdlib/yield.tl: Renamed to stdlib/yield.tl. * share/txr/stdlib/{txr-case,ver}.txr: Renamed to stdlib/{txr-case,ver}.txr. * gencadr.txr: Update to stdlib/place.tl. * genman.txr: Update to stdlib/cadr.tl.
* matcher: new looping macros.Kaz Kylheku2021-06-245-0/+163
| | | | | | | | | | | | | | * lisplib.c (match_set_entries): Autoload on new while-match, while-match-case and while-true-match-case symbols. * share/txr/stdlib/match.tl (while-match, while-match-case, while-true-match-case): New macros. * tests/011/patmatch.tl: Tests. * txr.1: Documented. * share/txr/stdlib/doc-syms.tl: Updated.
* matcher: remove unused gensym.Kaz Kylheku2021-06-241-1/+0
| | | | | * share/txr/stdlib/match.tl (if-match): match-p gensym is not used; remove it.
* New: stack overflow protection.Kaz Kylheku2021-06-2310-3/+169
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * configure: detect getrlimit, producing HAVE_RLIMIT in config.h. * eval.c (do_eval, do_expand): Call gc_stack_check inline function to check stack pointer against limit. * gc.c (gc_stack_bottom): Static becomes extern, so inline function in gc.h can refer to it. (gc_stack_limit): New global variable. (gc_init): If we have rlimit, then probe RLIMIT_STACK. If the stack is sufficiently large, then enable the stack overflow protection, which kicks in when the stack pointer appears to be within a certain percentage of the limit. (set_stack_limit, get_stack_limit): New static functions. (gc_late_init): Register set-stack-limit and get-stack-limit intrinsics. (gc_stack_overflow): New function. * gc.h (gc_stack_bottom, gc_stack_limit, gc_stack_overflow): Declared. (gc_stack_check): New inline function. * lib.c (stack_overflow_s): New symbol variable. (obj_print_impl): Call gc_stack_check to protect recursive printing againts overflow. * lib.h (stack_overflow_s): Declared. * unwind.c (uw_init): Register stack-overflow symbol as a an exception symbol subtyped from error. (uw_unwind_to_exit_point): When dealing with an unhandled exception, turn off the stack limit, so we can print the messages without triggering it in a loop. * vm.c (vm_execute_closure, vm_funcall_common): Insert gc_stack_check to the top of the execution of every VM function. * txr.1: Documented. * share/txr/stdlib/doc-syms.tl: Updated.
* lib: use existing self variable.Kaz Kylheku2021-06-231-2/+2
| | | | | * lib.c (populate_obj_hash): Refer to self, rather than hard coding function name prefix.
* cmp_str: whitespace.Kaz Kylheku2021-06-231-37/+37
| | | | * lib.c (cmp_str): Fix incorrect indentation.
* long-suffix: fix crash reproducing on Cygwin.Kaz Kylheku2021-06-231-3/+1
| | | | | | * stream.c (long_suffix): Remove stray wcspbrk(dot, psc) call from the body of loop, which sometimes occurs when dot is null. It may have been optimized away, so I didn't notice.
* cyr: broken on 64 bit.Kaz Kylheku2021-06-231-2/+3
| | | | * lib.c (cyr): We must use the cnum type for the mask, not int.
* c_str now takes a self argument.Kaz Kylheku2021-06-2321-231/+297
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Adding a self parameter to c_str so that when a non-string occurs, the error is reported against a function. Legend: A - Pass existing self to c_str. B - Define self and pass to c_str and possibly other functions. C - Take new self parameter and pass to c_str and possibly other functions. D - Pass existing self to c_str and/or other functions. E - Define self and pass to other functions, not c_str. X - Pass nil to c_str. * buf.c (buf_strm_put_string, buf_str): B. * chksum.c (sha256_str, md5_str): C. (sha256_hash, md5_hash): D. * eval.c (load): D. * ffi.c (ffi_varray_dynsize, ffi_str_put, ffi_wstr_put, ffi_bstr_put): A. (ffi_char_array_put, ffi_wchar_array_put): C. (ffi_bchar_array_put): A. (ffi_array_put, ffi_array_out, ffi_varray_put): D. * ftw.c (ftw_wrap): A. * glob.c (glob_wrap): A. * lib.c (copy_str, length_str, coded_length,split_str_set, list_str, cmp_str, num_str, out_json_str, out_json_rec, display_width): B. (upcase_str, downcase_str, string_extend, search_str, do_match_str, do_rmatch_str, sub_str, replace_str, cat_str_append, split_str_keep, trim_str, int_str, chr_str, span_str, compl_span_str, break_str, length_str_gt, length_str_ge, length_str_lt, length_str_le, find, rfind, pos, rpos, mismatch, rmismatch): A. (c_str): Add self parameter and use in type mismatch diagnostic. If the parameter is nil, use "internal error". (flo_str): B, and correction to "flot-str" typo. (out_lazy_str, out_quasi_str, obj_print_impl): D. * lib.h (c_str): Declaration updated. * match.c (dump_var): X. (v_load): D. * parser.c (open_txr_file): C. (load_rcfile): E. (find_matching_syms, provide_atom): X. (hist_save, repl): B. * parser.h (open_txr_file): Declaration updated. * parser.y (chrlit): X. * regex.c (search_regex): A. * socket.c (getaddrinfo_wrap, sockaddr_pack): A. (dgram_put_string): B. (open_sockfd): D. (sock_connect): E. * stream.c (stdio_put_string, tail_strategy, vformat_str, open_directory, open_file, open_tail, remove_path, rename_path, tmpfile_wrap, mkdtemp_wrap, mkstemp_wrap): B. (do_parse_mode, parse_mode, make_string_byte_input_stream): B. (normalize_mode, normalize_mode_no_bin): E. (string_out_put_string, formatv, put_string, open_fileno, open_subprocess, open_command, base_name, dir_name, short_suffix, long_suffix): A. (run): D. (win_escape_cmd, win_escape_arg): X. * stream.h (parse_mode, normalize_mode, normalize_mode_no_bin): Declarations updated. * sysif.c (mkdir_wrap, do_utimes, dlopen_wrap, dlsym_wrap, dlvsym_wrap): A. (do_stat, do_lstat): C. (mkdir_nothrow_exists, ensure_dir): E. (chdir_wrap, rmdir_wrap, mkfifo_wrap, chmod_wrap, symlink_wrap, link_wrap, readlink_wrap, exec_wrap, getenv_wrap, setenv_wrap, unsetenv_wrap, getpwnam_wrap, getgrnam_wrap, crypt_wrap, fnmatch_wrap, realpath_wrap, opendir_wrap): B. (stat_impl): statfn pointer-to-function argument now takes self parameter. When calling it, we pass name. * syslog.c (openlog_wrap, syslog_wrapv): A. * time.c (time_string_local, time_string_utc, time_string_meth, time_parse_meth): A. (strptime_wrap): B. * txr.c (txr_main): D. * y.tab.c.shipped: Updated.
* short-suffix, long-suffix: no match across slash.Kaz Kylheku2021-06-223-8/+96
| | | | | | | | | | | * stream.c (short_suffix, long_suffix): Take path separator characters into account; the suffix must not span across separators. The trailing separator must also not appear in the suffix. * tests/018/path.tl: Test cases added. * txr.1: Redocumented.
* lib: rmismatch tests and bugfix.Kaz Kylheku2021-06-222-1/+40
| | | | | | | * lib.c (rmismatch): when left is an empty string or vector, and right is nil: we must return -1 not zero. * tests/012/seq.tl: More rmismatch tests.
* lib: optimize mismatch, rmismatch for strings.Kaz Kylheku2021-06-222-3/+95
| | | | | | | | | | | * lib (mismatch, rmismatch): If the arguments are strings or literals, other than lazy strings, keyfun is identity, and equality is by character identity, the operation can be done with an efficient loop over the wchar_t strings. * tests/012/seq.tl: Tests for string case of mismatch, via starts-with function. Test mismatch via ends-with, and also directly for vectors and strings.
* New functions: long-suffix, short-suffix.Kaz Kylheku2021-06-225-0/+116
| | | | | | | | | | | | | | * stream.c (short_suffix, long_suffix): New functions. (stream_init): short-suffix and long-suffix intrinsics registered. * stream.c (short_suffix, long_suffix): Declared. * tests/018/path.tl: New file. * txr.1: Documented. * share/txr/stdlib/doc-syms.tl: Updated.
* doc: formatting in quasiquote pattern matching.Kaz Kylheku2021-06-221-1/+1
| | | | | | * txr.1: Remove stray space in marking up the JSON hash pattern syntax. This space has no ill effect when reading the man page via nroff, but messes up HTML rendering.
* Dubious new functions cxr/cyr.Kaz Kylheku2021-06-216-0/+141
| | | | | | | | | | | | | | * lib.c (cxr, cyr): New functions. * lib.h (cxr, cyr): Declared. * eval.c (eval_init): Intrinsics cxr and cyr registered. * tests/012/cadr.tl: New file. * txr.1: Documented. * share/txr/stdlib/doc-syms.tl: Updated.
* lib: remove useless coerce.Kaz Kylheku2021-06-211-1/+1
| | | | | * lib.c (string): chk_strdup returns the wchar_t * type already; there is no need to coerce.
* subprocesses: cygwin fixes.Kaz Kylheku2021-06-202-5/+11
| | | | | | | | | | | | | | * stream.c (fds_subst, fds_swizzle, fds_restore): These functions are used on Cygwin by some functions that are based on spawn, and so must be avaiable. * tests/018/process.tl: Two fixes here. Firstly, the 1>&1 redirection in one of the tests should be 1>&2 on all platforms. That a temporary edit left behind from massaging the tests and the code. Secondly, some weird results are observed on Cygwin. The communication with the child sees an extra carriage return and space sneak in. I'm just codifying this as Cygwin's way of passing the test case; it's likely unfixable at the application level.
* glob: support sequence of patterns argument.Kaz Kylheku2021-06-202-12/+53
| | | | | | | | | | | * glob.c (glob_wrap): Negate the GLOB_WRAP flag should it be present on input. If the input is a string, process it much like before. Otherwise loop over the pattern and call globl multiple times, adding the GLOB_APPEND flag for the second and subsequent times, and breaking out of the loop if a control transfer is detected. * txr.1: Document.
* doc: recent typo.Kaz Kylheku2021-06-201-1/+1
| | | | | | * txr.1: Fix typo introduced during recent work related to txr-case. aspell caught this but I overwrote the fix with a stale edit buffer.
* subprocesses: a few test cases.Kaz Kylheku2021-06-201-0/+18
| | | | * tests/018/process.tl: New file.
* read/get-json: reject trailing junk in string input.Kaz Kylheku2021-06-204-4/+106
| | | | | | | | | | | | | | | | | * parser.c (lisp_parse_impl): If parsing from string, check for trailing junk and diagnose. JSON parsing doesn't use lookahead because it doesn't have a.b syntax, so the recent_tok gives the last token that actually went into the syntax, and not a lookahead token. So in the case of JSON, we call yylex to see if there is any trailing token. * tests/010/json.tl: Extend get-json tests to more kinds of objects, and then replicate with trailing whitespace and trailing junk to provide coverage for these cases. * tests/012/parse.t: Slew of new read tests and iread also. * txr.1: Documented.
* streams: pipe streams not needed if we have fork.Kaz Kylheku2021-06-192-3/+7
| | | | | | | | | | * stream.c (se_pclose, make_pipe_stream): define only if !HAVE_FORK_STUFF. (pipe_close): Don't call se_pclose in HAVE_FORK_STUFF build. The pipe_ops are no longer shared between popen pipes and open-process pipes. * stream.h (make_pipe_stream): Conditionally declare.
* system: stop using popen if we have fork.Kaz Kylheku2021-06-193-35/+62
| | | | | | | | | | | | | | | | | | | | | | | | On all the platforms on which TXR is regularly built, there is fork, including the Windows port via Cygwin. We therefore don't have to be using popen for the open-command function. The handling the stream redirection will be more efficient if we implement open-command in terms of open-process, and the special mode string extensions for redirection like ">21" will work, too. * stream.c (fds_subst, fds_swizzle): Define these functions only for !HAVE_FORK_STUFF platforms. (open_command): popen-based implementation moved down into a section of code for !HAVE_FORK_STUFF platforms. A separate implementation is written for HAVE_FORK_STUFF platforms which passes the command to an appropriate Windows native or POSIX interpreter. * utf8.c (w_popen): Nothing but open-command calls this function, so we don't need it on HAVE_FORK_STUFF platforms. Wrap with #if. * utf8.h (w_popen): wrap with #if.
* Version 263txr-263Kaz Kylheku2021-06-177-828/+882
| | | | | | | | | | | | * RELNOTES: Updated. * configure, txr.1: Bumped version and date. * share/txr/stdlib/ver.tl: Bumped. * txr.vim, tl.vim: Regenerated. * protsym.c: Likewise.
* sysif: define passwd_s if we have <grp.h>.Kaz Kylheku2021-06-171-1/+5
| | | | | | * sysif.c (passwd_s): Define for either HAVE_PWUID or HAVE_GRGID, since this symbol variable is referenced in code protected by either preprocessor symbol.
* android: fix compiler noise about __ANDROID_API__Kaz Kylheku2021-06-171-1/+1
| | | | | | | | * configure: Termux's gcc now defines __ANDROID_API__ which clashes with our definition. We cannot remove our definition because Termux's uses a uselessly low, conservative value. Thus, add a -U to undefine it before our -D redefinition.
* op: add comment why cadr is not used.Kaz Kylheku2021-06-171-0/+1
| | | | | | | * share/txr/stdlib/op.tl (sys:op-expand): Comment added. We cannot use cadr because that triggers autoload of cadr.tl, which uses defplace, which requires place.tl, which requires op, which is what we are in the middle of defining.
* doc: bad grammar under FFI ptr typeKaz Kylheku2021-06-171-1/+1
| | | | * txr.1: "update *the* datum".
* doc: formatting of meta-symbols under op.Kaz Kylheku2021-06-171-16/+26
| | | | | | * txr.1: Fix the typesetting of @rest, @rec, @num, @n and @n-1 in the op macro section. Missing word inserted: "recursive call *to* the function".
* op: tests, and fix (op progn ...) situationKaz Kylheku2021-06-172-1/+73
| | | | | | | | | | * share/txr/stdlib/op.tl (op-expand): For the sake of special processing applied to support the lop operator, the code assumes that the expanded syntax-2 is a list with at least two elements, such that we can do (cddr syntax-2). This is not true for instance in (op progn). * tests/012/op.tl: New file.
* New macro: flow.Kaz Kylheku2021-06-164-1/+45
| | | | | | | | | | | | | | | There has been interest in the community in a threading macro; a triviality which which applies the opip syntax to a value. Eschewing unusual glyph combinations like ->> and whatnot, I've chosen the English word flow for this. * share/txr/stdlib/op.tl (flow): New macro. * lisplib.c (op_set_entries): Set up autoload for flow. * txr.1: Documented. * share/txr/stdlib/doc-syms.tl: Updated.
* listener: new --noprofile option.Kaz Kylheku2021-06-164-2/+16
| | | | | | | | | | | | | * parser.c (repl): Set the rcfile variable to nil if opt_noprofile is true, to suppress reading it. * txr.c (op_noprofile): New global variable. (help): Add help text. (txr_main): Recognize noprofile option and set variable. * txr.h (opt_noprofile): Declared. * txr.1: Documented.
* txr-case: upkeep.Kaz Kylheku2021-06-164-5/+48
| | | | | | | | | | | | | | | | | | | * share/txr/stdlib/txr-case.tl (txr-case-impl): If the input is a stream, then convert it to a lazy list of lines, so that running multiple functions against it produces sane, backtracking behavior, like a @(cases) construct. * tests/011/txr-case.expected: Updated. * tests/011/txr-case.txr: Now actually contains a test case for txr-case. * txr.1: Address an issue reported by Paul A. Patience: the input to match-fun, txr-if and txr-when may be a stream. That has always been the case in the implementation. Also document that when the input is a single string, it is treated as a list. Document the new requirement in txr-case that a stream is converted into lazy list of lines.
* math: forbid dubious inequality comparisons.Kaz Kylheku2021-06-153-1/+71
| | | | | | | | | | | | | | | | | | | The issue, reported by Paul A. Patience, is that code like (< 1 "abc") is successfully producing a result. The root cause is that 1 is an iterable object, and so is treated as a sequence opposite to the "abc" operand. We should allow only true sequences in this situation. * arith.c (seq_lt_compat_check): New static function. Checks that neither of two sequences is SEQ_NOTSEQ or SEQ_HASHLIKE. (seq_lt, seq_le): Use seq_lt_compat_check to reject dubious inputs. * txr.1: Minor wording change in the related documentation, removing a gratuitous adjective. * tests/016/arith.tl: Inequality tests.
* subprocesses: don't bother saving descriptors.Kaz Kylheku2021-06-151-2/+21
| | | | | | | | | | * stream.c (fds_subst_nosave): New static function. (fds_clobber): New static function. Like fds_swizzle, without the saving. (open_subprocess, run): Use fds_clobber instead of fds_swizzle in the child process. It makes no sense to use fds_swizzle, which saves duplicates of the old descriptors, if fds_restore is not going to be called.
* subprocesses: move fds_swizzle to child process.Kaz Kylheku2021-06-151-29/+41
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | In all cases in which we control fork and exec, we should be doing the fds_swizzle setup in the child process. This has several benefits. We do not disturb the file descriptor layout of the parent. We don't have to set up a catch to do the cleanup. We don't have to do the clean-up in the child either; but just let it terminate. * stream.c (struct save_fds): New members subin, subout and suberr to hold the substitute file descriptors. The reason for this is so that we can calculate these in the parent process so that all the error checking is done in the parent and and carry them over to the child somehow. This structure is the natural place for that. (fds_getfd): New static function: just does the job of fetching and validating the file descriptor from the given stream: the first part of fds_subst. This will be done in the parent. (fds_subst): Now just does the substitution using a given file descriptor. Done in the child. Some variable renaming here; a better name for fd_orig is fd_sub, the substitute descriptor. (fds_prepare): New function: called in the parent process, it obtains the three file descriptors from the streams. (fds_swizzle): Just do fd_subst with the ready-made file descriptors, not dealing with the streams. Called in the child. (open_command): Add fds_prepare call that is now needed. (open_subprocess, run): Get rid of catch frame. Just call fds_prepare where fds_swizzle was previously called, and only call fds_swizzle in the child process.
* expander: bug: atoms in quasiliteral.Kaz Kylheku2021-06-152-0/+18
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Via macros, atoms can sneak into a quasiliteral which then blow up because they get treated as strings without being converted. Example: (defmacro two () 2) `@(two)xyz` -> ;; error The expansion produces the invalid form, in which the 2 is subsequently treated as a string. (sys:quasi 2 "xyz") On the other hand, symbol macros don't have this problem: (defsymacro two 2) `@{two}xyz` -> "2xyz" The reason is that the (sys:var two) syntax will expand to (sys:var 2), and not 2. The straightforward, consistent fix is to ensure that the first case will also go to (sys:var 2). * eval.c (expand_quasi): If the expanded form is an atom which is not a bindable symbol, wrap it in a sys:var. * tests/012/quasi.tl: Test cases added. Also adding a compilation test for this file, cribbed from patmatch.tl.
* defsymacro: regression: don't expand replacement.Kaz Kylheku2021-06-153-26/+66
| | | | | | | | | | | | | | | | | | | | | | | | | | This is a regression that was introduced in 191. The change in 191 was trying to prevent defsymacro from being expanded immediately by the expander except in 190 compatibility. Unfortunately, this caused the whole defsymacro block not to be entered unless in 190 compatibility, otherwise taking the common exit which returns form_ex, containing the expanded replacement form. * eval.c (do_expand): Split up implementation of defvarl and defsymacro. In the defsymacro block, do not do any expanding on entry. Absent of compatibility mode, we just do some sanity checks and pass the entire form through. In 262 compatibility, we do the expansion to obtain form_ex. Then all the previous compat logic is wrapped in that block. * tests/011/macros-3.tl: Add a test case which confirms that symbol macros are lazily expanded. Weakness in the test suite is how these regressions creep in. * txr.1: Improve defsymacro documentation, spelling out clearly that the unexpanded replacement form is associated with the symbol. Eliminate obsolescent text suggesting that defsymacro is evaluated at macro time.
* txr.1: Correction to compat note for 190.Kaz Kylheku2021-06-151-1/+14
| | | | | | | | * txr.1: When writing up the compatibility notes for version 190, I missed the change that defmacro and defsymacro were switched away from instant evaluation by the macroexpander. That change introduced a regression which I just noticed, and this is noted also.
* doc: problems in mkstempKaz Kylheku2021-06-151-3/+3
| | | | | * txr.1: mkstemp: fix formatting problem in heading, and incorrect reference to mkdtemp with bad grammar.
* subprocesses: diagnose streams with no fileno.Kaz Kylheku2021-06-151-3/+9
| | | | | | | | * stream.c (fds_subst): Check that stream_fd returns a non-integer; if so, put out a more meaningful diagnostic, rather than allowing c_num to generate a "nil is not an integer" error. Also, let's incorporate the self string into the existing failed dup diagnostic.
* vim: recognize #/.../ inside brace variables.Kaz Kylheku2021-06-151-1/+1
| | | | | | * genvim.txr (bvar): Add tl_regex as a constituent. This occurs in quasiliteral pattern matching syntax, like `...@{a #/regex}...`.
* unwind: unbind dyn env when diagnosing unhandled exception.Kaz Kylheku2021-06-141-0/+2
| | | | | | | | | | * unwind.c (uw_unwind_to_exit_point): The error reporting relies on the *stderr* stream, but that stream may be bound in some funny way in a context where an exception has happened, without anything around it to restore that during unwinding. Before diagnosing, we should reset the dynamic environment to the top level, so that the global values of all dynamic variables are revealed.
generated by cgit v1.2.3 (git 2.25.1) at 2025-03-19 15:52:15 +0000