📄 changelog
字号:
* Makefile.in (dist): Survive ln failures. Create .tar.z (gzipped tar) file as well as .tar.Z (compressed tar) file.Fri Jan 8 22:31:41 1993 Paul Eggert (eggert@twinsun.com) * side.c (print_half_line): When the input position falls outside the column, do not output a tab even if the output position still falls within the column.Mon Dec 21 13:54:36 1992 David J. MacKenzie (djm@kropotkin.gnu.ai.mit.edu) * Makefile.in (.c.o): Add -I.Fri Dec 18 14:08:20 1992 Paul Eggert (eggert@twinsun.com) * configure.in: Add HAVE_FCNTL_H, since system.h uses it.Tue Nov 24 10:06:48 1992 David J. MacKenzie (djm@goldman.gnu.ai.mit.edu) * Makefile.in: Note change from USG to HAVE_STRING_H.Mon Nov 23 18:44:00 1992 Paul Eggert (eggert@twinsun.com) * io.c (find_and_hash_each_line): When running out of lines, double the number of allocated lines, instead of just doubling that number minus the prefix lines. This is more likely to avoid the need for further memory allocation.Wed Nov 18 20:40:28 1992 Paul Eggert (eggert@twinsun.com) * dir.c (dir_sort): Plug memory leak: space holding directory contents was not being reclaimed. Get directory size from struct file_data for initial guess at memory needed. Detect errors when reading and closing directory. (diff_dirs): Pass struct file_data to dir_sort. Finish plugging leak. * diff.c (compare_files): Pass struct file_data to diff_dirs. * io.c (find_and_hash_each_line): Don't assume alloc_lines is nonzero when allocating more lines.Thu Nov 12 16:02:18 1992 Paul Eggert (eggert@twinsun.com) * diff.c (main): Add `-U lines' as an alias for `--unified=lines'. * diff3.c (usage): Add third --label option in example. * util.c (analyze_hunk): Fix test for ignoring blank lines. * configure.in, system.h: Avoid USG; use HAVE_TIME_H etc. instead.Mon Nov 9 05:13:25 1992 Paul Eggert (eggert@twinsun.com) * diff3.c (main, usage): Add -A or --show-all. -m now defaults to -A, not -E. Allow up to three -L options. (output_diff3_edscript, output_diff3_merge): Remove spurious differences between these two functions. Output ||||||| for -A. Distinguish between conflicts and overlaps. (dotlines, undotlines): New functions that output `Ns', not `N,Ns'. (output_diff3_edscript, output_diff3_merge): Use them. * io.c (find_identical_ends): shift_boundaries needs an extra identical line at the end, not at the beginning. * sdiff.c (edit): execvp wants char **, not const char **.Mon Oct 19 04:39:32 1992 Paul Eggert (eggert@twinsun.com) * context.c (print_context_script, find_function): Context line numbers start with - file->prefix_lines, not 0. * io.c (binary_file_p): Undo last change; it was a library bug.Sun Oct 18 00:17:29 1992 Richard Stallman (rms@mole.gnu.ai.mit.edu) * io.c (binary_file_p): Consider empty file as non-binary.Mon Oct 5 05:18:46 1992 Paul Eggert (eggert@twinsun.com) * diff3.c (main, make_3way_diff, using_to_diff3_block): Don't report bogus differences (for one of -mexEX3) just because the file0-file1 diffs don't line up with the file0-file2 diffs. (This is entirely possible since we don't use diff's -n option.) Always compare file1 to file2, so that diff3 sees those changes directly. Typically, file2 is now the common file, not file0. (output_diff3_merge): The input file is file 0, not the common file. (FC, FO): New macros; they replace FILE1, FILE0 for two-way diffs, to distinguish them from three-way diffs. * diff3.c (using_to_diff3_block): Fold repeated code into loops. * diff3.c (make_3way_diff, process_diff): Have the *_end variable point to the next field to be changed, not to the last object allocated; this saves an if-then-else. * diff3.c (process_diff): Use D_NUMLINES instead of its definiens. * diff3.c: Make fns and vars static unless they must be external.Wed Sep 30 09:21:59 1992 Paul Eggert (eggert@twinsun.com) * analyze.c (diff_2_files): OUTPUT_IFDEF is now robust. * diff.h (ROBUST_OUTPUT_STYLE): Likewise. (default_line_format): Remove. All refs removed. * ifdef.c (print_ifdef_lines): Add %L. Optimize %l\n even if user specified it, as opposed to its being the default.Tue Sep 29 19:01:28 1992 Paul Eggert (eggert@twinsun.com) * diff.c (longopts, main): --{old,new,unchanged,changed}--group-format are new options, so that -D is no longer overloaded. Set no_diff_means_no_output if --unchanged-{line,group}-format allows it. * diff.h (enum line_class): New type. (group_format, line_format): Use it to regularize option flags. All refs changed. * ifdef.c (format_ifdef, print_ifdef_lines): %n is no longer a format.Mon Sep 28 04:51:42 1992 Paul Eggert (eggert@twinsun.com) * diff.c (main, usage): Replace --line-prefix with the more general --{old,new,unchanged}-line-format options. * ifdef.c (format_ifdef, print_ifdef_lines): Likewise. * diff.h (line_format): Renamed from line_prefix. All refs changed. * diff.h, ifdef.c (default_line_format): New variable. * util.c (output_1_line): New function. (print_1_line): Use it. * ifdef.c: (format_ifdef, print_ifdef_lines): Add %0 format.Sun Sep 27 05:38:13 1992 Paul Eggert (eggert@twinsun.com) * diff.c (main): Add -E or --line-prefix option. Add -D'=xxx' for common lines. Change default -D< format from copy of -D> format to to -D<%<; similarly for default -D> format. * diff.h (common_format, line_prefix): New variables. * ifdef.c (format_ifdef): New function. (print_ifdef_script, print_ifdef_hunk, print_ifdef_lines): Use it for -D'=xxx', -E. * context.c (find_hunk): Glue together two non-ignorable changes that are exactly CONTEXT * 2 lines apart. This shortens output, removes a behavioral discontinuity at CONTEXT = 0, and is more compatible with traditional diff. * io.c (find_identical_ends): Slurp stdin at most once. * util.c (print_line_line): line_flag is const char *.Thu Sep 24 15:18:07 1992 Paul Eggert (eggert@twinsun.com) * ifdef.c (print_ifdef_lines): New function, which fwrites a sequence of lines all at once for speed. (print_ifdef_script, print_ifdef_hunk): Use it.Thu Sep 24 05:54:14 1992 Paul Eggert (eggert@twinsun.com) * diff.c (main): Support new -D options for if-then-else formats. (specify_format): New function. * diff.h (ifndef_format, ifdef_format, ifnelse_format): New variables. * ifdef.c (print_ifdef_hunk): Use the new variables instead of a hardwired format. * side.c (print_1sdiff_line): Represent incomplete lines on output. (print_sdiff_script): Likewise. Don't print 'q' at end, since that doesn't work with incomplete lines. * sdiff.c (interact): Don't assume diff output ends with 'q' line. * diff.h (ROBUST_OUTPUT_STYLE): OUTPUT_SDIFF is now robust. * sdiff.c (lf_copy, lf_snarf): Use memchr instead of index, to avoid dumping core when files contain null characters. (memchr): New function (if memchr is missing). * io.c (sip): New arg SKIP_TEST to skip test for binary file. (read_files): Don't bother testing second file if first is binary.Thu Sep 17 21:17:49 1992 David J. MacKenzie (djm@nutrimat.gnu.ai.mit.edu) * system.h [!USG && !_POSIX_VERSION]: Protect from conflicting prototype for wait in sys/wait.h.Wed Sep 16 12:32:18 1992 David J. MacKenzie (djm@nutrimat.gnu.ai.mit.edu) * Makefile.in: Include binprefix in -DDIFF_PROGRAM.Tue Sep 15 14:27:25 1992 David J. MacKenzie (djm@nutrimat.gnu.ai.mit.edu) * Version 2.0.Sat Sep 12 01:31:19 1992 David J. MacKenzie (djm@nutrimat.gnu.ai.mit.edu) * util.c, diff.h, system.h [!HAVE_MEMCHR]: Don't use void * and const when declaring memchr replacement. Declare memchr if !STDC_HEADERS && !USG.Thu Sep 10 15:17:32 1992 David J. MacKenzie (djm@nutrimat.gnu.ai.mit.edu) * Makefile.in (uninstall): New target. * diff.c (excluded_filename): Use fnmatch, not wildmat. (usage): Document -x, -X, --exclude, --exclude-from. Makefile.in: Use fnmatch.c, not wildmat.c.Sun Sep 6 23:46:25 1992 Paul Eggert (eggert@twinsun.com) * configure.in: Add HAVE_MEMCHR. * diff.h, util.c: Use it instead of MEMCHR_MISSING.Sun Sep 6 07:25:49 1992 Paul Eggert (eggert@twinsun.com) * diff.h: (struct line_def): Replace this 3-word struct with char *. This uses less memory, particularly for large files with short lines. (struct file_data): New member linbuf_base counts number of lines in common prefix that are not recorded in linbuf; this uses less memory if files are identical or differ only at end. New member buffered_lines counts possibly differing lines. New member valid_lines counts valid data. New member alloc_lines - linbuf_base replaces old linbufsize. linbuf[0] now always points at first differing line. Remove unused members ltran, suffix_lines. Add const where appropriate. (Is_space): New macro, for consistent definition of `white space'. (excluded_filename, memchr, sip, slurp): New declarations. * ed.c (print_ed_hunk): Adjust to diff.h's struct changes. * context.c (pr_context_hunk): Likewise. * ifdef.c (print_ifdef_script): Likewise. * side.c (print_sdiff_script, print_half_line): Likewise. * util.c (analyze_hunk, line_cmp, print_1_line): Likewise. * analyze.c (shift_boundaries): Remove unneeded variable `end' and unnecessary comparisons of `preceding' and `other_preceding' against 0. (diff_2_files): When comparing files byte-by-byte for equality, don't slurp them all in at once; just compare them a buffer at a time. This can win big if they differ early on. Move some code to compare_files to enable this change. Use only one buffer for stdin with `diff - -'. (discard_confusing_lines, diff_2_files): Coalesce malloc/free calls. (build_script): Remove obsolete OUTPUT_RCS code. * diff.c (add_exclude, add_exclude_file, excluded_filename): New fns. (main): Use them for the new --exclude and --exclude-from options. (compare_files): Don't open a file unless it must be read. Treat `diff file file' and `diff file dir' similarly. Move some code here from diff_2_files to enable this. Simplify file vs dir warning. * dir.c (dir_sort): Support new --exclude* options. * io.c (struct equivclass): Put hash code and line length here instead of struct line_def, so that they can be shared. (find_and_hash_each_line): Compute equivalence class as we go, instead of doing it in a separate pass; this thrashes memory less. Make buckets realloc-able, since we can't preallocate them. Record one more line start than there are lines, so that we can compute any line's length by subtracting its start from the next line's, instead of storing the length explicitly. This saves memory. Move prefix-handling code to find_identical_ends; this wins with large prefixes. Use Is_space, not is_space, for consistent treatment of white space. (prepare_text_end): New function. (find_identical_ends): Move slurping here, so it's only done when needed. Work even if the buffers are the same (because of `diff - -'). Compare prefixes a word at a time for speed. (find_equiv_class): Delete; now done by find_and_hash_each_line. (read_files): Don't slurp unless needed. find_equiv_class's work is now folded into find_and_hash_each_line. Don't copy stdin buffer if `diff - -'. Check for running out of primes. (sip, slurp): Split first part of `slurp' into another function `sip'. `sip' sets things up and perhaps reads the first ST_BLKSIZE buffer to see whether the file is binary; `slurp' now just finishes the job. This lets diff_2_files compare binary files lazily. Allocate a one-word sentinel to allow word-at-a-time prefix comparison. Count prefix lines only if needed, only count the first file's prefix. Don't bother to count suffix lines; it's never needed. Set up linbuf[0] to point at first differing line. (binary_file_p): Change test for binary files: if it has a null byte in its first buffer, it's binary. (primes): Add more primes. * util.c (line_cmp): Use bcmp for speed. Use Is_space, not is_space, for consistent treatment of white space. (translate_line_number): Internal line numbers now count from 0 starting after the prefix. (memchr): New function (if memchr is missing). * Makefile.in: Document HAVE_ST_BLKSIZE. Link with wildmat.o. * system.h (STAT_BLOCKSIZE): New macro based on HAVE_ST_BLKSIZE. * configure.in: Add AC_ST_BLKSIZE. * wildmat.c: New file.Fri Sep 4 01:28:51 1992 Richard Stallman (rms@mole.gnu.ai.mit.edu) * sdiff.c (xmalloc): Renamed from ck_malloc. Callers changed.Thu Sep 3 15:28:59 1992 David J. MacKenzie (djm@nutrimat.gnu.ai.mit.edu) * diff.h: Don't declare free, index, rindex.Tue Aug 11 22:18:06 1992 John Gilmore (gnu at cygnus.com) * io.c (binary_file_p): Use heuristic to avoid declaring info files as binary files. Allow about 1.5% non-printing characters (in info's case, ^_).Tue Jul 7 01:09:26 1992 David J. MacKenzie (djm@nutrimat.gnu.ai.mit.edu) * diff.h: Replace function_regexp and ignore_regexp with lists of compiled regexps. * analyze.c, context.c, util.c: Test whether the lists, not the old variables, are empty. * util.c (analyze_hunk), context.c (find_function): Compare lines with the lists of regexps. * diff.c (add_regexp): New function. (main): Use it. * diff3: Add -v --version option. * Makefile.in: Link with version.o. * system.h: New file. * diff.h, cmp.c, diff3.c, sdiff.c: Use it. * diff.h, diff3.c: Include string.h or strings.h, as appropriate. Declare malloc and realloc. * diff3.c (perror_with_exit): Include program name in message. * diff3.c: Lowercase error messages for GNU standards. * sdiff.c [USG || STDC_HEADERS]: Define bcopy in terms of memcpy. * sdiff.c: Use the version number from version.c. * Makefile.in: Link with version.o. * cmp.c error.c xmalloc.c: New files from textutils. * Makefile.in: Add rules for them. * diff.c (longopts): --unidirectional-new-file is like -P, not -N. Rename --file-label to --label (leave old name, but undocumented). * sdiff.c, diff.c (usage): Condense messages and fix some errors. * diff3.c (main, usage): Add long-named options.
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -