error.
(pfatal_with_name): Pass error string as format parameter instead
of as part of format string.
(pfatal_and_delete): Likewise.
(main): Adjust call to fatal.
(xmalloc): Likewise.
<stdlib.h> is available.
(enum sym_type): New st_C_attribute value for parsing
gcc's __attribute__. Deleted st_C_typespec value.
(gperf, in_word_set): Use gperf 3, options changed. Added the
__attribute__ keyword, removed all the st_C_typespec keywords,
changed attribute for Java to (C_JAVA & !C_PLPL).
(inattribute): New global bool, part of the C state machine.
(cblev): Identifier renamed to bracelev throughout.
(consider_token, C_entries): Numerous changes for making the
parser more robust and adding support for __attribute__.
Tiny change by David A. Capello <dacap@users.sourceforge.net> -
(Lua_suffixes, Lua_help, lang_names, Lua_functions):
Support the Lua scripting language <http://www.lua.org>.
Tweak arch tagging to make build/install-in-place less annoying
Previously, autoconf-created Makefiles and the like would contain
duplicate taglines (unfortunately, autoconf doesn't seem to have a
`strip in generated file' comment mechanism) leading to conflicts, and
installing in place would create unknown directories and copies of
source directories (leading to conflicts with the source directories).
This changeset makes all autoconf-processed files use explicit id-tags
and adds .arch-inventory entries to ignore installation directories.
(main): Tidy up usage string. Fix "Use of uninitialized value" warning.
Set version to 0.2. Parse the directory listing to get any ChangeLog.n
file, not just 1..9.
(header_match_p, entry_match_p, print_log, parse_changelog): Remove Perl
prototypes (their purpose is to help the parser, which isn't needed here,
not declare arguments).
(parse_changelog): Make --reverse faster on big batches by not modifying
the entries list.
:local: methods, or omits the colon between the hostname
and the path. Allow :/ in repository path, since CVS does.
Fix typo: "pository" should be set from $CVSROOT, not $repository.
This fixes a bug reported by Wolfgang Scherer in
<http://mail.gnu.org/archive/html/bug-gnu-emacs/2004-02/msg00085.html>,
along with some related bugs I discovered by inspecting how
CVS itself parses $CVSROOT.
returning it. Fix the return type accordingly.
(main): Under --eval, don't fail if left with additional
arguments after decoding options. Quote file names.
lines for readability.
(decode_options): Don't call print_help_and_exit in default case. Print version
information to stdout.
(main): Don't call print_help_and_exit.
Waroquiers <wao@cfmu.eurocontrol.be> applied and largely revised.
(language): Added a `help' member.
(arg_type): Added an at_end constant.
(plain_C_suffixes): Some items removed from here.
(Objc_suffixes): And put here (new constant).
(Ada_help, Asm_help, default_C_help, Cplusplus_help, Cjava_help,
Cobol_help, Erlang_help, Fortran_help, HTML_help, Lisp_help,
Makefile_help, Objc_help, Pascal_help, Perl_help, PHP_help,
PS_help, Prolog_help, Python_help, Scheme_help, TeX_help,
Texinfo_help, Yacc_help, auto_help, none-help, no_lang_help): New
constants.
(PS_functions, PS_suffixes): Renamed from Postscript_functions and
Postscript_suffixes.
(lang_names): Adapted to the new language structure, new language
"objc" added (was previously merged with "proc").
(print_language_names): Some help strings corrected.
(print_help): Now takes an argument and possibly prints lang help.
(print_help): Some help strings corrected. Documents
language-specific help.
(main): Only print help after having parsed all the arguments.
(libintl.h): Include this if _LIBC. Otherwise include gettext.h.
(wchar.h): Include, maybe.
(attribute_hidden): Define if not defind.
(__getopt_initialized): Use attribute_hidden.
(__libc_argc, __libc_argv): Renamed from original_argc, etc.
(__getopt_nonoption_flags, nonoption_flags_max_len, nonoption_flags_len):
Conditional on USE_NONOPTION_FLAGS.
(SWAP_FLAGS): New definitions.
(exchange): Test USE_NONOPTION_FLAGS.
(_getopt_initialize): Test USE_NONOPTION_FLAGS.
(_getopt_internal): Error if argc < 1. New local var print_errors.
Improve test for ambiguous long option.
Add LIBIO support for error message output.
(NONOPTION_P): Test USE_NONOPTION_FLAGS.
(eval, display): New vars.
(longopts): Add --eval and --display.
(decode_options): Add -e and -d processing.
(print_help_and_exit): Update the usage string.
(main): Add support for --eval and --display.
auto-detection is enabled and the `::' qualifier is met.
(consider_token, C_entries): Several bugs corrected that tagged
some declarations even though --declarations was not used.
(plainc): New macro.
(C_entries): Use it.
(C_entries): Several cosmetic changes.
(C_entries): Invalidate the token is some cases.
(Ada_funcs): Do not tag "use type Xxxx;".
New language HTML.
(make_tag): Never generate null length tag names.
(linebuffer_init): Renamed from initbuffer. All callers changed.
(pattern): Structure renamed to `regexp', member regex renamed to
pattern.
(node_st): Member pat renamed to regex.
(pattern); New member force_explicit_name, for future use. Now
always set to true, cannot be reset.
(add_regex, regex_tag_multiline, readline): Use it.
(main): Free some global structures.
(fdesc): New member `written'.
(readline, process_file): Initialise it.
(put_entries): Set it.
(main): Use it to create entries for files without tags.
(total_size_of_entries): Do not count invalid tags.
(etags_strcasecmp): Like BSD's, for compatibility.
(strcaseeq): Make it into a macro.
which are almost always unnamed. etags.el looks for an explicit
tag name, then for an implicit one. See make_tag for details.
The change is both forwards and backwards compatible.
(make_tag): New function (was the disabled function new_pfnote).
(make_C_tag): Use it.
(arg_type): at_icregexp label removed (obsolete).
(pattern): New member multi_line for multi-line regexps.
(filebuf): A global buffer containing the whole file as a string
for multi-line regexp matching.
(need_filebuf): Global flag raised if multi-line regexps used.
(print_help): Document new regexp modifiers, remove references to
obsolete option --ignore-case-regexp.
(main): Do not set regexp syntax and translation table here.
(main): Treat -c option as a backward compatibility hack.
(main, find_entries): Init and free filebuf.
(find_entries): Call regex_tag_multiline after the regular parser.
(scan_separators): Check for untermintaed regexp and return NULL.
(analyse_regex, add_regex): Remove the ignore_case argument, which
is now a modifier to the regexp. All callers changed.
(add_regex): Manage the regexp modifiers.
(regex_tag_multiline): New function. Reads from filebuf.
(readline_internal): If necessary, copy the whole file into filebuf.
(readline): Skip multi-line regexps, leave them to regex_tag_multiline.
(add_regex): Better check for null regexps.
(readline): Check for regex matching null string.
(find_entries): Reorganisation.
sequences supported by gcc.
(find_entries): rewind unconditionally.
(find_entries): Do not call language functions directly, now calls
itself.
(find_entries): Do general initialisations here.
(CNL_SAVE_DEFINEDEF, C_entries, LOOP_ON_INPUT_LINES, F_getit)
(Ada_getit, Pascal_functions, Pascal_functions)
(prolog_skip_comment): Do not do them here.
(readline_internal): Increment lineno here.
(readline): Conditionally undo readline_internal increment.
(readline): Do not return a value.
(STDIN): New constant.
(parsing_stdin): New flag.
(longopts): New option --parse-stdin=NAME.
(print_help): Document it.
(main): Handle it.
(process_file): Split into process_file and process_file_name.
(process_file_name): New function.
(TEX_defenv): Now contains more contructs.
(TEX_cmt): Make it a static char and move it before TeX_commands.
(TeX_commands): Shorten the tag to the brace after the name.
(TeX_commands): Names now include the initial backslash.
(TeX_commands): Names do not include numeric args #n.
(TeX_commands): Correct line char number in tags.
(TEX_tabent, TEX_token): Deleted.
(TeX_commands, TEX_decode_env): Streamlined.
inadvertently lost on 2002-04-13.
* etags.c (main): Use `sort -o TAGFILE TAGFILE' instead of
`sort TAGFILE -o TAGFILE', as POSIX 1003.1-2001 disallows
the latter usage.
inadvertently lost on 2002-04-13.
(main): Use `sort -o TAGFILE TAGFILE' instead of `sort TAGFILE -o
TAGFILE', as POSIX 1003.1-2001 disallows the latter usage.
(main): New argument -d, for specifying directory.
(usage): Document.
(get_user_id): Compute
(get_home_dir): Deleted.
(get_prefix): New function, taken from main.
(main): Check whether or not we are running setuid. Move prefix
computation to get_prefix. Don't call getpwent; we don't need to
any more. Instead, move it to get_user_id().
(process_file): Print the tags from the nodes as soon as
possible, and delete the nodes. This brings down the memory
occupancy as etags to almost the same level as when the #line
directives were not parsed.
(free_fdesc): New function.
(find_entries): Use it.
(invalidate_nodes): In etags mode, do not just mark the nodes as
invalid, do delete them.
(strcaseeq): New function.
(get_language_from_filename): Use it to do a case insenstitive
comparison if called with appropriate args.
(find_entries): Try with case insensitive match.
(process_file): Bug fixed.
the command line, in either order.
* etags.c (find_entries): Delete tags previously obtained from
file xxx.c's #line directives when parsing file xxx.y. This is
generally done for automatically generated files containing
#line directives. This handles the case when xxx.y is tagged
before xxx.c, and the entries of xxx.c pointing to xxx.y should
be discarded.
(language): Added the metasource member. Initializers changed.
(invalidate_nodes): New function.
(readline): Discard lines after having found a #line
directive pointing to an already tagged file. This handles the
case when xxx.y is tagged before xxx.c, and the entries of
xxx.c pointing to xxx.y should be discarded.
(fdesc): New structure for keeping track of input files.
(fdesc): Remove `file' member (a string) and use instead a pointer
to a file description structure.
(curfile, curfiledir, curtagfname, curlang, nocharno,
forced_lang): Global variables removed in favor of fdhead and
curfdp, pointers to file description strucures.
(longopts, main, print_help): Use the CTAGS conditional to include
or exclude options that work on etags or ctags only.
(process_file, find_entries, pfnote, add_node, put_entries,
readline): Use fdhead and curfdp.
(process_file, find_entries): Do not take an arg string, all
callers changed.
* etags.c (longopts, print_help, main): Test CTAGS to disallow
options that are not right for either etags or ctags.
* etags.c (number_len, total_size_of_entries): Define them also
in CTAGS mode, because gcc does not compile all refs away.
(MAX_DATA_LEN, MAX_SCORES): New.
(SCORE_FILE_PREFIX): If HAVE_SHARED_GAME_DIR is not defined,
default to ~/.emacs.d/games.
(get_user_id): Don't zero uid in the case where we can't get the
username.
(lose): New function.
(main): Actually use `max', and default it to MAX_SCORES.
Correctly handle new default for SCORE_FILE_PREFIX. Use `lose'
function.
(read_score): Handle the case of reading unamelen characters, then
finishing. Use mktemp if mkstemp isn't available.
(lock_file, unlock_file): Delete unused versions.
(lock_file): Always sleep, even if we unlinked the lock file.
(get_user_id): Take struct passwd as an argument.
(get_home_dir): New function.
(main): Read in user information here. Discover home directory if
necessary.
(read_score): Trim newline only in `getline' case.
(struct score_entry): Add username field.
(push_score): Use it.
(get_user_id): New function.
(main): Don't malloc excessively.
(main): Use username field.
(read_score): Read it.
(push_score): Handle it.
(write_scores) Write it.
(read_score): Handle arbitrary length data.
(Python_functions, PHP_functions): Name tags, for ctags' sake.
(TeX_commands): Name tags. Correction of old disabled code.
* etags.c (curfiledir, curtagfname): New global variables.
(process_file): Initialise them.
(readline): Canonicalize the name found in #line directive.
(no_line_directive): New global var; set it for old behaviour.
(main): Remove some #ifdef in the getopt switch.
(add_node, put_entries): Code added to merge different chunks of
nodes referring to the same file. Currently the tags are just
appended, without any check for duplicates.
(Perl_functions): Do not special case ctags.
(readline): Identify #line directives and do the right thing.
(nocharno, invalidcharno): New global vars.
(process_file): Reset nocharno.
(readline): Set nocharno.
(pfnote): Read nocharno and maybe put invalidcharno in node.
(total_size_of_entries, put_entries): Use invalidcharno.
* etags.c: Keep the whole tag table in memory, even in etags mode.
(main): Call put_entries here even in CTAGS mode.
(main, process_file): Check the return values of fclose and pclose.
(process_file): Do not call put_entries after parsing each file.
(process_file): Canonicalise file names even for ctags.
(process_file): Set curfile here...
(find_entries): ... not here any more.
(add_node): In etags mode, build a linked list of entries (on
right pointer) for each file, and link the first entry of each
file on left nodes.
(put_entries): Print here the name of the file.
(put_entries): Print the entries starting from the first file.
(number_len, total_size_of_entries): Define these only iin etags
mode, make the second work only on the right nodes.
* etags.c: Make all global variables static.
