--- - branch: MAIN date: Sun May 19 03:30:44 UTC 2013 files: - new: '1.31' old: '1.30' path: pkgsrc/devel/rcs/Makefile pathrev: pkgsrc/devel/rcs/Makefile@1.31 type: modified - new: '1.3' old: '1.2' path: pkgsrc/devel/rcs/PLIST pathrev: pkgsrc/devel/rcs/PLIST@1.3 type: modified - new: '1.11' old: '1.10' path: pkgsrc/devel/rcs/distinfo pathrev: pkgsrc/devel/rcs/distinfo@1.11 type: modified - new: '0' old: '1.4' path: pkgsrc/devel/rcs/patches/patch-ab pathrev: pkgsrc/devel/rcs/patches/patch-ab@0 type: deleted - new: '0' old: '1.2' path: pkgsrc/devel/rcs/patches/patch-ac pathrev: pkgsrc/devel/rcs/patches/patch-ac@0 type: deleted - new: '0' old: '1.1' path: pkgsrc/devel/rcs/patches/patch-ad pathrev: pkgsrc/devel/rcs/patches/patch-ad@0 type: deleted id: 20130519T033044Z.14ae975fae5a7ffa513bf35323a2f1ce9e45b321 log: "Update to 3.9.0\n\n* Use gmake.\n* MAKE_JOBS_SAFE=yes (at least, with MAKE_JOBS=5)\n\nChangelog:\nNEWS for GNU RCS (Revision Control System)\nSee the end for copying conditions.\n\n- 5.9.0 | 2013-05-06\n\n - distribution now .tar.lz and .tar.xz\n\n\tIf you have GNU tar, you can use \"tar xf\" and it will DTRT.\n\tIf not, you can use \"lzip -dc TARBALL | tar xf -\" to unpack\n\tthe .tar.lz, or \"xz -dc TARBALL | tar xf -\" for the .tar.xz.\n\n - planned retirement\n\n - configure option â\x80\x98--enable-compat2â\x80\x99\n\n\tThis option enables reading of files written by RCS 2.x (before\n\tRCS was GNU, even), but the file format became obsolete in 1982.\n\tSupport for it WILL BE REMOVED in a near future GNU RCS release.\n\n - common option â\x80\x98-Vâ\x80\x99\n\n\tThis option is obsoleted by â\x80\x98--versionâ\x80\x99 (since 5.8, 2011-08-30).\n\tSupport for it WILL BE REMOVED in some future GNU RCS release.\n\tIts use now produces a warning to stderr.\n\n\tPlease note that â\x80\x98-VNâ\x80\x99 (N â\x88\x88 {3,4,5}) is a separate issue.\n\n - bugs fixed\n\n - â\x80\x98rcsmerge --helpâ\x80\x99 mentions â\x80\x98-Aâ\x80\x99, â\x80\x98-Eâ\x80\x99, â\x80\x98-eâ\x80\x99\n\n\tThese options are accepted and internally passed to diff3(1).\n\n - â\x80\x98ident -VNâ\x80\x99 and â\x80\x98merge -VNâ\x80\x99 now signal error\n\n\tFor these commands, the argument to â\x80\x98-Vâ\x80\x99 has no meaning.\n\tPreviously, such invocations would display version info,\n\tignoring the arg. Now they signal a \"bad option\" error.\n\n - new features\n\n - ident(1) recognizes Subversion \"fixed-width keyword syntax\"\n\n\tIn addition to the normal keyword pattern, for Subversion 1.2\n\t(and later) compatability, ident(1) also recognizes patterns\n\thaving one of the forms:\n\n\t$KEYWORD:: TEXT $\n\t;; two colons and space after keyword\n\t;; space before ending $\n\n\t$KEYWORD:: TEXT#$\n\t;; two colons and space after keyword\n\t;; hash before ending $\n\n - new co(1) option â\x80\x98-Sâ\x80\x99 for \"self-same\" mode\n\n\tIn this mode, the owner of a lock is unimportant, just that it\n\texists. Effectively, this prevents you from checking out the\n\tsame revision twice.\n\n\t$ whoami\n\t ttn\n\n\t$ co -l -f z\n\t RCS/z,v --> z\n\t \ revision 1.1 (locked)\n\t done\n\n\t$ co -S -l -f z\n\t RCS/z,v --> z\n\t \ co: RCS/z,v: Revision 1.1 is already locked by ttn.\n\n - several RCS commands \"internalized\" into rcs(1)\n\n\tAs part of an ongoing effort to modernize the command-line\n\tinterface of GNU RCS, the previously standalone programs:\n\n\t \ ci, co, rcs, rcsclean, rcsdiff, rcsmerge, rlog\n\n\tcan now be invoked as \"rcs PROGRAM\". In this case, we call\n\tPROGRAM a \"command\", and also provide some aliases. E.g., you\n\tcan type \"rcs diff\" in addition to \"rcs rcsdiff\" (and \"rcsdiff\",\n\tof course). We plan to support standalone (w/o \"rcs\" prefix)\n\tinvocation from the shell at least through the 5.x series of\n\treleases -- not indefinitely, but for a good while, yet.\n\n\tThe rcs(1) program itself now supports â\x80\x98--commandsâ\x80\x99 to list all\n\tthe commands, â\x80\x98--aliasesâ\x80\x99 to list their aliases, as well as\n\tâ\x80\x98--help COMAMNDâ\x80\x99 to show the help for COMMAND.\n\n\tNote that programs ident(1) and merge(1) remain standalone.\n\n\tLastly, in the manual, section \"Invoking rcs\" now describes both\n\t\"modern\" and \"legacy\" usages. \ Also, the internalized commands\n\tinvocations mentions both \"rcs COMMAND\" and \"PROGRAM\" styles.\n\n - â\x80\x98--helpâ\x80\x99 output includes a one-line description\n\n\tE.g., \"merge --help\" says: \"Three-way file merge\".\n\n - most long options can be recognized if partially specified\n\n\tWith the exception of \"rcs --help COMMAND\", where \"--help\" must\n\tbe spelled out in full, all long options can now be recognized\n\teven if partially specified. For example, \"rcs --al\" is\n\trecognized as \"rcs --aliases\".\n\n - updated portability\n\n\tSeveral more Gnulib (http://www.gnu.org/software/gnulib) modules\n\tare now in use.\n\n \ - new cross-compilation support\n\n\tThe configure script now assigns \"pessimistic defaults\" when\n\tcross-compiling. See new section \"cross-compilation\" in README.\n\n - effects of â\x80\x98-VNâ\x80\x99 (N â\x88\x88 {3,4,5}) documented\n\n\tSee (info \"(rcs) Misc common options\").\n\n - maintenance tools updated\n - gnulib-tool (GNU gnulib 2013-05-01 06:14:19) 0.0.7913-5191\n\n- 5.8.2 | 2013-04-04\n\n \ - Bugs fixed\n\n - Wrong symbolic name dereference\n\n\tRCS 5.8 introduced a bug whereby commands would incorrectly\n\tdereference a symbolic name (into a numerical revision number)\n\tin the presence of multiple symbolic names that share a common\n\tprefix. See tests/t803.\n\n - â\x80\x98integrityâ\x80\x99 value syntax better specified\n\n\tThe â\x80\x98integrityâ\x80\x99 value (if present) is a string composed of a\n\tsystem part followed by an optional user part, with formfeed\n\t(^L, U+0C) separating the two. Unlike other string values, the\n\tâ\x80\x98integrityâ\x80\x99 string (either part) must NOT contain '@' (U+40).\n\tIf it does, RCS displays a \"spurious '@' in `integrity' value\"\n\terror message and exits failurefully.\n\n\tThis change restores interop play-space for third-party tools\n\tthat was stricken w/ the RCS 5.8 top-level grammar freeze. Left\n\tunspecified is how such tools should divvy up the user part.\n\n - New manual chapter: RCS file format\n\n\tThis documents the RCS file format grammar and particulars,\n\tadapted from rcsfile(5).\n\n \ - Manpages refer to info documentation\n\n\tThey now recommend using \"info rcs\" for full documentation.\n\n - Script to trim \"junk at end of file\" posted\n\n\tThe \"junk at end of file\" error occurs for some files due to RCS\n\t5.8 being more strict about the syntax it accepts. You can use\n\tâ\x80\x98trimrcsâ\x80\x99 by Warren Jones:\n\n\thttp://lists.gnu.org/archive/html/help-rcs/2013-01/msg00006.html\n\n\tto remedy such files while we figure out how best to move such\n\tfunctionality into RCS proper.\n\n - Maintenance tools updated\n - automake (GNU Automake) 1.13.1\n \ - gnulib-tool (GNU gnulib 2013-03-19 16:08:47) 0.0.7899-34f84\n - makeinfo (GNU Texinfo) 5.1\n\n- 5.8.1 | 2012-06-05\n\n - Bugs fixed\n\n - Debug output removed\n\n\tDue to an oversight, release 5.8 rlog included code to write\n\tdebugging output to stderr for invocations using the â\x80\x98-dâ\x80\x99 option\n\tand a date range (e.g., â\x80\x98rlog -d 2010<2012â\x80\x99).\n\n - Criteria for avoiding read-only checks refined\n\n\tFor \"make check\", some test cases are inhibited if the user\n\trunning the test is not effectively blocked from writing a\n\tpurportedly read-only file, such as when \"make check\" is run by\n\tthe super user, or for certain (weird) NFS situations.\n\n\tPreviously, to determine this condition, we considered only the\n\toperating system type, a very crude (and incomplete) proxy.\n\tNow, we explicitly check with a shell sequence comprising the\n\tumask(1) and test(1) commands, plus output-redirection.\n\n - Regression in â\x80\x98-zLTâ\x80\x99 handling\n\n\tOn a 64-bit x86 system, RCS 5.8 introduced a regression whereby:\n\n\t rlog -zLT -d>2011-05-04\n\n\twould select the correct entry but display its date always with\n\tdefault (01) month and day, i.e., YYYY-01-01. \ This is now fixed\n\t(see also tests/t320).\n\n - Regression in â\x80\x98ci -d -Tâ\x80\x99 handling\n\n\tRCS 5.8 introduced a regression whereby:\n\n\t ci -l -d -T FILE\n\n\twould set the mtime of RCS/FILE,v (the comma-v file) to the\n\tepoch. \ This is now fixed (see also tests/t810).\n\n - Use â\x80\x98diff --labelâ\x80\x99 instead of â\x80\x98diff -Lâ\x80\x99\n\n\tPreviously, RCS used GNU diff's â\x80\x98-Lâ\x80\x99 option. According to Paul\n\tEggert (a GNU diffutils maintainer):\n\n\t That option has been undocumented since diffutils 2.8\n\t (released in March 2002) and the option is intended to\n\t be replaced sometime soon with a different meaning.\n\n\tNow, RCS uses â\x80\x98diff --labelâ\x80\x99, thus immune to the planned change.\n\n - Miscellaneous changes\n - Make help extraction noisy (on failure)\n - Silence some compiler warnings\n - Increase coverage of \"make check\"\n\n - Documentation improvements\n\n - Manpage rcsintro(1) dropped\n\n\tThis manpage is redundant, and (arguably) should not have been\n\tin section 1 in the first place.\n\n - Use \"Invoking COMMAND\" instead of \"COMMAND\" as node names\n\n\tThis makes it easier for â\x80\x98info --usage COMMANDâ\x80\x99 to DTRT, and\n\tmakes GNU (info \"(standards) Manual Structure Details\") happy.\n\n \ - explicitly UTF-8\n\n\tThis is to prepare for a (future) GNU Texinfo release that\n\trenders @code in a more pretty way when the encoding is UTF-8.\n\tIf you're reading this (from the future) with such a Texinfo at\n\thand, feel free to regenerate the docs in doc/ prior to install.\n\n - CVS is not GNU\n\n\tPreviously, we incorrectly said \"GNU CVS\", succumbing to a\n\tcommon misunderstanding. Now we know better.\n\n - TAGS file no longer distributed\n\n\tTo create, configure normally and do \"make TAGS\".\n\n - New configure script option â\x80\x98--enable-coverageâ\x80\x99\n\n\tSpecifying â\x80\x98--enable-coverageâ\x80\x99 causes â\x80\x98_Exitâ\x80\x99 to be an alias for\n\tâ\x80\x98exitâ\x80\x99 and CFLAGS to append â\x80\x98--coverageâ\x80\x99 if the compiler is GCC.\n\tThis is needed because the coverage machinery writes the .gcda\n\tfiles only on â\x80\x98exitâ\x80\x99.\n\n\tThis option is for maintainers; most people can ignore it.\n\n - Portability improvements\n - Use more gnulib modules\n - Use portable Makefile subst-ref variable syntax\n - Use portable shell command-output interpolation syntax\n - Maintenance tools upgraded\n - GNU Automake 1.12\n - GNU Autoconf 2.69\n - gnulib-tool (GNU gnulib 2012-06-03 16:29:00) 0.0.7432-f6c24-modified\n\n- 5.8 | 2011-08-30\n\n - License now GPLv3+ (see COPYING)\n\n - Change in terminology: from \"path\" to \"file name\" (or \"file-name\")\n\n\tHowever, if \"path\" intends \"search path\", we say so explicitly.\n\n \ - Changes to the RCS package\n\n - New documentation in Info format\n\n\tOn \"make install\", rcs.info is installed in $(infodir), with\n\ttitle \"GNU RCS \" in dircategory \"Version control\".\n\tThe doc source is texinfo (released under GNU FDL 1.3), so you\n\tcan easily create output in HTML, PDF, etc.\n\n \ - Dropped configure option: --with-diffutils\n\n\tTo specify non-GNU diffutils programs diff(1) and diff3(1), name\n\tthem using variables on the configure command-line. \ See README.\n\n - Configuration more strict in some ways, more lax in others.\n\n\tBefore, part of the configuration was done at compile time.\n\tNow, all of it is done by the configure script. Here are the\n\tset of conditions which will cause the configure script to fail\n\t(with a \"could not find...\" message):\n\n\t - for --enable-mailer=PROG, â\x80\x98PROGâ\x80\x99 not absolute;\n\t - for diff(1), value of env var â\x80\x98DIFFâ\x80\x99 not absolute;\n\t - diff(1) not GNU diffutils compatible;\n\t - for diff3(1), value of env var â\x80\x98DIFF3â\x80\x99 not absolute;\n\t - for ed(1), value of env var â\x80\x98EDâ\x80\x99 not absolute;\n\t - no C99-capable compiler.\n\n\tHere, \"absolute\" means \"specified as an absolute filename\".\n\tOn the flip side, configuration no longer checks for some\n\tsituations such as â\x80\x98sigactionâ\x80\x99 yes, but â\x80\x98SA_SIGINFOâ\x80\x99 no.\n\n\tMost of the portability duties are now handled by gnulib.\n\n - New configure option: --enable-suid[=setreuid]\n\n\tThis builds RCS with setuid support (the default). Optional\n\targ â\x80\x98setreuidâ\x80\x99 means use setreuid(2) instead of seteuid(2).\n\n - New configure option: --disable-mmap\n\n\tThis builds RCS without mmap(2), even if available. See README.\n\n - New configure option: --enable-mailer=PROG\n\n\tThe feature whereby ci(1) sends mail when breaking a lock is now\n\tdisabled by default. To enable, specify â\x80\x98--enable-mailer=PROGâ\x80\x99\n\tto the configure script. See README.\n\n - New configure option: --enable-compat2\n\n\tThis preponderantly unlikely to be used option allows RCS\n\tcommands to read RCS files written by RCS 2. See README.\n\n - You can \"make check\" prior to \"make install\".\n\n\tDoing \"make check\" automatically prepends to the â\x80\x98PATHâ\x80\x99 env var\n\tthe value of â\x80\x98$(abs_top_builddir)/srcâ\x80\x99, so that the programs\n\tco, rcsdiff, and rcsmerge can find their peers (co and merge).\n\n\tLikewise, \"make installcheck\" prepends â\x80\x98$(DESTDIR)$(bindir)â\x80\x99.\n\n\tPreviously, you had to \"make install\" first and then arrange\n\tfor â\x80\x98$(DESTDIR)$(bindir)â\x80\x99 to be on â\x80\x98PATHâ\x80\x99 \"manually\".\n\n\tSee tests/README for more info on the test suite.\n\n - Bug fixes\n\n - Remove all edit info when removing all revisions.\n\n\tPreviously, \"rcs -o\" (outdating) all revisions failed to\n\tleave the RCS file in a consistent state; edit info (i.e.,\n\tlog message + diff(1) output) remained for the deleted revisions.\n\n\tFor example, this sequence of commands:\n\t echo foo > foo\n\t ci -q -i -t-desc -mHELLO foo\n\t rcs -q -o1.1: foo\n\t grep '@H' foo,v\n\tused to display \"@HELLO\" to stdout.\n\n\tNow, all revisions are completely removed.\n\n - Code no longer uses mktemp(3).\n\n\tUsing mktemp(3) is a security risk. We use mkstemp(3) now.\n\tLikewise, rcsfreeze.sh now uses mktemp(1).\n\n - Misc manpage tweaks / fixes.\n\n\tDocument â\x80\x98rlog -qâ\x80\x99; fix small merge.1 omission; add branch\n\tlabels in rcsfile.5; say \"GNU RCS \" in footer.\n\n - Other changes\n\n - All commands accept â\x80\x98--helpâ\x80\x99 and â\x80\x98--versionâ\x80\x99.\n\n\tThe help output includes an email address for bug reports.\n\tFor continuity, option â\x80\x98-Vâ\x80\x99 is now a synonym for â\x80\x98--versionâ\x80\x99.\n\n\tRelatedly, commands no longer display usage info if given\n\ta bad or malformed option. You can use â\x80\x98--helpâ\x80\x99 for that.\n\n - A string of all digits is now valid for author, state.\n\n\tThis means you can set the author or state to, for example,\n\t\"000000\" or \"42\". Previously, these would have caused a\n\t\"invalid identifier\" or \"invalid symbol\" error.\n\n - Env var RCS_MEM_LIMIT controls stdio threshold.\n\n\tFor speed, RCS uses memory-based routines for files up to\n\t256 kilobytes, and stream-based (stdio) routines otherwise.\n\tYou can change this threshold value by setting the environment\n\tvariable â\x80\x98RCS_MEM_LIMITâ\x80\x99 to a non-negative integer, measured in\n\tkilobytes. An empty â\x80\x98RCS_MEM_LIMITâ\x80\x99 value is silently ignored.\n\n - RCS can now work with files larger than 2 gigabytes.\n\n\tRCS now uses large file offsets (#define _FILE_OFFSET_BITS 64).\n\n - Pass-through for RCS file â\x80\x98commitid SYMBOLâ\x80\x99 now builtin.\n\n\tDue to GNU CVS (Concurrent Versions System) using a compatible\n\tfile format as RCS for the RCS file, you can use RCS commands\n\tto view and manipulate its contents. (Note, however, the next\n\tNEWS item.)\n\n\tThis works well enough except for a small annoyance: CVS adds a\n\tper-revision field called the \"commitid\" with an opaque (to RCS)\n\tsymbolic value. Previously, RCS commands would emit a warning\n\t\"unrecognized phrases\" (unless given â\x80\x98-qâ\x80\x99 on the command-line).\n\n\tRCS commands now automatically support pass-through handling of\n\tâ\x80\x98commitid SYMBOLâ\x80\x99 (so â\x80\x98-qâ\x80\x99 is no longer necessary).\n\n - RCS file top-level grammar frozen.\n\n\tThe RCS file top-level grammar is now frozen; RCS no longer\n\tsupports pass-through operation of unrecognized key/data pairs\n\t(called \"newphrases\" in the RCS 5.7 rcsfile(5) manpage).\n\n\tTo avoid painting ourselves into a corner, the grammar now\n\tincludes a new key â\x80\x98integrityâ\x80\x99 with @-string value, whose\n\tsub-grammar is not yet specified. (We intend to keep checksums\n\tand other compacted redundancies in this field, for manipulation\n\tby the commands in a future RCS 5.x release.) For upward\n\tcompatability, the commands in this release do not change this\n\tfield, although they silently read and write it (pass-through),\n\tif present.\n\n - RCS file syntax-validated earlier, completely.\n\n\tPreviously, RCS file syntax was validated lazily, and trailing\n\tgarbage was not detected (see bugfix above). Now, a top-level\n\tvalidation is done on each access.\n\n \ - Possible to specify an empty log message with ci -m, rcs -m.\n\n\tThe commands \"ci -m\" and \"rcs -m\" no longer error on an\n\tempty log message. Their non-interactive behavior is now\n\tconsistent with the interactive invocation.\n\n\twas: ci -m file < /dev/null # use stdin to avoid error\n\tnow: ci -m file # works fine, like so\n\n\tNote that these commands actually store as the log message\n\tthe string: \"*** empty log message ***\".\n\n - Date option accepts some more date-only formats\n\n\tDate format â\x80\x98YYYY-DDDâ\x80\x99 specifies a year and a day (1-366),\n\twhile format â\x80\x98YYYY-wWW-Dâ\x80\x99 specifies a year, an ISO week number\n\t(0-53, 0 is a GNU RCS extension), and a day number (1-7, for\n\tMonday through Sunday).\n\n - Changes to rcsdiff\n\n - New handling for option: -U N\n\n\tThis arranges to output N lines of unified-diff context.\n\tRelatedly, the list of possible options passed to the underlying\n\tdiff(1) appears in both \"rcsdiff --help\" and in the manual.\n\n - Refined \"same-revision don't call diff\" optimization\n\n\tNormally, if the two revisions specified are the same, we avoid\n\tcalling the underlying diff(1) on the theory that it will\n\tproduce no output. This does not hold generally for â\x80\x98-yâ\x80\x99\n\t(--side-by-side) and â\x80\x98-Dâ\x80\x99 (--ifdef), such as when the revision\n\tspecified is by different symbolic names, so for those options\n\tthe optimization is disabled.\n" module: pkgsrc subject: 'CVS commit: pkgsrc/devel/rcs' unixtime: '1368934244' user: ryoon