- Pacman-dev - lists.archlinux.org

[pacman-dev] [PATCH 1/2] bash-completion: don't complete filenames when they're not wanted
by Eli Schwartz 21 Oct '18

21 Oct '18

Filename completion should only be generated for makepkg, when using the options -p or --config... which means we should offer option completions by default. Filename completion for pacman, should not be generated when using -Qu, or -F without -o. Signed-off-by: Eli Schwartz <eschwartz(a)archlinux.org> --- scripts/completion/bash_completion.in | 18 ++++++++++++------ 1 file changed, 12 insertions(+), 6 deletions(-) diff --git a/scripts/completion/bash_completion.in b/scripts/completion/bash_completion.in index d99fba53..31b41f71 100644 --- a/scripts/completion/bash_completion.in +++ b/scripts/completion/bash_completion.in @@ -71,10 +71,13 @@ _pacman_key() { } _makepkg() { + compopt +o default local cur opts prev COMPREPLY=() _get_comp_words_by_ref cur prev - if [[ $cur = -* && ! $prev =~ ^-(-(config|help|key|version)$|\w*[Vhp]) ]]; then + if [[ $prev = @(-p|--config) ]]; then + compopt -o default + elif [[ ! $prev =~ ^-(-(config|help|key|version)$|\w*[Vh]) ]]; then opts=('allsource asdeps check clean cleanbuild config force geninteg help holdver ignorearch install key log needed noarchive nobuild nocheck nocolor noconfirm nodeps noextract noprepare noprogressbar nosign @@ -101,6 +104,7 @@ _pacman_repo_list() { } _pacman() { + compopt -o default local common core cur database files prev query remove sync upgrade o COMPREPLY=() _get_comp_words_by_ref cur prev @@ -131,12 +135,14 @@ _pacman() { D|R) _pacman_pkg Qq;; F) - _arch_incomp 'l list' && _pacman_pkg Slq - ;; + { _arch_incomp 'l list' && _pacman_pkg Slq ; } || + _arch_incomp 'o owns' || + compopt +o default;; Q) { _arch_incomp 'g groups' && _pacman_pkg Qg sort; } || { _arch_incomp 'p file' && _pacman_file; } || - _arch_incomp 'o owns' || _arch_incomp 'u upgrades' || + { _arch_incomp 'u upgrades' && compopt +o default; } || + _arch_incomp 'o owns' || _pacman_pkg Qq;; S) { _arch_incomp 'g groups' && _pacman_pkg Sg; } || @@ -153,8 +159,8 @@ _pacman_file() { compopt -o filenames; _filedir 'pkg.tar*' } -complete -F _pacman -o default pacman -complete -F _makepkg -o default makepkg +complete -F _pacman pacman +complete -F _makepkg makepkg complete -F _pacman_key -o default pacman-key # ex:et ts=2 sw=2 ft=sh -- 2.18.0

2 2

[pacman-dev] [PATCH] handle EINTR while polling scripts/hooks
by Andrew Gregory 21 Oct '18

21 Oct '18

If poll() is interrupted by a signal, alpm was closing the socket it uses for listening to script/hook output. This would drop script output at the least and kill the script at the worst. Fixes FS#60396 Signed-off-by: Andrew Gregory <andrew.gregory.8(a)gmail.com> --- lib/libalpm/util.c | 10 +++++++++- test/pacman/tests/TESTS | 1 + test/pacman/tests/scriptlet-signal-handling.py | 15 +++++++++++++++ 3 files changed, 25 insertions(+), 1 deletion(-) create mode 100644 test/pacman/tests/scriptlet-signal-handling.py diff --git a/lib/libalpm/util.c b/lib/libalpm/util.c index eaf85e93..d33eef2a 100644 --- a/lib/libalpm/util.c +++ b/lib/libalpm/util.c @@ -665,6 +665,7 @@ int _alpm_run_chroot(alpm_handle_t *handle, const char *cmd, char *const argv[], ssize_t olen = 0, ilen = 0; nfds_t nfds = 2; struct pollfd fds[2], *child2parent = &(fds[0]), *parent2child = &(fds[1]); + int poll_ret; child2parent->fd = child2parent_pipefd[TAIL]; child2parent->events = POLLIN; @@ -685,7 +686,14 @@ int _alpm_run_chroot(alpm_handle_t *handle, const char *cmd, char *const argv[], #define STOP_POLLING(p) do { close(p->fd); p->fd = -1; } while(0) while((child2parent->fd != -1 || parent2child->fd != -1) - && poll(fds, nfds, -1) > 0) { + && (poll_ret = poll(fds, nfds, -1)) != 0) { + if(poll_ret == -1) { + if(errno == EINTR) { + continue; + } else { + break; + } + } if(child2parent->revents & POLLIN) { if(_alpm_chroot_read_from_child(handle, child2parent->fd, ibuf, &ilen, sizeof(ibuf)) != 0) { diff --git a/test/pacman/tests/TESTS b/test/pacman/tests/TESTS index 5deb93c4..dc0b4ec3 100644 --- a/test/pacman/tests/TESTS +++ b/test/pacman/tests/TESTS @@ -150,6 +150,7 @@ TESTS += test/pacman/tests/replace102.py TESTS += test/pacman/tests/replace103.py TESTS += test/pacman/tests/replace104.py TESTS += test/pacman/tests/replace110.py +TESTS += test/pacman/tests/scriptlet-signal-handling.py TESTS += test/pacman/tests/scriptlet-signal-reset.py TESTS += test/pacman/tests/scriptlet001.py TESTS += test/pacman/tests/scriptlet002.py diff --git a/test/pacman/tests/scriptlet-signal-handling.py b/test/pacman/tests/scriptlet-signal-handling.py new file mode 100644 index 00000000..e65df55b --- /dev/null +++ b/test/pacman/tests/scriptlet-signal-handling.py @@ -0,0 +1,15 @@ +self.description = "Handle signal interrupts while running scriptlets/hooks" + +p1 = pmpkg("dummy") +p1.install['post_install'] = """ + kill -INT $PPID # send an arbitrary signal that pacman catches + sleep 1 # give alpm time to close the socket if EINTR was not handled + echo to-parent # if the interrupt is not handled this will die with SIGPIPE + echo success > interrupt_was_handled + """ +self.addpkg(p1) + +self.args = "-U %s" % p1.filename() + +self.addrule("PACMAN_RETCODE=0") +self.addrule("FILE_EXIST=interrupt_was_handled") -- 2.19.0

2 1

[pacman-dev] [PATCH] reset signal handlers before running scripts/hooks
by Andrew Gregory 21 Oct '18

21 Oct '18

Front-ends or libraries may set signals to be ignored, which gets inherited across fork and exec. This can cause scripts to malfunction if they expect the signal. To make matters worse, scripts written in bash can't reset signals that were ignored when bash was started. Fixes FS#56756 Signed-off-by: Andrew Gregory <andrew.gregory.8(a)gmail.com> --- Hopefully nobody ignores or depends on signals outside of the POSIX list. As far as I can tell, the only way to reset all signal handlers is to iterate over the entire range of positive integers. lib/libalpm/util.c | 20 ++++++++++++++++++++ test/pacman/tests/TESTS | 1 + test/pacman/tests/scriptlet-signal-reset.py | 11 +++++++++++ 3 files changed, 32 insertions(+) create mode 100644 test/pacman/tests/scriptlet-signal-reset.py diff --git a/lib/libalpm/util.c b/lib/libalpm/util.c index a06f5bfd..eaf85e93 100644 --- a/lib/libalpm/util.c +++ b/lib/libalpm/util.c @@ -548,6 +548,25 @@ static int _alpm_chroot_read_from_child(alpm_handle_t *handle, int fd, return 0; } +static void _alpm_reset_signals(void) +{ + /* reset POSIX defined signals (see signal.h) */ + /* there are likely more but there is no easy way + * to get the full list of valid signals */ + int *i, signals[] = { + SIGABRT, SIGALRM, SIGBUS, SIGCHLD, SIGCONT, SIGFPE, SIGHUP, SIGILL, + SIGINT, SIGKILL, SIGPIPE, SIGQUIT, SIGSEGV, SIGSTOP, SIGTERM, SIGTSTP, + SIGTTIN, SIGTTOU, SIGUSR1, SIGUSR2, SIGPOLL, SIGPROF, SIGSYS, SIGTRAP, + SIGURG, SIGVTALRM, SIGXCPU, SIGXFSZ, + 0 + }; + struct sigaction def; + def.sa_handler = SIG_DFL; + for(i = signals; *i; i++) { + sigaction(*i, &def, NULL); + } +} + /** Execute a command with arguments in a chroot. * @param handle the context handle * @param cmd command to execute @@ -633,6 +652,7 @@ int _alpm_run_chroot(alpm_handle_t *handle, const char *cmd, char *const argv[], exit(1); } umask(0022); + _alpm_reset_signals(); execv(cmd, argv); /* execv only returns if there was an error */ fprintf(stderr, _("call to execv failed (%s)\n"), strerror(errno)); diff --git a/test/pacman/tests/TESTS b/test/pacman/tests/TESTS index b11cb511..5deb93c4 100644 --- a/test/pacman/tests/TESTS +++ b/test/pacman/tests/TESTS @@ -150,6 +150,7 @@ TESTS += test/pacman/tests/replace102.py TESTS += test/pacman/tests/replace103.py TESTS += test/pacman/tests/replace104.py TESTS += test/pacman/tests/replace110.py +TESTS += test/pacman/tests/scriptlet-signal-reset.py TESTS += test/pacman/tests/scriptlet001.py TESTS += test/pacman/tests/scriptlet002.py TESTS += test/pacman/tests/sign001.py diff --git a/test/pacman/tests/scriptlet-signal-reset.py b/test/pacman/tests/scriptlet-signal-reset.py new file mode 100644 index 00000000..27246d12 --- /dev/null +++ b/test/pacman/tests/scriptlet-signal-reset.py @@ -0,0 +1,11 @@ +self.description = "Reset signals before running scriptlets/hooks" + +p1 = pmpkg("dummy") +# check if SIGPIPE is ignored, it should be fatal, but GPGME ignores it +p1.install['post_install'] = "kill -PIPE $$; echo fail > sigpipe_was_ignored" +self.addpkg(p1) + +self.args = "-U %s" % p1.filename() + +self.addrule("PACMAN_RETCODE=0") +self.addrule("!FILE_EXIST=sigpipe_was_ignored") -- 2.19.0

2 1

[pacman-dev] [GIT] The official pacman repository annotated tag, v3.0.0-rc2, updated. v3.0.0-rc2
by Andrew Gregory 20 Oct '18

20 Oct '18

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "The official pacman repository". The annotated tag, v3.0.0-rc2 has been updated to fc4cd3864aae4edd5835116d9191b48f60ffa4a4 (tag) from b977b9c030861decd22376e4d2a9147e830ebfae (which is now obsolete) tagging 9e3a1853451230d887c3c89f49f04e502fb69621 (commit) replaces v3.0.0-rc1 tagged by Andrew Gregory on Sat Oct 20 12:53:22 2018 -0700 - Log ----------------------------------------------------------------- replaces broken tag b977b9c030861decd22376e4d2a9147e830ebfae -----BEGIN PGP SIGNATURE----- iQEzBAABCAAdFiEEuBUbEXA3eBCVUUynu9/8kjBrESEFAlvLh7IACgkQu9/8kjBr ESFJ9QgAgulpF3KQDRNRgGIpoIlblE8U0H6AziEr+lG3RRUGZs4gxJybD2aMkcrB KSQ0L4dTHfcQ1d//d+9N2TnEf60erz+ryC+lBDzNenqZ4ocLQwiYeRCzyc8DqIoU 7Rl8mFgv0IB+4IOh5r9nPFzsGqV1UJkTjOP5guHVsy7U/1ZbXEEjhBUwkhWPfpke 3QV+7T2I2oq8aFMm6Xx30WuOIUeFpjQ+HzyI3evVWQW/B3sL5CsTm/iyav3MaAZs rtCrGkNpfH1EI2+N8WW/sA6ZON+LpnEoUnovaYD2wU2KRlnZDc5sZt3jXp9ETBvp poKY0EgIJulQ5US59AdIWg2zWKDfSw== =oJsw -----END PGP SIGNATURE----- Aaron Griffin (7): * Fixed an issue with globbing the --test argument Added this test to check the XferCommand functionality * Fix the double package name URL when using XferCommand * Bug fix for makepkg dependency testing. This requires that we * Updated -V output to include the 2007 copyright date. * Two fixes when running under a new root (-r|--root) * Fixed an error with 'cascade' removal due to creation of a new pmpkg_t struct Dan McGee (33): * rankmirrors updates from Scott Horowitz <stonecrest(a)gmail.com>. * Removed a mirrorlist that codemac says is quite outdated. * Failure to #include config.h cost us here, we lost all NLS in alpm.c. This commit looks much more monumental than it is. Almost all just #include * Added missing header include guards in md5.h and sha1.h. Trying to fix up this autotools stuff a bit more. * Remove sha1 checksums for now from the INTEGRITY_CHECK array. * Updated Italian translation * Fix FS #6534- unclear IgnorePkg message. Sorry translators, had to update * Sorry tranlators, another string update. .pacorig was displayed twice in * Slight updates of Hungarian translation * -Qs was returning an error if no package found, which is not the same behavior as -Ss. * Oops, that last commit had some debug stuff in it. Removed it and added * Updated Italian translation * Added a readme file to the contrib/ directory. * Updated Brazilian Portuguese translation * Updated German translation * Updated pot files for hopefully the last time before release. * Slight updates to NEWS file. * Mark a function as static that is only used in deptest.c. * Updated Italian translation * Added confirmation step to makepkg -C operation to ensure user is deleting * Another slight update, getting rid of -rf flags and clarifying a message. * Updated Italian translation * Added Russian language translation. Thanks! * Fix group comparison issue and associated compilation warnings by using * Updated Hungarian translation, thanks! * Updates to Hungarian translation from Nagy, mostly cleanup stuff. * Fixed -Qil regression, now both flags are honored. (FS #1355) * Removed ${CFLAGS} from Makefile.am(s) as it was causing all CFLAGS to be * Fix an issue where the same dependency was recorded multiple times in the * Fix wrong filesize being recorded to local DB. Reported by Andreas Radke. * Noted a misspelling for later (after we are not in string freeze). Jürgen Hötzel (1): fixed string comparison callback (patch from Nagy Gabor) ----------------------------------------------------------------------- hooks/post-receive -- The official pacman repository

1 0

[pacman-dev] [GIT] The official pacman repository annotated tag, v3.0.0-rc1, updated. v3.0.0-rc1
by Andrew Gregory 20 Oct '18

20 Oct '18

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "The official pacman repository". The annotated tag, v3.0.0-rc1 has been updated to bdb94daf7256d315c39ce73b78ffcf2c8b5a77cf (tag) from 59ae306a5453e0113ffba8395db9df1b5507c650 (which is now obsolete) tagging 11fbc595d48d55be929e259285bb65e53ef33007 (commit) tagged by Andrew Gregory on Sat Oct 20 12:47:37 2018 -0700 - Log ----------------------------------------------------------------- replaces broken tag 59ae306a5453e0113ffba8395db9df1b5507c650 -----BEGIN PGP SIGNATURE----- iQEzBAABCAAdFiEEuBUbEXA3eBCVUUynu9/8kjBrESEFAlvLhlkACgkQu9/8kjBr ESFf1AgAn3mykXPagGnkXeunD9xjXmzo4Fnx7OLbOCBzBL0nZATLUjAyu0j26HO7 PaScKOuSOc5m9WHYdmbqQy8fIsu1mI9jyNC0q1iTCpX54DCp1RZSmMIHlO7wgHCJ kMHVHfmnt5Zz7Y9bmYm0MSbLPklbUzcSeOh1DzgIERsJD2aCJ3Dwstfpg9uKTGty jM2n3ZhB5WSw4Sz6cKt0eH+y64YKkwvQVTNoMS7kgQBXrEHITh4gtmDeEz4FcYAT OEJc7xXFQFYbblcpAMeYKRfyjN54F5fkXHj7jWijBSa4LBOlwTl6UGDZxKXnjXR1 Z6ZBWoROrhOs3ASDlubfFIGPzC3Bsw== =z+VF -----END PGP SIGNATURE----- Aaron Griffin (313): Merging Frugalware changes - these need to be checked for instances of "Frugal" Added pactest to repository, from Aurelien Foret: Frugalware changes - mainly architecture updates and signed/unsigned changes Merged frugalware changes (too many to list). Also added some config file Merged frugalware changes. Added a few other minor things too, but there's alot Final frugalware changes commit include changes so that this compiles Fixed java includes - java bindings now compile Corrected documentation compilation - succeeds now Added alpm function docs along *** empty log message *** Whoops forgot some .in files *** empty log message *** Added po files... not sure if this is the right approach Added sha1 support (for now?) Another forgotten file set Removed convertdb Whoops, actually remove convertdb this time Added po files Applied changes from frugalware: Frugalware patch for AC_CHECK_PROGS fixups: *** empty log message *** Applied Frugalware patch from Christian Hamar alias krix <krics(a)linuxforum.hu> configure changes for java detection (require swig) Copyright changes Adjust progress bar to align with frugalware's progressbar Removed extra includes Removed three checks for po4a (patching issue) Yet another "added three times" issue. Fixed doxygen comments Reverted debug parameter to -1 Added copyright holder Removed old db files (unused) From VMiklos <vmiklos(a)frugalware.org> A handful of minor changes: Whoops, this file got left out Minor changes: More extern moving - keep extern decls in the headers makes for easier/better Fixed this test so that it succeeds - it is probably not 100% appropriate, but Fixes from frugalware: few 'typos' included while patching Reverted the "out of memory" error to 1 to prevent API changes. Added re-pacman Added PM_DLFNM_LEN define, via VMiklos Attempted fix for x86_64 - switched some unsigned char variables to ints, and Fixed library directory for bindings - we need the craptastic .libs libtool dir Moved downloaded db unpacking to the backend files, to easier allow conversion Remove generated man pages from CVS Fri Oct 27 21:54:32 CEST 2006 VMiklos <vmiklos(a)frugalware.org> Added libfetch linux port + misc changes Numerous changes: Numerous changes: * Makefile changes for libfetch integration Forgot to cycle to next server on download error - fixed, as well as the Removed user-CFLAGS during a debug build. Added -Wall to normal build(I like -Wall) Fixed pacman -U: Whoops forgot this Added a newline upong progress completion - this was accidentally left out causing the second progress bar to overwrite the first * Fixed some alpm_get_option calls (long params were used for C99 compliance, * Modified some error output and logging Removed antesis mirror, as it no longer works Added mcheck support for memory debugging * Fixed an error message that should be a debug message * Improved mcheck output mcheck() seems to cause segfaults. Annoying. Switched back to useing mtrace() - if anything valgrind is superior to mcheck anyway Whoops - I fail at setenv Last mtrace/setenv change, I swear Skip root check on -Sp *** empty log message *** * Numerous mini valgrind fixes. * autotool fixes * has_archname changes * has_archname additions * Initial changes to gensync - makepkg changes were not checked in from another * Changes to some of the TODOs and a brief glance at the NEWS file... * integrity check looping - support multiple algorithms at once * makepkg.conf.in variable changes (missed the checkin) Minor makepkg changes - mainly from Daniel McGee * repo-add script - to add entries to a db file directly from package data (no PKGBUILD) libalpm appends the trailing / to the config root, no need to do it here * Fixed a double free which happened in 2 cases in --sync * Fixed an error when trying to download to an unwritable location Removed global maxcols - it is looked up on the fly now, so the progress bar is * Completed getinfo api changes (pmmissing_t, pmtrans_t, etc) * usize -> isize conversion Added Dan McGee to the Authors file for his contributions on the mailing list * Whoops util Makefile *** empty log message *** -std=c99 seems to undefine PATH_MAX in this case. I don't care enough to fix it * Move sync *after* transaction init (lock file). We don't want to stomp on the * Fixed some error output for "-Sd" and "-Rsc" dealing with missing/ignored * cosmetic: double space in "full system upgrade" * Cosmetic changes and typo fixes * Fixed the bug where -R pkgA pkgB failed when pkgA depended on pkgB Added dbpath validation Corrected -Rsc operation - dependancy info was not being read from the DB Changed "Total Uncompressed Size" to "Total Installed Size" Added 'abs' and configs to CVS Corrected -Qg description Check status of final tar/compress package creation for final status (FS#3382) Changed upgrade message to use the term "installation" to make the meaning more Removed an extra call to alpm_option_set_root() - it is set by alpm_init() A few minor updates so --root works again. Merged some proposed changes from "Dan McGee" <dpmcgee(a)gmail.com> rankmirrors script from Dan McGee <dpmcgee(a)gmail.com> Copy/pasta errors from mail client. Fixed. * Dan McGee's makepkg updates http://www.archlinux.org/pipermail/pacman-dev/2006-December/000792.html FS #3289 fix from Dan McGee <dpmcgee(a)gmail.com> Makefile adjustments for installation rules: * Added etc/abs dir to configure generated makefiles More abs build fixes *** empty log message *** * modeline fixes / reformatting * Jürgen Hötzel <juergen(a)hoetzel.info> * Dan McGee <dpmcgee(a)gmail.com> Added default SRCDST setting to pacman's source cache Restored NoUpgrade options - see http://www.archlinux.org/pipermail/pacman-dev/2006-December/000838.html for details *untested* potential fix for the NoUpgrade issues. Errors with last merge (manual patching) * Dan McGee <dpmcgee(a)gmail.com> Jürgen Hötzel <juergen(a)hoetzel.info> Fixes from Dan McGee <dpmcgee(a)gmail.com> and myself * FS#5990 fix (trailing slash) * Unused variable 'path' * Properly fix the NoUpgrade issues. Modified front end output routines to accept a "padding' setting, which pads any Cleanup from Dan McGee <dpmcgee(a)gmail.com> and JÃŒrgen HÃ¶tzel <juergen(a)hoetzel.info> *** empty log message *** Configuration changes: Removed archlinux.de mirror by request Column fix when stdout is redirected. Default to 80. * Removed autoconf TODO fixed in last commit * Removed latest fix from TODO.autoconf * Fix a missing slash separator in printed uri * James Rosten <seinfeld90(a)gmail.com> * Missing error label in sync.c * Added instructions for patch submission * James Rosten <seinfeld90(a)gmail.com> *** empty log message *** * Bug fix FS#6058 * bug fix FS#6100 - fix --print-uris output * cosmetic: newline after "local database is up to date" * James Rosten <seinfeld90(a)gmail.com> * Dan McGee <dpmcgee(a)gmail.com> * Dan McGee <dpmcgee(a)gmail.com> * Jürgen Hötzel <juergen(a)hoetzel.info> *** empty log message *** Added Jamie (yankees26) cvsignore cleanup and additions James Rosten <seinfeld90(a)gmail.com> JÃŒrgen HÃ¶tzel <juergen(a)hoetzel.info> James Rosten <seinfeld90(a)gmail.com> Roman Kyrylych <Roman.Kyrylych(a)gmail.com> Dan McGee <dpmcgee(a)gmail.com> Dan McGee <dpmcgee(a)gmail.com> * Misc logging changes Added strerror() output on unlink failure Dan McGee <dpmcgee(a)gmail.com> Dan McGee <dpmcgee(a)gmail.com> Some confusion with variable names in unlink_file caused unlink to try and find the _relative_ file instead of absolute. The absolute path name has also been used in some of the output there too. * remove static neednl - no accessor functions (yet) *** empty log message *** Blah blah commit message for testing cvslog -> email Preliminary checkin for alpm_list conversion Added some things I noticed when making alpm_list changes, but didn't want to change too much at once. Johannes Weiner <hannes(a)saeurebad.de> Johannes Weiner <hannes(a)saeurebad.de> Whoops. There's a 0 where there should be a 1. Moved some decls to make the "pkg_invalid" patch compile again. makepkg file-type detection changes from James Rosten <seinfeld90(a)gmail.com> Removed list.h * Dan McGee <dpmcgee(a)gmail.com> Introduced an infinite loop during the pmlist -> alpm_list transition, whoops Changelog support for makepkg - this has been in pacman itself for some time, * Added a check for .FILELIST for package validity in addition to .PKGINFO Rename Changelog -> ChangeLog (the standard way of casing a ChangeLog) Lowercase "changelog" in output message * Added some calloc calls to replace the malloc-then-set-to-zero functionality James Rosten <seinfeld90(a)gmail.com> James Rosten <seinfeld90(a)gmail.com> Removed some debugging cruft left in in the last commit This mainly deals with code clarity- removing currently unneeded * Shuffled some of the alpm_list free funtions - still not perfect, but better Cleanup 'neednl' usage - make it static James Rosten <seinfeld90 at gmail.com> * Fix FS #3107- Use ISO date style * copy and paste error, "resolving deps" -> "cleaning up" Forgot this patch from Dan McGee as part of the bug #5409 fixes Dan McGee <dpmcgee(a)gmail.com> Missed that 8) This is a silly script now that a fancy aurbuild is out in the wild (on the AUR, K. Piche <kpiche(a)rogers.com> K. Piche <kevin.piche(a)cgi.com> Turn debug off by default TODO changes Well that silly memory clobber was plain old "uninitialized data" - fixed More uninitialized stuff... I believe these were removed at some point. Should Debug logging changes: Fix corrupt DB entry handling when loading package entries. TODO changes 8) Switched an int to an enum * Make -Qo work on multiple files again * Forgot the int->enum header checkin Removed the extra newline based on general opinion * !strcmp reads funny (as "not string compare" - switch those to == 0, which has Move filelist creation ABOVE the other metafiles to keep them out of the list install scriptlet is _always_ .INSTALL -Ql needs the 'root' prepended to each file Cleaned up some of the metadata control It appears I misunderstood what this bash-ism did Loading package data from a file was ALWAYS generating the filelist, instead of * Minor output fix (move colon so it's not part of the URL) * unified the progress bars (fill_progress function) * Added archive verification when loading package metadata for -u and -A Implemented a crappy version of -Qu (query upgrades). This simply outputs the Added the --upgrades option for -Qu (--query --upgrades) Attempt to unify the -S and -Su version comparissons for when a package is newer Copy-pasta errors. 8) Swapped out PKGBUILD.8 for PKGBUILD.5 from Dan McGee - original PKGBUILD.8 had too many Frugal specific changes. Minor fixups for the moment, still partially complete Attempt #2 at -S/-Su unification: verify that the 'force' option has been read before comparing versions Minor experimental changes to the download progress bar: I broke the progress bars earlier. Put that back to normal. * Fix total transfered size on download bar Cleanup and more debug info for alpm_pkg_compare_versions Only honor 'force' if versions are NOT identical Copy/pasta error in alpm_get_upgrades that caused a segfault *** empty log message *** *** empty log message *** Attempt to NOT remove packages on filesystem errors (like a read-only * Fixed some missing error values in strerror * Cleaned up direct pm_fprintf usage (move to MSG/ERR macros for now) * Remove "RETRIEVE_LOCAL" trans event as libdownload handles local files * Fix #2982 - use -Rs to remove deps * I made "alpm_versioncmp" public a while back, without noticing this was already done (alpm_pkg_vercmp). I dropped this change, making versioncmp "private" again. (alpm_versioncmp -> _alpm_versioncmp, hidden symbol) Cleaned up some more output * 'add' percentages now cap at 100% (see the comment) * Package file parsing - fixed size and isize - isize is the "size" variable *** empty log message *** This should be an ERROR, not a WARNING Added 'testpkg' binary to test for validity of a package: * Updated the README file Updated gensync to use repo-add better error message Rewrote updatesync in terms of repo-add and repo-remove Wasn't clearing out the 'pkgname' variable, which was causing goofy output * Removed the 'vercmp' op from pacman. The standalone 'vercmp' binary should be * Added --cachedir commandline option Exit status of 2 seems wrong for -V Added changes from Daniel YC Lin via bug #3266 * Added long options (missing in patch) Fix some x86_64 configuration handling Changed "MARCHFLAG" to "ARCHSWITCH" as the name was confusing (it's not a FLAG These were in the wrong order, so errored for me. Reverted the exit 2 -> exit 0 change. This was there for a reason (so that Removed desc_localized for the time being. We don't have localized * Refactored conflict checking within packages. Profiling from Dan showed an From Phil Dillon-Thiselton <dibblethewrecker(a)gmail.com>: TODO update While trying to change the number of missing deps output by makepkg, I noticed * Fixed _alpm_pkg_compare_versions so that it does not output warnings if versions match. (reported by Travis Willard) * Bugfix for FS#6427: Allow -Si to use "repository/package" syntax * Bugfix FS#6422 - spacing for warning output. Due to the fact that we fixed Added a 'contrib' directory for other stuff that might be a good idea to maintain here. * A whole mess of backup changes Added a config file for the vim-project plugin for kicks. To use: * Fix version comparisons when one pkgrel doesn't exist (this ony fails on an == Added new italian translation from Giovanni Scafora <linuxmania(a)gmail.com> Moved the update_depends function to trans.c, as it depends on a transaction * Some cascade removal changes. Mainly code cleanup, but this is an attempt to * Looks like I left some debugging code in there * Indentation changes - 4 spaces for python, we standardized this before (and * Use the libtool'd .la file as a link target so we can run pactest properly * Re-added a compare function for syncpkg's - it was removed without thinking TODO updates * Using positional params (POSIX compliant, not C standard) for printfs that * Cleaned up some debug output * Forgot the second printf in the last commit * debug cleanup (more) * Prepend all lists with _ for consistency * Modified the handling and output w.r.t. IgnorePkg entries. -Sy always ignores Just some changes to my TODO We don't need this anymore * Removed a stray print statement left in there * Fixed globbing for pactest --test argument *** empty log message *** Ignore that... wasn't supposed to happen * conflicts are returned with the full path, so prepending root is redundant * Too much debugging output in the alpm_pkg_compare_versions function * The info level for this was wrong, causing replacements to not really work I had this laying around, figured someone might find it useful Whoops, added to Makefile.am Syntax error else if --> elif (thanks to yankees26 for finding this) * corrected (IMO) --debug usage with pactest. --debug is now passed straight * Writing 'replaces' info to both /desc and /depends files for testing - see the * Enforce const char* params when using strings * pacman hidden arguments: removed -Y and -D. -T is the only hidden arg now, to * Added a small note to HACKING * This test was off. pactest doesn't do all the DB work, so we need to manually * Fixed a pacman warning when these fake packages didn't create an (empty) * Just a note for future reference *** empty log message *** * REQUIREDBY test was reversed (pkg3 requiredby pkg1, not pkg1 requiredby pkg3) * Switched some functions to alpm_pkg_get_* usage as I came across them * Comment clarification Big commit this time: Minor changes Added a test for checking 'requiredby' removal * libtool fixes for building against the right libs * handle changes: callbacks should check handle first, as there's a few * Fixed a whole mess of extra '/' pathing issues when a different root is * Fixed the handle realroot stuff * By Roman's request - proactive backup=() entry test case Aurelien Foret (470): Fixed a typo in add_loadtarget() Renamed PM_TRANS_CB_XXX to PM_TRANS_EVT_XXX Improved error handling in add_loadtarget() Renamed PACXXX defines to PM_XXX Added a db_update() function to manage sync databases updates. Removed a hardcoded reference to the 'local' database in db_create() Added LOG_STR_LEN define to avoid hardcoded length for log strings Removed a leftover debug printf Fixed a typo Added a missing argument in db_create() Removed the "__" prefix from __pm_logcb and __pm_logmask Make sure we generate a UPGRADE_DONE event during upgrades One more ORE comment Added DB_TREENAME_LEN define to avoid hardcoded length for database string New ORE tag *** empty log message *** Moved log callback definition to alpm.h - added PM_EXT_PKG and PM_EXT_DB defines to alpm.h (taken from pacman 2.9.3) - reworked sync_synctree() to make use of alpm_db_update() Avoided usage of global variables pmo_root and pmo_dbpath. Added missing extern declaration Bring back db_scan() modifications from pacman 2.9.1 Code formatting Dropped pmo_nosave parameter (now included in pmo_flags) Code reformatting Added a missing FtpQuit() call Code cleanup Renamed PM_RET_ERR to RET_ERR (no need for a PM_ prefix for internal Removed hardcoded reference to fields of packages One more sanity check, and one more ORE tag Removed unuseful getuid() calls Fixed some typo... Avoided extern references to pmo_root. Relocated log and transaction stuffs from pacman.c to log.c and trans.c Copyright update Removed an uneeded global reference Revert to previous version Removed an hardcoded string length Removed an hardcoded string length pacman_deptest() rework Rework for fixed length strings Added support for .lastupdate files (from pacman 2.9.1) - added HTTP support for .lastupdate files removed a deprecated placeholder added support for the XFERCOMMAND option - added pmo_chomp support (from pacman 2.9.3) replaced rewinddir() calls by db_rewind() fetch_pkgurl: do not download a file if it's already in the current dir added display of backup info in dump_pkg_full() Let the C preprocessor concatenates CACHEDIR string - added db_setlastupdate to db.c moved db_find_conflicts from db.c to conflict.c removed leftover debug printf - removed db_find_conflicts() declaration from db.h uncommented a call to alpm_logaction reorganization removed duplicated prototype cleanup for add_loadtarget() logs fix: one can't add a package twice in transaction targets list added a missing goto! replaced hardcoded references to the cache by CACHEDIR backported fixes for trim() from pacman 2.9.5 Backport from pacman 2.9.5 Backport from pacman 2.9.5 (code cleanup) fixed a typo updated comments _alpm_runscriplet fixes Backport from pacman 2.9.5 More backport from pacman 2.9.5 (mainly for resolvedeps) Backport from pacman 2.9.5 (--ignore in usage) reworked PM_LOG_XXX messages during install/remove/upgrade operations backport from pacman 2.9.5 (splitdep returns an int) backport from pacman 2.9.5 (formatting, commentaries) rewrote list_free to not be recursive anymore - fixed a small memory leak made 'provides' actually work when updating 'requiredby' fields from deps renamed temp files from pacman_XXXXXX to alpm_XXXXXX fixed an error message added a FREELISTPTR macro to free a PMList without freeing its data improved log message for alpm_db_update improved logs (use _alpm_log instead of fprintf) replaced occ of list_free() by FREELIST/FREELISTPTR macros reworked dbpath handling to make it possible to use databases stored at different locations at the same time maked use of db->path instead of handle->root+handle->dbpath+db->treename Replaced snprintf calls by the SNPRINTF macro to avoid buffer overflows when copying strings clearer string length for a strncpy call Turned pmdepmissing_t into an opaque structure turned pmdepmissing_t into an opaque structure changed alpm_db_register() prototype put back 5 _alpm_log() calls put back 3 _alpm_log() calls changed alpm_db_register() prototype *** empty log message *** backport from pacman 2.9.5 - chdir fixed a log mask removed an uneeded call to db_scan checked for db_update() archive existence started the work on "pacman_sync" fixed a compilation warning reworked sync_cleanup added RESOLVEDEPS to the list of transaction events reworked resolvedeps: sync with changes in alpm.h and new RESOLVEDEPS event added support for PACKAGES field in alpm_trans_getinfo - updated transaction events fixed a log level - added 'lastupdate' field to pmdb_t more work to backport pacman_sync from pacman 2.9.5 added list_is_ptrin() reworked a log - added transaction flags for the removal of a package to be upgrade fixed a typo added logs for cache generation replaced transaction 'packages' field by 2 queues: one for packages to be installed and the other for the ones to be removed updated structure names (PM_SYNCPKG), and added transaction queues (install & remove) package 'reason' field should default to 0 reworked MSG and ERR outputs reverted back to trans->packages instead of install and remove queues more work to backport the sync features from pacman 2.9.5 synchronization with libalpm for latest sync feature work fixed a small memory leak removed a leftover comment - small bugfixes fixed a potential segfault more work in sync_prepare reworked alpm_db_readpkg added function pkg_splitname (not used for now) added option -p in pacman usage screen backported support for print-uris from pacman 2.9.x updated a log - updated sync feature replaced a call to list_free by the macro FREELIST -fixed 2 logs added packages integrity checking more work for sync_commit() reworked transaction prototypes reworked handle structure initialization - reworked transaction prototypes - reworked pmsyncpkg_t structure - general cleanup cleanup to keep in sync with alpm changes fixed a user message more sanity checks for pkg_load() renamed PM_ERR_INVALID_NAME to PM_ERR_PKG_INVALID_NAME improved logs for package removal add_loadtarget: load package info only when needed fixed a log message more work for sync_commit() fixed a segfault bypass trans_prepare() if there is no target avoided a call to db_scan() - fixed args order in list_is_ptrin() fixed args order in list_is_ptrin() - fixed args order in list_is_ptrin() avoided two calls to db_scan() in checkdeps() display the list of targets to be removed before trans_commit() call code beautifying first steps to support REPLACES feature renamed list_is_ptrin() in list_is_in() some remork for pacman_deptest() code cleanup rework the cache handling to avoid as much as possible calls to db_scan() fixed some potential resource leaks outputs cleanup added transaction events for sync_commit operations removed deprecated comments in add_commit() logs cleanup code cleanup use the cache instead of calling db_scan() - fixed a cut'n'paste bug with %REASON% entry fixed the format of %REASON% entry when writing it code cleanup slighlty improved efficiency for dump_pkg_full() fixed a commented MLK in db_find_conflicts()/CHECK1 some more efficiency in remove_prepare() cascade loop got ride of an ORE tag in resolvedeps() added support for the package reason field avoided to load twice the same target in remove_loadtarget() added a new ORE tag fixed potential memory leaks code cleanup removed an uneeded check reimplemented list_free() a safer way - rework to support pacman_deptest() reworked pacman_deptest() *** empty log message *** removed unused function list_is_ptrin() removed the extra line feed appended by asctime() fixed targets display for cascade or recurse commands added a --debug option to externally set the library log level updated REASON constants to be consistent with databases from pacman 2.9.x code cleanup code cleanup ../../README get ride of ORE tags do not register an already registered db code cleanup fixed a small memory leak with md5sums handling moved pkg_cmp() to cache.c reimplemented list_remove to deal with lists of any kind of data fixed a small memory leak during the cleanup of config data added some debug logs found a possible MLK: to be fixed make use of the new list_remove implementation code cleanup - fixed a small memory leak fixed a memory leak and avoided to modify a read-only parameter code cleanup event transaction callback rework to prepare the introduction of a conversation callback code cleanup code cleanup fixed some small memory leaks reordered operations in add_commit (as in pacman 2.9.x) fixed a typo fixed a pointer error code cleanup fixed a memory leak with md5 sums handling - indented properly sync_cleancache() fixed cb_trans() prototype changed license package field to a list changed license package field to a list Fixed some typos Small update Added a PKG_FULLNAME_LEN macro Applied reworked patch from VMiklos (vmiklos(a)frugalware.org) Fixed an error log Added a CYGWIN define to include header files required to build in a Cygwin environment Added a missing include for CYGWIN target Applied patch from Christian Hamar (krics(a)linuxforum.hu) Moved the REPLACES and FORCE package fields from DESC to DEPENDS db entry Reworked the cache handling to only load DEPENDS info by default Code update to make use of PKG_FULLNAME_LEN Fixed an text formatting issue update Added an error string for lock file creation failure Fixed missing parameter in trans_init() call Fixed EVENT and QUESTION macros - more sanity checks pacman should work without its configuration file! improved query_fileowner() function removed a leftover comment pacman update to cope with the CACHEDIR option from the library added PM_OPT_CACHEDIR library option added a missing header file declaration removed useless getenv() calls added alpm.h to delivery targets (Jason) removed an uneeded variable declaration do not free the cache right before using it! added a missing line feed in a MSG call added pmconfig_t structure to hold all the configuration removed entry dealing with pacman configuration structure stripped end of lines renamed pmconfig_t to config_t added PACCONF, PACDB and PACROOT defines fixed some compilation warnings fixed a compilation warning fixed parseconfig usage Used the chroot syscall instead of the chroot binary removed the declaration of an unused variable added more pm_errnos fixed compilation warnings added a FREESYNC macro to release pointers to pmsyncpkg_t structures fixed wrong prototype usage for trans_new() cleanup fixed compilation warnings fixed a file descriptor leak Fixed some memory leaks (patch from Essien Ita Essien <essiene(a)datavibe.net>) Fixed some logging issues (patch from VMiklos <vmiklos(a)frugalware.org>) Fixed a typo (spelling) fix (patch from VMiklos <vmiklos(a)frugalware.org>) Allowed C++ usage (patch from VMiklos <vmiklos(a)frugalware.org>) added a debug log when freeing the package cache removed an uneeded call to strtrim used defines from alpm.h instead of local ones (patch from VMiklos <vmiklos(a)frugalware.org>) do not consider duplicated targets as an error (patch from VMiklos <vmiklos(a)frugalware.org>) don't complain on PM_ERR_DB_UPTODATE errors (patch from VMiklos <vmiklos(a)frugalware.org>) added a missing code chunk of the ignore command line option (patch from VMiklos <vmiklos(a)frugalware.org>) fixed a segfault and added more verbose outputs in group handling (patch from VMiklos <vmiklos(a)frugalware.org>) *** empty log message *** renamed error titles from XXX to xxx - fixed a regression with pacman 2.x (patch from VMiklos <vmiklos(a)frugalware.org>) removed some overlapping pm_errno assignments - reworked verbose levels fixed a possible memory corruption fixed a possible memory leak Added Cygwin target defines code cleanup fixed a typo fixed various memory leaks fixed possible memory corruption code cleanup code cleanup more sanity checks: do not add or remove databases while a trasaction is running code cleanup fixed a corrupted argument code cleanup fixed a segfault in removdeps added .lastudate support for HTTP protocol terminated .lastupdate implementation for HTTP protocol fixed a regression - fix for sync conflicts (patch from VMiklos <vmiklos(a)frugalware.org>) tweaked a log (patch from VMiklos <vmiklos(a)frugalware.org>) added a missing cachedir registration various code cleanup - started to rename list_XXX calls to _alpm_list_XXX code cleanup - fixed a regression with pacman 2.x (reworked patch from VMiklos <vmiklos(a)frugalware.org>): fixed the usage of 4 list_is_in calls (should have been list_is_strin) moved pacman_deptest function in its own file removed ORE tags (after tests, no real performance increase by implementing them) - sync: populates *data when a conflict can't be resolved pacman_sync: handle trans_prepare error cases (patch from VMiklos <vmiklos(a)frugalware.org>) removed a misleading PM_ERR_XXX occurrence code cleanup (mainly removed line spaces at the beginning of lines by tabulations) - removed uid checks (CYGWIN) removed handle/uid checks (CYGWIN) code cleanup added more debug logs in the sync conflict/replace code Fixed a regression in the conflict handling code code cleanup removed a restriction: the data parameter of XXX_prepare functions can be NULL sync_commit can now return conflicting files with a trans_prepare like data structure (patch from VMiklos <vmiklos(a)frugalware.org>) removed all code related to PM_SYNC_TYPE_DEPEND - merged pkg_new and pkg_dummy functions code cleanup added ALLDEPS transaction flag mainly to handle "makepkg -s" made use of the new ALLDEPS transaction flag to handle "makepkg -s" fixed a memory leak in the transaction cleanup code code cleanup updated README and download comments about mtimes support fixed an awfull typo - added 2 more ORE tags (to be reviewed) in the replace/conflict sync code added more doxygen comments (patch from VMiklos <vmiklos(a)frugalware.org>) removed "^M" symbols at end of lines improved log and debug messages code cleanup code cleanup removed unuseful error codes fixed a possible memory corruption added a missing commentary from pacman 2.9.x code code cleanup maked verbose group listing enabled by -Sgg instead of -Sgv code cleanup sysupgrade: skip version cmp for pkg elected for removal (patch from VMiklos <vmiklos(a)frugalware.org>) checkdeps: sync_prepare: added a missing chunk of code for dependency checks of packages elected for removal pulled out conflict checkings from checkdeps() in its own function: checkconflicts() fixed tabs more code cleanup - code cleanup reworked log messages even more code cleanup code cleanup code cleanup added the possibility to "-Syy" (can be used to force synctrees update, even if mtimes are ok) fixed detection for duplicate entries in list of deps/conflicts more code cleanup yet added a FAKEROOT define allowing to use pacman in a fakeroot enivronment (for tests purpose) resolvedeps: return the depmiss info to the frontend in case of failure fixed a typo document package, group, sync and transaction functions (patch from VMiklos <vmiklos(a)frugalware.org>) doxygen comments update (patch from VMiklos <vmiklos(a)frugalware.org>) - fixed a compilation warning reordered group entry position in db_write (to be in line with alpm/db_write) reworked lock handling (patch from VMiklos <vmiklos(a)frugalware.fr>) reworked lock handling (patch from VMiklos <vmiklos(a)frugalware.org>) fixed a compilation warning added missing retval values in trans_release error cases added a missing error message in case of lock failure fixed potential file descriptor leaks cleaned up db_write error cases handling improved log messages for add_commit cleanup up error cases handling in pacman_deptest added a missing trans_release before pacman_sync call removed unlock code from alpm_release() Added a missing trans_release call upon lib shutdown (patch from VMiklos <vmiklos(a)frugalware.org>) doxygen support: added @brief field (patch from VMiklos <vmiklos(a)frugalware.org>) added libalpm manpage (patch from VMiklos <vmiklos(a)frugalware.org>) added line feeds when needed (patch from VMiklos <vmiklos(a)frugalware.org>) fixed typos in 2 debug logs fixed vprint implementation (patch from VMiklos <vmiklos(a)frugalware.org>) - moved yesno() from util.c to log.c fixed a possible memory leak - sync_prepare: fixed a case where a conflict should be removed from the target list code cleanup copyright update added a pmconflict_t structure to handle file conflicts added support for the PM_CONFLICT structure to handle file conflicts added alpm_pkg_checkmd5sum() to check md5sums for package from the cache added support for alpm_pkg_checkmd5sum() pacman 2.9.8 synchro copyright update copyright update synchro with pacman 2.9.8 code cleanup code factorization code factorization: take 2 error codes cleanup db_write cleanup: only write non-empty fields db_write: only write non-empty fields to database added more logs - reworked debug messages reworked debug messages added a add_faketarget function reworked logs added the possibility to force synctree update by using -Syy better use ERR than fprintf for error messages fixed a compilation warning - added a WARN macro to display warnings used MALLOC macro when possible fixed a typo fixed a compilation warning - merged db_open and db_create into one single function moved the .lastupdate support from libalpm to pacman removed a leftover comment - db_write: add support to write both local and sync entries - grep is now a static function fixed directory roots when running scriplets code cleanup *** empty log message *** sync with pacman 2.9.8 code cleanup *** empty log message *** doxygen support for autotools (patch from Christian Hamar <krics(a)linuxforum.hu> partial support for version comparators (patch from VMiklos <vmiklos(a)frugalware.org>) commented out doxygen support as it produces an error prepend library function names with _alpm (helped with the patch from VMiklos <vmiklos(a)frugalware.org>) synchro with pacman 2.9.8 fixed doxygen support (patch from VMiklos <vmiklos(a)frugalware.org>) dropped the MALLOC macro checkmd5sums enhancements the library must provide default values for main options skipped conflict resolution for targets removed from transaction set fixed outputs when downloadonly flag is set (patch from VMiklos <vmiklos(a)frugalware.org>) improved _alpm_list_free handling fixed indentation disable progress bar if the output is redirected (patch from VMiklos <vmiklos(a)frugalware.org>) added ERROR logs in case of malloc failures (patch from VMiklos <vmiklos(a)frugalware.org>) added a debug log code cleanup code cleanup reworked the db object to prepare future integration with different backends removed HoldPkg option (frontend only feature) fixed HoldPkg implementation make cache works when loading it with DEPENDS info only added more debug logs added a FLAG_NOCONFLICT fixed alpm_db_update extraction path - changed flags type from char to int added a NOSCRIPLET flag to transactions (patch from VMiklos <vmiklos(a)frugalware.org>) fixed reason loading if the pkg cache is loaded with DEPENDS only prepare code for cache loaded with only DEPENDS code cleanup removed an unexpected log fixed a commentary removed db_scan calls removed db_scan calls - take 2 fixed pkgdesc usage keep gensync and updatesync in sync db_write: fixed md5sum field all _alpm_XXX_cmp functions are now public code cleanup removed an uneeded error code (DB_UPTODATE) build be_files.c as a additional object code cleanup - removed pacman.h removed pacman.h put back treename in db->path fixed a regression (patch from VMiklos <vmiklos(a)frugalware.org>) fixed dbpath initialization (found out by VMiklos <vmiklos(a)frugalware.org>) added .cvsignore files Dan McGee (119): Discussed on IRC for a bit, this makes the following changes for clarity: * Modified info query so it doesn't show more than available when querying a * Remove -fno-strict-aliasing as it is now unnecessary to compile. adding my personal TODO file Fixing line spacings of download/install size messages. * Updated repo-add script to remove same package, different version when * Refactored some functions to clean up variable declaration. slight line spacing/indent updates just an update removal of unnecessary '\' in integrity check generation This removes the warning about skipping dependency checking when the user is Added mirror in Finland to mirrorlist * Tried to clean up newline display a bit in the frontend. Updated news file to be more relevant with previous release of pacman; Fix problem with PKGDEST variable not being used correctly. No longer necessary to explicitly specify SRCDEST, leave it up to user to Add newlines to error messages * Updated makepkg to use a new form of options. These two arrays now replace oops, ENVIRONMENT -> BUILDENV in comments * Added some backwards compatability stuff to makepkg, along with helpful fix makefiles for PKGBUILD.8 -> PKGBUILD.5 one more change for man page move I should test code before I check it in. Also a fix for the annoying wget output in PKGBUILD bug. Cleaning up mistakes. :P * Makepkg updates, and small bugfix on man page compression. * Update of all the *.po files and *.pot template files. Oops. Actuall add two new translation templates as requested on the ML- * Added a 'translation-help' file for use by our translators. Spruced up the TODOs a bit, got rid of some of them since a lot of it was Greatly simplify the way our non-existant translated manpages were being Forgot to remove a few files from configure.ac Fix newlines for real, and add a missing _() gettext wrapper. * Nice overhaul of manpages. It is at least a start. * Oops. Added some files to make "make update-po" work. Small changes to README, mostly just to postpone making any real changes. :) * Hopefully fixed the download progres bar for real. We should no longer Another round of manpage updates. Cleaned up the PKGBUILD man a bit. Man pages, man pages, man pages. Just a few more additions and clarifications, * Remove "error: " text from ERR() call since it is appended by ERR anyway. New manpages were going nowhere without an updated makefile * Fix compile errors/warnings. A few more slight changes to the pacman man page Adding repo-remove script, allows for a package to be removed from a repo. Forgot to update the Makefile with the new script, adding it. * Implemented FS #5641 as suggested in Aaron's comment- '-S' option now means * Unified description of arch variable. Recommend using single quotes in TODO update and .cvsignore addition * Fixed an i18n bug; length of description string was hardcoded previously; * Updated translation-help guide with instructions for updating one po file. Fixed locale setting issues in the frontend, and fixed description of --cachedir. Adding official translations. Thanks a lot, guys! Pierre Schmitz <pierre(a)archlinux.de> Translation Update, pt_BR: Slightly optimized to remove duplicate strcmp operation. * Updated a too-verbose message in pacman/add.c. (shouldn't be an issue withi * Removed needless checking of package infolevel- leave it to the * Updated makepkg.8 manpage to reflect shift of -S to --usesudo. * pt_BR translation updates. * strlen -> mbstowcs (multibyte str to wide char str) conversion where we * Updated Italian translation * Updated pt_BR translation (Douglas Soares de Andrade), and merged other * Updated Italian translation Updated NEWS file with some of Roman's suggestions. * Updated German translation * Removed unnecessary #define * Adding pacsearch - a script to search both the sync repos and locally Small updates, read diff for details * Added some improved debug output to file conflict checking. * Updated bash_completion script to 3.0 options. * Updated German translations. * Updated conflict checking one last time. You can finally have a file move * Slightly fixed up the check target- we now pass 16 instead of 10 tests in Thought about adding this a while back, finally remembered it. Basically a copy of the pacman-coding.html document in text form. * Update Italian Translation Fixed up the output a bit. * Slight message change for consistancy- add a ' ' (space char) at the * Fixed inconsistency of args- _alpm_db_read, _alpm_db_write. Ignore pactest-created root directory * Updated Italian translation * A lot of Makefile.am updates to try to get 'make dist' and 'make distcheck' * Rewrote pactest with an OptionParser so we can have helpful help text, and * Updated German translations. * Update the translation POT files- I don't want to do this daily but the Big commit here, I'll try to cover all the bases. * Updated distcheck target- added some flags so pactest is 100% correct. * Removed man2html stuff from configure. It is now available in the doc/ * Changes to quite a few of the tests, mostly minimal changes (better * Oops. Messed a few things up, fixing them now. * added contrib/.cvsignore, and slight change to supfile.testing * Updated Italian translation Implementing feature request: <http://bugs.archlinux.org/task/4706> * Updated Italian translation * Added some undocumented options to the manpage. * TODO updates. * db search optimization- why don't we compile the regex once instead of once * Added 5 new upgrade tests, which check various things with depends, * Updated all the supfiles for consistancy, and added a commented out line * Updated description of test upgrade052. * Prevent this error when creating empty packages. * Added another test- I believe it is a possible situation and one we need to * Some new tests. That bring new failures. :) * Added another test as suggested by Nagy on the ML- a package both provides * Implemented Douglas' suggestion from the ML- print all integrity checks * Unifying placement of REPLACES in desc file, as pacman2 does. We'll worry * Fix up _alpm_db_free, remove use of void*. * Silly mistake on making sync895 and sync896- I didn't create the field * Updated Italian translation * Modified values of INFRQ_* options so we start at 0x01 and not 0x00. This * Updated Italian translation * Slight documentation update. * Oops- I need to learn to count in hex. * A little more hacking with wchar_t output, but nothing really changed in * file utility was updated to version 4.20, removing code that was there to * Updated Italian translation * Remove unnecessary call to _alpm_trans_update_depends that was erroring * Maybe that call to _alpm_trans_update_depends was necessary. Added it back * Update distcheck flags, we no longer need to pass --disable-fakeroot. * Hopefully the LAST big update of the language files before release. Judd Vinet (72): Initial revision made makepkg echo the version-release number as well (#2375) added i18n to TODO changed license package field to a list test test patch from vmiklos -- properly remove each db entry instead of rmrf'ing the whole thing Merging in recent fixes/additions from 2.9.7 added more error messages, enabled the db permission check in trans_commit correct a trans_commit protocol fixed a segfault w/ the skiplist added conversation callback support for transactions started some doxygen comment blocks removed debug stmt added first part of the autoconf stuff - patch from Christian Hamar <krics(a)linuxforum.hu> bumped up PACMAN_VERSION added idea from christian to TODO added patch to use the system memcpy for md5 stuff added INFRQ_DESC back to the package cache so the group cache will work remove dupes from group lists in sync removed debug comment added a default cachedir Added CacheDir directive copied up-to-date scripts from pacman 2.9.7 improved the Replace question format a bit fixed a segfault that occured in special cases when propagating requiredby fields during package replacements bugfix: before searching databases for a dependency, make sure one of the packages in the final list doesn't already provide that dependency added patch from VMiklos - fixes ownership problems when building in a chroot w/ fakeroot added pacman-optimize to Makefile.am added --no-check-certificate added Doxyfile don't die on missing package files, just warn (#3535) Applied unset_lcall_lang.diff -- unset LANG/LC_ALL before building packages added Frugalware patch: -Q --foreign to display packages not found in sync repos typo added frugalware patch for DistCC support patch from VMiklos: regarding the documentation, groups have to be renamed from foo to alpm_foo verbosity fix from VMiklos VMiklos: --noconfirm fix VMiklos: add STRNCPY macro from alpm to pacman applied first try at sync conflict handling (VMiklos) plus a couple changes renamed rpmvercmp to versioncmp, added some fixes from isteve(a)bofh.cz Added .tbz2 extension to makepkg patch from VMiklos - use PACKAGE_VERSION instead of PACMAN_VERSION renamed configure.in to configure.ac renamed ChangeLog to NEWS Added AUTHORS file added Miklos and Christian typo fix In the database-against-targets conflict checks, scan db packages' provides removed old/disable provides-conflict checking code cleanup added new options field to PKGBUILDs for package-by-package overrides cleaned up format a bit, added global KEEPDOCS and NOSTRIP flags changed behaviour with original=X,current=Y,new=Z backup scenario -- install new file as .pacnew and keep old one in place little fix for erroneous output use sudo for -i if available (fw) makeworld sudo support Search through package provides if no literal matches are found when scanning for targets with -S added a --noprogressbar switch for scripts to use changes merged from 2.9.8 -- clarified license usage, new options var in PKGBUILDs, --noprogressbar pacman switch... FW: re-add CacheDir setting FW: spaces -> tabs cosmetic fix conflict checks: when doing db-vs-target checks, opt to use the NEWER, to-be-installed package instead of the local db version if there's one available first stage of i18n stuff from VMiklos patch: check strlen of str, NOT msg, which is always true i18n stuff bugfix: when looking at provides, defer to the new, to-be-installed package's provisios instead of the the existing package's spelling fix Patch from FW: Better control over CTRL-C interruptions -- do not leave the DB in an inconsistent state removed libtar support in favour of libarchive small typo fix JÃŒrgen HÃ¶tzel (9): * Fix building outside $(srcdir). * fixed path to automake (removed hardcoded version) * doxygen check cleanup * removal of swig generated bindings * removal of swig generated bindings * fixed clean target: no error if repository files do not exist * gen-mirrorlist.sh not needed anymore * fixed dist target * unset LC_ALL, LANG after sourcing /etc/profile.d/* ----------------------------------------------------------------------- hooks/post-receive -- The official pacman repository

1 0

[pacman-dev] [GIT] The official pacman repository annotated tag, v3.0.0, updated. v3.0.0
by Andrew Gregory 20 Oct '18

20 Oct '18

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "The official pacman repository". The annotated tag, v3.0.0 has been updated to c914ef466fc08048b95f999cb99c4e8431548c18 (tag) from c6b17b224f264973f32291ec0e4101d1a9fd0af1 (which is now obsolete) tagging 58fe79eef64fb87b2553e6514a47beaa6d0249c1 (commit) replaces v3.0.0-rc2 tagged by Andrew Gregory on Sat Oct 20 12:45:51 2018 -0700 - Log ----------------------------------------------------------------- replaces broken tag c6b17b224f264973f32291ec0e4101d1a9fd0af1 -----BEGIN PGP SIGNATURE----- iQEzBAABCAAdFiEEuBUbEXA3eBCVUUynu9/8kjBrESEFAlvLhe8ACgkQu9/8kjBr ESGH8QgArLmBhjuFwAaOc7Sj1THPF5BkwWqDcTQDSqMjmBdv3LskC1/ija/IsV2J k6e9FGqHZk2H/EsSuxflpDd0M0XpLAxmcHnNe1qTPoqyCBBmNpwNZDeVE6115FBX GX3q4mHQLmDJduhQJR0hHBvmUGXfKZoxNznHl/0MYZEosl6lAibwLAxdda1ro6RN XXbAVc/5b8bIsqL2o1Jj66VJnOCarUQ2KoygQ8kMjVyLcxw9g4qsYsLGKzDXgUUm 5dCIskNPDdJDGCtIPNOyPQemS5BJl3e8QFwo6OP9VgeSPhKnSTqnNlxy2mDtcwOP gPPWAN4Aya5zZa31NzifbV58GyuDhA== =gB59 -----END PGP SIGNATURE----- Aaron Griffin (14): * Missed a path when requiring that all paths end with / - this caused -U not to * Sebastian Sareyko * makepkg.conf.in: Added /usr/share/gtk-doc back into default DOC_DIRS James Rosten <seinfeld90(a)gmail.com> Nagy Gabor <ngaba(a)petra.hos.u-szeged.hu> * Fix pacman -Se which installs depends only, as it was broken * Fix asking the user to upgrade when using -Sp Giovanni Scafora <linuxmania(a)gmail.com> * Quick fixup for the translations due to a typo fix committed earlier. * -Qo now properly resolves paths when finding an owner * Correct install scriptlet usage (reuse of handle->root when not needed) I was mistaken, newpkg->data IS useful here, as the scriptlet isn't in the FS at Quick pactest changes: * Added a mirror list for [testing] (it appears most mirrors mirror testing) Dan McGee (29): * Resizing terminal while progress bars were displayed caused some weird * Small updates to my TODO file. * Fixed up broken translation * Updated Russian translation * Updated Hungarian translation * Missing a 'msg' in makepkg- pointed out by wain on Flyspray. Thanks! * Added quoting on several makepkg paths that were lacking it. James Rosten <seinfeld90(a)gmail.com> * Adding new Turkish mirror as posted on the forums. * Updated Italian translation * TODO list updates. * makepkg: fix installation of dependencies with version comparitors. * makepkg: Fix behavior of -Ssr: deps were not being correctly removed before. * makepkg: a few more changes to get dependencies working correctly. If a * NEWS: add quick line about backup files. * TODO.dan updates. * pacman.c: Add CacheDir to -v --verbose output. * Typo fix. * Updated Italian translation * Fix conflict checking to ignore symlinks that were in previous version of Lots of translation updates before we release. Roman Kyrylych <roman.kyrylych(a)gmail.com> * Updated Brazilian Portuguese translation Clarify some English messages as suggested by Nagy Gabor. I even did the hard Fixed alpm_log call- too many parameters passed. (bardo on #archlinux-pacman) * Updated German translation * Removed three dead mirrors (Dale Blount <dale(a)archlinux.org>). Andrew Fyfe <andrew(a)neptune-one.net> * Fix bug where 'makepkg -L' doesn't stop on build failure. Jürgen Hötzel (1): * dont use XferCommand if retrieving local files, fixes wget error: ----------------------------------------------------------------------- hooks/post-receive -- The official pacman repository

1 0

[pacman-dev] [PATCH] pacman: don't error when a group exists but all packages are ignored
by morganamilo 20 Oct '18

20 Oct '18

Currently when attempting to sync a group where all packages are ignored (either by ignorepkg, ignoregroup or --needed) pacman will error with "target not found". Instead, if a group has no packages, check if the group exists and only throw an error if it does not. Signed-off-by: morganamilo <morganamilo(a)gmail.com> diff --git a/lib/libalpm/alpm.h b/lib/libalpm/alpm.h index 2d3d198a..316853bb 100644 --- a/lib/libalpm/alpm.h +++ b/lib/libalpm/alpm.h @@ -1446,6 +1446,8 @@ int alpm_extract_keyid(alpm_handle_t *handle, const char *identifier, alpm_list_t *alpm_find_group_pkgs(alpm_list_t *dbs, const char *name); +int alpm_group_exists(alpm_list_t *dbs, const char *name); + /* * Sync */ diff --git a/lib/libalpm/sync.c b/lib/libalpm/sync.c index 05f58fad..57058782 100644 --- a/lib/libalpm/sync.c +++ b/lib/libalpm/sync.c @@ -313,6 +313,26 @@ alpm_list_t SYMEXPORT *alpm_find_group_pkgs(alpm_list_t *dbs, return pkgs; } +/** Check if a group exists across a list of databases. + * @param dbs the list of alpm_db_t * + * @param name the name of the group + * @return 1 if the group exists, 0 if it does not + */ +int SYMEXPORT alpm_group_exists(alpm_list_t *dbs, + const char *name) +{ + alpm_list_t *i; + for(i = dbs; i; i = i->next) { + alpm_db_t *db = i->data; + + if (alpm_db_get_group(db, name)) { + return 1; + } + } + + return 0; +} + /** Compute the size of the files that will be downloaded to install a * package. * @param newpkg the new package to upgrade to diff --git a/src/pacman/sync.c b/src/pacman/sync.c index ef8faedf..32df6e04 100644 --- a/src/pacman/sync.c +++ b/src/pacman/sync.c @@ -543,6 +543,10 @@ static int process_group(alpm_list_t *dbs, const char *group, int error) int count = alpm_list_count(pkgs); if(!count) { + if(alpm_group_exists(dbs, group)) { + return 0; + } + pm_printf(ALPM_LOG_ERROR, _("target not found: %s\n"), group); return 1; } -- 2.19.1

2 3

[pacman-dev] [PATCH v2] Port pactest to python3
by Dave Reisner 18 Oct '18

18 Oct '18

Use BytesIO instead of StringIO, and ensure that we unicode-encode data where needed. --- * Make sure pactest --review is happy configure.ac | 2 +- test/pacman/pactest.py | 5 +++-- test/pacman/pmdb.py | 4 ++-- test/pacman/pmpkg.py | 6 +++--- test/pacman/util.py | 2 +- 5 files changed, 10 insertions(+), 9 deletions(-) diff --git a/configure.ac b/configure.ac index a569ffeb..c369ca74 100644 --- a/configure.ac +++ b/configure.ac @@ -179,7 +179,7 @@ AC_SUBST(LFS_CFLAGS) AC_PROG_AWK AC_PROG_CC_C99 AC_PROG_INSTALL -AC_CHECK_PROGS([PYTHON], [python2.7 python2 python], [false]) +AC_CHECK_PROGS([PYTHON], [python3 python], [false]) AC_PATH_PROGS([BASH_SHELL], [bash bash4], [false]) # check for perl 5.10.1 (needed by makepkg-template) diff --git a/test/pacman/pactest.py b/test/pacman/pactest.py index 1f5b8483..85cce6a1 100755 --- a/test/pacman/pactest.py +++ b/test/pacman/pactest.py @@ -1,4 +1,4 @@ -#! /usr/bin/python2 +#! /usr/bin/python3 # # pactest : run automated testing on the pacman binary # @@ -45,7 +45,8 @@ def write(self, message): # duplicate stdout/stderr to a temporary file class OutputSaver(): def __init__(self): - self.save_file = tempfile.NamedTemporaryFile(prefix='pactest-output-') + self.save_file = tempfile.NamedTemporaryFile( + prefix='pactest-output-', mode='w') def __enter__(self): sys.stdout = MultiWriter(sys.stdout, self.save_file) diff --git a/test/pacman/pmdb.py b/test/pacman/pmdb.py index f7671987..95aa8756 100644 --- a/test/pacman/pmdb.py +++ b/test/pacman/pmdb.py @@ -15,9 +15,9 @@ # along with this program. If not, see <http://www.gnu.org/licenses/>. +from io import BytesIO import os import shutil -from StringIO import StringIO import tarfile import pmpkg @@ -251,7 +251,7 @@ def generate(self): filename = os.path.join(pkg.fullname(), name) info = tarfile.TarInfo(filename) info.size = len(data) - tar.addfile(info, StringIO(data)) + tar.addfile(info, BytesIO(data.encode('utf8'))) tar.close() # TODO: this is a bit unnecessary considering only one test uses it serverpath = os.path.join(self.root, util.SYNCREPO, self.treename) diff --git a/test/pacman/pmpkg.py b/test/pacman/pmpkg.py index 5a32ccd6..4667ebc1 100644 --- a/test/pacman/pmpkg.py +++ b/test/pacman/pmpkg.py @@ -14,8 +14,8 @@ # You should have received a copy of the GNU General Public License # along with this program. If not, see <http://www.gnu.org/licenses/>. +from io import BytesIO import os -from StringIO import StringIO import tarfile import util @@ -146,7 +146,7 @@ def makepkg(self, path): for name, data in archive_files: info = tarfile.TarInfo(name) info.size = len(data) - tar.addfile(info, StringIO(data)) + tar.addfile(info, BytesIO(data.encode('utf8'))) # Generate package file system for name in self.files: @@ -167,7 +167,7 @@ def makepkg(self, path): # TODO wow what a hack, adding a newline to match mkfile? filedata = name + "\n" info.size = len(filedata) - tar.addfile(info, StringIO(filedata)) + tar.addfile(info, BytesIO(filedata.encode('utf8'))) tar.close() diff --git a/test/pacman/util.py b/test/pacman/util.py index 5fbe4c35..544bdd8d 100644 --- a/test/pacman/util.py +++ b/test/pacman/util.py @@ -152,7 +152,7 @@ def getmd5sum(filename): def mkmd5sum(data): checksum = hashlib.md5() - checksum.update("%s\n" % data) + checksum.update(("%s\n" % data).encode('utf8')) return checksum.hexdigest() -- 2.19.1

1 0

[pacman-dev] [PATCH] libalpm: process needed before group selection
by morganamilo 18 Oct '18

18 Oct '18

When --needed is used, up to date packages are now filtered out before showing the group select. Signed-off-by: morganamilo <morganamilo(a)gmail.com> --- This patch set is currently incomplete. There is a problem where if every package in the group is already installed you end up with the eror. "error: target not found: groupname". Instead "there is nothing to do" should be produced instead. I'm unsure of how to solve this so I am submitting this for discussion. Currently my idea is to have alpm_find_group_pkgs gain a new param `int *exists` which the front end can then check instead of the length of the return. Or instead the needed check could just be moved to the front end. Let me know if theres a better way. diff --git a/lib/libalpm/sync.c b/lib/libalpm/sync.c index b6ae7b72..f1c02417 100644 --- a/lib/libalpm/sync.c +++ b/lib/libalpm/sync.c @@ -270,6 +270,8 @@ alpm_list_t SYMEXPORT *alpm_find_group_pkgs(alpm_list_t *dbs, for(i = dbs; i; i = i->next) { alpm_db_t *db = i->data; alpm_group_t *grp = alpm_db_get_group(db, name); + alpm_handle_t *handle = db->handle; + alpm_trans_t *trans = handle->trans; if(!grp) { continue; @@ -277,10 +279,26 @@ alpm_list_t SYMEXPORT *alpm_find_group_pkgs(alpm_list_t *dbs, for(j = grp->packages; j; j = j->next) { alpm_pkg_t *pkg = j->data; + alpm_pkg_t *local = _alpm_db_get_pkgfromcache(handle->db_local, pkg->name); if(alpm_pkg_find(ignorelist, pkg->name)) { continue; } + if(local) { + const char *localpkgname = local->name; + const char *localpkgver = local->version; + int cmp = _alpm_pkg_compare_versions(pkg, local); + + if(cmp == 0) { + if(trans != NULL && trans->flags & ALPM_TRANS_FLAG_NEEDED) { + /* with the NEEDED flag, packages up to date are not reinstalled */ + _alpm_log(handle, ALPM_LOG_WARNING, _("%s-%s is up to date -- skipping\n"), + localpkgname, localpkgver); + ignorelist = alpm_list_add(ignorelist, pkg); + continue; + } + } + } if(alpm_pkg_should_ignore(db->handle, pkg)) { alpm_question_install_ignorepkg_t question = { .type = ALPM_QUESTION_INSTALL_IGNOREPKG, -- 2.19.0

3 4

[pacman-dev] [PATCH] alpm: Fix SIGINT handling re: aborting download
by Olivier Brunel 17 Oct '18

17 Oct '18

Upon receiving SIGINT a flag is set to abort the (curl) download. However, since it was never reset/initialized, if a front-end doesn't actually exit on SIGINT, and later tries any operation that needs to perform a new download, said download would always get aborted right away due to the flag not having been reset. Signed-off-by: Olivier Brunel <jjk(a)jjacky.com> --- lib/libalpm/dload.c | 1 + 1 file changed, 1 insertion(+) diff --git a/lib/libalpm/dload.c b/lib/libalpm/dload.c index cca39470..0a3293cf 100644 --- a/lib/libalpm/dload.c +++ b/lib/libalpm/dload.c @@ -253,6 +253,7 @@ static void curl_set_handle_opts(struct dload_payload *payload, const char *useragent = getenv("HTTP_USER_AGENT"); struct stat st; + dload_interrupted = 0; /* the curl_easy handle is initialized with the alpm handle, so we only need * to reset the handle's parameters for each time it's used. */ curl_easy_reset(curl); -- 2.19.0

2 4