April 2016 - Pacman-dev - lists.archlinux.org

[pacman-dev] [PATCH v2] Report local file URL for -Sp operations if package is in cache
by Allan McRae 19 Apr '16

19 Apr '16

When using "pacman -Sp" operation to get URLs of packages to download, it is useful to know which packages are already in the file cache and do not need downloaded. Print packages in the cache with a file:// prefix. e.g $ pacman -Sp glibc file:///var/cache/pacman/glibc-2.23-1-x86_64.pkg.tar.xz Also reordered the case statements so that appropriate output is printed if an impossible fallthrough occurs in the sync case. Fixes FS#15868 Signed-off-by: Allan McRae <allan(a)archlinux.org> --- v2: - move "servers =" block in case PM_OP_SYNC out of else block - move PM_OP_SYNC below PM_OP_UPGRADE as technically a fallthrough could happen and that should go to the default output. src/pacman/util.c | 24 ++++++++++++++++++++++-- 1 file changed, 22 insertions(+), 2 deletions(-) diff --git a/src/pacman/util.c b/src/pacman/util.c index d924bf3..ed21bc8 100644 --- a/src/pacman/util.c +++ b/src/pacman/util.c @@ -1026,15 +1026,35 @@ static char *pkg_get_location(alpm_pkg_t *pkg) alpm_list_t *servers; char *string = NULL; switch(config->op) { + case PM_OP_UPGRADE: + return strdup(alpm_pkg_get_filename(pkg)); + case PM_OP_SYNC: + if(alpm_pkg_download_size(pkg) == 0) { + /* file is already in the package cache */ + alpm_list_t *i; + const char *pkgfile = alpm_pkg_get_filename(pkg); + char path[PATH_MAX]; + struct stat buf; + + for(i = alpm_option_get_cachedirs(config->handle); i; i = i->next) { + snprintf(path, PATH_MAX, "%s%s", (char *)i->data, pkgfile); + if(stat(path, &buf) == 0 && S_ISREG(buf.st_mode)) { + pm_asprintf(&string, "file://%s", path); + return string; + } + } + } + servers = alpm_db_get_servers(alpm_pkg_get_db(pkg)); if(servers) { pm_asprintf(&string, "%s/%s", (char *)(servers->data), alpm_pkg_get_filename(pkg)); return string; } - case PM_OP_UPGRADE: - return strdup(alpm_pkg_get_filename(pkg)); + + /* fallthrough should never occur */ + default: pm_asprintf(&string, "%s-%s", alpm_pkg_get_name(pkg), alpm_pkg_get_version(pkg)); return string; -- 2.7.4

2 1

[pacman-dev] [PATCH] makepkg: Add DEBUGDEST for a pkgtype of debug
by Earnestly 18 Apr '16

18 Apr '16

Fixes FS#48902. This patch piggybacks on makepkg's internal knowledge of the package name to move packages of type "debug" (more on this later) to a user defined directory. The goal is to make moving only debug packages to a debug repo in a reliable manner without using globbing to try and guess if a package is a debug package or not. This concept patch adds a new global variable called "pkgtype" which defines the kind of package it is, currently the only pkgtype is "debug". I've left non-debug pkgtypes to indicate a normal package but it may be potentially expanded with different types of package in future. Currently pkg_file is updated based on the variable and prefers DEBUGDEST if it is defined otherwise falling back on PKGDEST which essentially treats debug packages as normal packages if DEBUGDEST is not configured. Signed-off-by: Earnestly <zibeon(a)gmail.com> --- doc/makepkg.conf.5.txt | 5 +++++ etc/makepkg.conf.in | 3 +++ scripts/makepkg.sh.in | 8 ++++++++ 3 files changed, 16 insertions(+) diff --git a/doc/makepkg.conf.5.txt b/doc/makepkg.conf.5.txt index 5bd3472..53502d6 100644 --- a/doc/makepkg.conf.5.txt +++ b/doc/makepkg.conf.5.txt @@ -224,6 +224,11 @@ Options like to keep all their packages in one place so this option allows for this behavior. A common location is ``/home/packages''. +**DEBUGDEST**"/path/to/directory":: + If this value is not set, debug packages will by default honor `PKGDEST`. + This allows people to configure a directory where debug packages will be + placed. + **SRCDEST=**"/path/to/directory":: If this value is not set, downloaded source files will only be stored in the current directory. Many people like to keep all source files in diff --git a/etc/makepkg.conf.in b/etc/makepkg.conf.in index efac16a..78220f8 100644 --- a/etc/makepkg.conf.in +++ b/etc/makepkg.conf.in @@ -108,6 +108,9 @@ PURGE_TARGETS=(usr/{,share}/info/dir .packlist *.pod) # #-- Destination: specify a fixed directory where all packages will be placed #PKGDEST=/home/packages +#-- Destination: specify a fixed directory where all debug packages will be +# placed, uses PKGDEST if not set +#DEBUGDEST=/home/packages #-- Source cache: specify a fixed directory where source files will be cached #SRCDEST=/home/sources #-- Source packages: specify a fixed directory where all src packages will be placed diff --git a/scripts/makepkg.sh.in b/scripts/makepkg.sh.in index f80e37a..4afa4f2 100644 --- a/scripts/makepkg.sh.in +++ b/scripts/makepkg.sh.in @@ -1106,6 +1106,13 @@ create_package() { done # tar it up + case $pkgtype in + # If the user hasn't configured DEBUGDEST then we need to fall back on + # PKGDEST so the package has somewhere to go. + debug) PKGDEST=${DEBUGDEST:-$PKGDEST} ;; + *) PKGDEST=$PKGDEST ;; + esac + local fullver=$(get_full_version) local pkg_file="$PKGDEST/${pkgname}-${fullver}-${pkgarch}${PKGEXT}" local ret=0 @@ -1174,6 +1181,7 @@ create_debug_package() { fi pkgdir="${pkgdir}-@DEBUGSUFFIX@" + pkgtype=debug # check if we have any debug symbols to package if dir_is_empty "$pkgdir/usr/lib/debug"; then -- 2.8.0

2 2

[pacman-dev] [PATCH] pactree: Add support for getting optional dependencies
by Johannes Löthberg 18 Apr '16

18 Apr '16

Signed-off-by: Johannes Löthberg <johannes(a)kyriasis.com> --- doc/pactree.8.txt | 3 +++ src/util/pactree.c | 41 ++++++++++++++++++++++++++++++++++++----- 2 files changed, 39 insertions(+), 5 deletions(-) diff --git a/doc/pactree.8.txt b/doc/pactree.8.txt index b177788..4c391c0 100644 --- a/doc/pactree.8.txt +++ b/doc/pactree.8.txt @@ -49,6 +49,9 @@ Options *-l, \--linear*:: Prints package names at the start of each line, one per line. +*-o, \--optdeps*:: + Show packages that are optionally depended upon by the named package. + *-r, \--reverse*:: Show packages that depend on the named package. diff --git a/src/util/pactree.c b/src/util/pactree.c index 67be9f9..fb91b39 100644 --- a/src/util/pactree.c +++ b/src/util/pactree.c @@ -118,6 +118,7 @@ struct graph_style *style = &graph_utf8; int graphviz = 0; int max_depth = -1; int reverse = 0; +int optdeps = 0; int unique = 0; int searchsyncs = 0; const char *dbpath = DBPATH; @@ -182,6 +183,7 @@ static int parse_options(int argc, char *argv[]) {"help", no_argument, 0, 'h'}, {"linear", no_argument, 0, 'l'}, {"reverse", no_argument, 0, 'r'}, + {"optdeps", no_argument, 0, 'o'}, {"sync", no_argument, 0, 's'}, {"unique", no_argument, 0, 'u'}, @@ -198,7 +200,7 @@ static int parse_options(int argc, char *argv[]) style = &graph_default; #endif - while((opt = getopt_long(argc, argv, "ab:cd:ghlrsu", opts, &option_index))) { + while((opt = getopt_long(argc, argv, "ab:cd:ghlorsu", opts, &option_index))) { if(opt < 0) { break; } @@ -236,6 +238,9 @@ static int parse_options(int argc, char *argv[]) case 's': searchsyncs = 1; break; + case 'o': + optdeps = 1; + break; case 'u': unique = 1; style = &graph_linear; @@ -266,6 +271,7 @@ static void usage(void) " -g, --graph generate output for graphviz's dot\n" " -h, --help display this help message\n" " -l, --linear enable linear output\n" + " -o, --optdeps list packages that are optionally depended upon by the named package\n" " -r, --reverse list packages that depend on the named package\n" " -s, --sync search sync databases instead of local\n" " -u, --unique show dependencies with no duplicates (implies -l)\n" @@ -381,12 +387,29 @@ static alpm_list_t *get_pkg_dep_names(alpm_pkg_t *pkg) return names; } +static alpm_list_t *get_pkg_optdep_names(alpm_pkg_t *pkg) +{ + alpm_list_t *i = NULL, *names = NULL; + for(i = alpm_pkg_get_optdepends(pkg); i; i = alpm_list_next(i)) { + alpm_depend_t *d = i->data; + if(searchsyncs) { + names = alpm_list_add(names, d->name); + } else { + alpm_db_t *localdb = alpm_get_localdb(handle); + if(alpm_find_satisfier(alpm_db_get_pkgcache(localdb), d->name)) { + names = alpm_list_add(names, d->name); + } + } + } + return names; +} + /** * walk dependencies, showing dependencies of the target */ -static void walk_deps(alpm_list_t *dblist, alpm_pkg_t *pkg, tdepth *depth, int rev) +static void walk_deps(alpm_list_t *dblist, alpm_pkg_t *pkg, tdepth *depth, int rev, int optdeps) { - alpm_list_t *deps, *i; + alpm_list_t *deps = NULL, *i = NULL; if(!pkg || ((max_depth >= 0) && (depth->level > max_depth))) { return; @@ -396,8 +419,16 @@ static void walk_deps(alpm_list_t *dblist, alpm_pkg_t *pkg, tdepth *depth, int r if(rev) { deps = alpm_pkg_compute_requiredby(pkg); + if(optdeps) { + alpm_list_t *odeps = alpm_pkg_compute_optionalfor(pkg); + deps = alpm_list_join(deps, odeps); + } } else { deps = get_pkg_dep_names(pkg); + if(optdeps) { + alpm_list_t *odeps = get_pkg_optdep_names(pkg); + deps = alpm_list_join(deps, odeps); + } } for(i = deps; i; i = alpm_list_next(i)) { @@ -431,7 +462,7 @@ static void walk_deps(alpm_list_t *dblist, alpm_pkg_t *pkg, tdepth *depth, int r d.prev = NULL; } } - walk_deps(dblist, dep_pkg, &d, rev); + walk_deps(dblist, dep_pkg, &d, rev, optdeps); depth->next = NULL; } } @@ -494,7 +525,7 @@ int main(int argc, char *argv[]) NULL, 1 }; - walk_deps(dblist, pkg, &d, reverse); + walk_deps(dblist, pkg, &d, reverse, optdeps); print_end(); -- 2.7.1

3 4

[pacman-dev] [PATCH] [RFC] makepkg: move build enviroment set-up to function
by Allan McRae 15 Apr '16

15 Apr '16

This is the first set require to split the preparation of the build environment into libmakepkg, which will allow dropping in extensions (e.g. to allow PGO). After this patch, disabling buildflags or makeflags and enabling debug CFLAGS will only effect the build() and check() functions. These are no longer exported for the prepare() and package() function. This should have zero impact for the prepare() and package() functions of a properly written PKGBUILD, as no building/linking is done there... Signed-off-by: Allan McRae <allan(a)archlinux.org> --- Can anyone think of a case where a well written PKGBUILD will be affected by this change to handling '!buildflags', '!makeflags' and 'debug'? scripts/makepkg.sh.in | 77 +++++++++++++++++++++++++++------------------------ 1 file changed, 41 insertions(+), 36 deletions(-) diff --git a/scripts/makepkg.sh.in b/scripts/makepkg.sh.in index f80e37a..beb90b9 100644 --- a/scripts/makepkg.sh.in +++ b/scripts/makepkg.sh.in @@ -763,6 +763,45 @@ source_buildfile() { source_safe "$@" } +prepare_buildenv() { + # clear user-specified buildflags if requested + if check_option "buildflags" "n"; then + unset CPPFLAGS CFLAGS CXXFLAGS LDFLAGS + fi + + if check_option "debug" "y"; then + CFLAGS+=" $DEBUG_CFLAGS" + CXXFLAGS+=" $DEBUG_CXXFLAGS" + fi + + # clear user-specified makeflags if requested + if check_option "makeflags" "n"; then + unset MAKEFLAGS + fi + + # ensure all necessary build variables are exported + export CPPFLAGS CFLAGS CXXFLAGS LDFLAGS MAKEFLAGS CHOST + + local ccache=0 + + # use ccache if it is requested (check buildenv and PKGBUILD opts) + if check_buildoption "ccache" "y" && [[ -d /usr/lib/ccache/bin ]]; then + export PATH="/usr/lib/ccache/bin:$PATH" + ccache=1 + fi + + # use distcc if it is requested (check buildenv and PKGBUILD opts) + if check_buildoption "distcc" "y"; then + if (( ccache )); then + export CCACHE_PREFIX="${CCACHE_PREFIX:+$CCACHE_PREFIX }distcc" + export CCACHE_BASEDIR="$srcdir" + elif [[ -d /usr/lib/distcc/bin ]]; then + export PATH="/usr/lib/distcc/bin:$PATH" + fi + export DISTCC_HOSTS + fi +} + run_function_safe() { local restoretrap @@ -786,26 +825,9 @@ run_function() { fi local pkgfunc="$1" - # clear user-specified buildflags if requested - if check_option "buildflags" "n"; then - unset CPPFLAGS CFLAGS CXXFLAGS LDFLAGS - fi - - if check_option "debug" "y"; then - CFLAGS+=" $DEBUG_CFLAGS" - CXXFLAGS+=" $DEBUG_CXXFLAGS" - fi - - # clear user-specified makeflags if requested - if check_option "makeflags" "n"; then - unset MAKEFLAGS - fi - msg "$(gettext "Starting %s()...")" "$pkgfunc" cd_safe "$srcdir" - # ensure all necessary build variables are exported - export CPPFLAGS CFLAGS CXXFLAGS LDFLAGS MAKEFLAGS CHOST # save our shell options so pkgfunc() can't override what we need local shellopts=$(shopt -p) @@ -847,25 +869,6 @@ run_prepare() { } run_build() { - local ccache=0 - - # use ccache if it is requested (check buildenv and PKGBUILD opts) - if check_buildoption "ccache" "y" && [[ -d /usr/lib/ccache/bin ]]; then - export PATH="/usr/lib/ccache/bin:$PATH" - ccache=1 - fi - - # use distcc if it is requested (check buildenv and PKGBUILD opts) - if check_buildoption "distcc" "y"; then - if (( ccache )); then - export CCACHE_PREFIX="${CCACHE_PREFIX:+$CCACHE_PREFIX }distcc" - export CCACHE_BASEDIR="$srcdir" - elif [[ -d /usr/lib/distcc/bin ]]; then - export PATH="/usr/lib/distcc/bin:$PATH" - fi - export DISTCC_HOSTS - fi - run_function_safe "build" } @@ -2122,6 +2125,8 @@ else chmod a-srw "$pkgdirbase" cd_safe "$startdir" + prepare_buildenv + if (( ! REPKG )); then (( BUILDFUNC )) && run_build (( CHECKFUNC )) && run_check -- 2.7.4

1 1

[pacman-dev] [PATCH v2] Fix handling of unavailable backup hashes
by Will Miles 11 Apr '16

11 Apr '16

This patch fixes several places where an unavailable (NULL) backup hash may be passed to a printf() call, which segfaults on some libcs. This includes local database writes where unavailable hashes are represented by the file name only. This condition may occur when an updated package changes the contents of its backup=() list. Signed-off-by: Will Miles <wmiles(a)sgl.com> --- lib/libalpm/add.c | 6 +++--- lib/libalpm/be_local.c | 6 +++++- 2 files changed, 8 insertions(+), 4 deletions(-) diff --git a/lib/libalpm/add.c b/lib/libalpm/add.c index f5c9a95..629e2a5 100644 --- a/lib/libalpm/add.c +++ b/lib/libalpm/add.c @@ -339,9 +339,9 @@ static int extract_single_file(alpm_handle_t *handle, struct archive *archive, hash_pkg = backup ? backup->hash : alpm_compute_md5sum(filename); _alpm_log(handle, ALPM_LOG_DEBUG, "checking hashes for %s\n", origfile); - _alpm_log(handle, ALPM_LOG_DEBUG, "current: %s\n", hash_local); - _alpm_log(handle, ALPM_LOG_DEBUG, "new: %s\n", hash_pkg); - _alpm_log(handle, ALPM_LOG_DEBUG, "original: %s\n", hash_orig); + _alpm_log(handle, ALPM_LOG_DEBUG, "current: %s\n", (hash_local ? hash_local : "NULL") ); + _alpm_log(handle, ALPM_LOG_DEBUG, "new: %s\n", (hash_pkg ? hash_pkg : "NULL") ); + _alpm_log(handle, ALPM_LOG_DEBUG, "original: %s\n", (hash_orig ? hash_orig : "NULL") ); if(hash_local && hash_pkg && strcmp(hash_local, hash_pkg) == 0) { /* local and new files are the same, updating anyway to get diff --git a/lib/libalpm/be_local.c b/lib/libalpm/be_local.c index f817822..21166cb 100644 --- a/lib/libalpm/be_local.c +++ b/lib/libalpm/be_local.c @@ -1037,7 +1037,11 @@ int _alpm_local_db_write(alpm_db_t *db, alpm_pkg_t *info, alpm_dbinfrq_t inforeq fputs("%BACKUP%\n", fp); for(lp = info->backup; lp; lp = lp->next) { const alpm_backup_t *backup = lp->data; - fprintf(fp, "%s\t%s\n", backup->name, backup->hash); + if (backup->hash) { + fprintf(fp, "%s\t%s\n", backup->name, backup->hash); + } else { + fprintf(fp, "%s\n", backup->name); + } } fputc('\n', fp); } -- 2.6.3

2 1

[pacman-dev] [PATCH] Fix segfault during install when a package update adds a new file to 'backup'
by Will Miles 08 Apr '16

08 Apr '16

When a new package release adds a new entry to the backup list, no hash exists for that file in the previously installed package version. Fix two places where that nonexistent hash may be dereferenced. Signed-off-by: Will Miles <wmiles(a)sgl.com> --- lib/libalpm/add.c | 2 +- lib/libalpm/be_local.c | 4 +++- 2 files changed, 4 insertions(+), 2 deletions(-) diff --git a/lib/libalpm/add.c b/lib/libalpm/add.c index f5c9a95..ff0b81d 100644 --- a/lib/libalpm/add.c +++ b/lib/libalpm/add.c @@ -341,7 +341,7 @@ static int extract_single_file(alpm_handle_t *handle, struct archive *archive, _alpm_log(handle, ALPM_LOG_DEBUG, "checking hashes for %s\n", origfile); _alpm_log(handle, ALPM_LOG_DEBUG, "current: %s\n", hash_local); _alpm_log(handle, ALPM_LOG_DEBUG, "new: %s\n", hash_pkg); - _alpm_log(handle, ALPM_LOG_DEBUG, "original: %s\n", hash_orig); + _alpm_log(handle, ALPM_LOG_DEBUG, "original: %s\n", (hash_orig ? hash_orig : "NULL") ); if(hash_local && hash_pkg && strcmp(hash_local, hash_pkg) == 0) { /* local and new files are the same, updating anyway to get diff --git a/lib/libalpm/be_local.c b/lib/libalpm/be_local.c index f817822..5e39199 100644 --- a/lib/libalpm/be_local.c +++ b/lib/libalpm/be_local.c @@ -1037,7 +1037,9 @@ int _alpm_local_db_write(alpm_db_t *db, alpm_pkg_t *info, alpm_dbinfrq_t inforeq fputs("%BACKUP%\n", fp); for(lp = info->backup; lp; lp = lp->next) { const alpm_backup_t *backup = lp->data; - fprintf(fp, "%s\t%s\n", backup->name, backup->hash); + if (backup->hash) { + fprintf(fp, "%s\t%s\n", backup->name, backup->hash); + } } fputc('\n', fp); } -- 2.6.3

3 5

[pacman-dev] [PATCH v2 0/1] libmakepkg: extension for additional BUILDENV options
by quequotion＠gmail.com 02 Apr '16

02 Apr '16

From: Que Quotion <quequotion(a)gmail.com> Resubmitting as requested, in the format requested, by the means requested. buildenv_ext.sh does for BUILDENV what tidy.sh does for OPTIONS --- scripts/Makefile.am | 2 ++ scripts/libmakepkg/buildenv_ext.sh.in | 42 +++++++++++++++++++++++++++++++++++ scripts/makepkg.sh.in | 3 +++ 3 files changed, 47 insertions(+) create mode 100644 scripts/libmakepkg/buildenv_ext.sh.in -- 2.7.4

4 9

[pacman-dev] Open up a place for BUILDENV extensions
by Que Quotion 01 Apr '16

01 Apr '16

>How exactly is this related to BUILDENV? This seems to be allowing arbitrary code to run before executing the build() function. You can use the word "arbitrary" to make it sound like a bad idea if you like, but it very simply does for BUILDENV exactly what tidy.sh does for OPTIONS: allow for supplemental scripts to add to makepkg's functionality at this stage. In fact, it does it in exactly the same way--because nearly all of this code was copied and pasted from tidy.sh. Just as tidy.sh allows any kind of anything to be run, it could allow any bash script to be run here--but the intention is for scripts that serve build-environment altering functions to be run here (as the intention for tidy.sh is to run cleanup and compression scripts). Your concern about cleanup seems misplaced. It will be up to the authors of supplemental scripts to make sure they work properly, and users who trust them to install and enable those features. This patch--in and of itself--will not cause the problems you are worried about. >This line does nothing, even if correctly spelled. What is misspelled? This is intended to serve the same purpose as the same line serves in tidy.sh: declare -a packaging_options tidy_remove tidy_modify The tidy scripts have names like upx.sh, the buildenv_ext scripts will have names like svgo.sh; these variables store the names of those options as they will be known in makepkg(-optimize).conf. >Since nullglob isn't set, an empty directory will cause this to throw an error about a missing file (as the glob will turn into a literal). As there's no "buildenv_ext" files provided by makepkg, this means we error by default. That could be fixed very easily, but again: copy pasta from tidy.sh. If you have a problem with this code--you should have a problem with that code too. Furthermore, as I have suggested in the forum (and posted a patch for), the first thing I would do with this if it were up to me is move ccache and distcc out of makepkg and ship them as builenv_ext scripts. Then you'd be shipping something that makes this unsafe code work just the way the same unsafe code works in tidy.sh. >quoting Sure, no problem. That also needs to be fixed in tidy.sh.

1 0

[pacman-dev] Fwd: Re: [PATCH] libmakepkg: extension for additional BUILDENV options
by Alad Wenter 01 Apr '16

01 Apr '16

https://www.archlinux.org/pacman/submitting-patches.html And please, click "Reply list" next time. -------- Forwarded Message -------- Subject: Re: [PATCH] libmakepkg: extension for additional BUILDENV options Date: Sat, 2 Apr 2016 04:14:58 +0900 From: Que Quotion <quequotion(a)gmail.com> To: alad(a)archlinux.info >Talking of unncessery email, can you please just follow convention instead of spamming our mailboxes with semantics? I have an even better idea: have your conventions documented somewhere. If they already are, send people who obviously weren't born knowing them to that place and exercise a little understanding.

2 1

Re: [pacman-dev] [PATCH] libmakepkg: extension for additional BUILDENV options
by Que Quotion 01 Apr '16

01 Apr '16

>>apg Until a very few moments ago, I was not subscribed to this mailing list. I do not intend to be subscribed after this is resolved. There is no other issue on the mailing list for which it is necessary for me to subscribe and I care very much when unnecessary email comes to my inbox. There's no way to get header information without being a subscriber, so I'm never going to get the Message-ID for any past messages. As a result, I'm left with no choice but to post my patches (with all the appropriate git-format bells and whistles) as either a new thread or a reply to this one. I'm going with new thread, since it is apparently standard practice to indicate patch version numbers in the subject line and--last week--I did bump the patch to v2.

2 1