The modification time on depends and desc file were changed to match the
modification time of the package file. I don't see why and we are actualling
losing information here. If we want to know the date of the package file, we
can just look inside the depends file. If we want to know when the entry was
created, we should not alter the modification time of depends and desc.
Besides, this had the non-obvious and undocumented side effect that the
depends file was always created, even if it was empty. And pacman actually
does require that. So I added a "touch depends" to always create the file.
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
Weird things could happen if several repo-add were run concurrently on the
same database. The introduced locking system will prevent this to happen.
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
- arch was missing
- backup is not used by repo-add. However makepkg still needs to put it in
PKGINFO because pacman uses it
- startdir is no longer used after the new delta implementation
- the declaration of group, depend, backup, etc is not needed because these
variables are always declared before being used :
declare $var="$val"
case "$var" in
group) _groups="$_groups$group\n" ;;
- reorder the variables declaration to follow the same order than they are
written to the depends and desc file, for making future checks easier
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
Use the correct database format
Use xdelta3 to get the source and destination files from the delta itself
Allow delta files to be added with repo-add just like package files. delta
files can also be removed with repo-remove. This is simply done by looking
for a .delta extension in the arguments, and calling the appropriate
db_write_delta or db_remove_delta functions.
Example usage:
repo-add repo/test.db.tar.gz repo/libx11-1.1.99.2-2-x86_64.pkg.tar.gz
repo-add repo/test.db.tar.gz repo/libx11-1.1.5-2_to_1.1.99.2-2-x86_64.delta
repo-remove repo/test.db.tar.gz libx11-1.1.5-2_to_1.1.99.2-2-x86_64.delta
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
The current implementation has several problems :
Wrong database format
All the info is taken from the filename, which is a bit ugly
It looks for .delta files in the current directory when adding a package,
which is not very flexible
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
* report when a package entry to be removed is not found
* backup and restore eventual "deltas" files
* slight optimization when looking for an entry : only look at the entries
starting with $pkgname
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
This should obsolete the delta support in makepkg. Having a separate script
should be more flexible.
Example usage:
$ pkgdelta repo/tzdata-2009a-1-x86_64.pkg.tar.gz repo/tzdata-2009b-1-x86_64.pkg.tar.gz
==> Generating delta from version 2009a-1 to version 2009b-1
==> Generated delta : 'repo/tzdata-2009a-1_to_2009b-1-x86_64.delta'
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
1) The changes to sync.c look big but there are mostly caused by
the indentation. Fix a bug where download_size == 0 because the packages and
deltas are already in the cache, but we still need to build the deltas list
and apply the deltas to create the final package.
2) Fix the gzip / md5sum issue by switching to xdelta3, disabling external
recompression and using gzip -n in pacman, and disable bsdtar compression
and using gzip -n in makepkg.
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
This reverts commit 9558639d80.
This change was wrong, popen does require /bin/sh in a subchroot.
1) pacman -S lilo -r root
Notice no error
2) rm root/bin/sh ; pacman -S lilo -r root
Notice an error :
error: scriptlet failed to execute correctly
Actually, we already get an explicit error here, when popen is run, so there
is no need to check for bin/sh explicitely.
Besides this check was problematic in some cases. For example, bash itself
has a scriptlet, but only post_install and post_upgrade, no pre_install and
pre_upgrade. However, since bash has a scriptlet, runscriptlet will also be
called before bash is installed. It won't do anything since the scriptlet
has no pre_install function. But if we keep the check, we will still get
"error : no /bin/sh".
Conflicts:
lib/libalpm/trans.c
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
This is a bug I noticed 2 years ago :
http://www.nabble.com/Re%3A-logging-output-crazy-to11437357.html#a11479679
I thought I fixed it with 57d77eab32
But the bug was still here. Reading man fork, this part caught my attention:
* The child inherits copies of the parent's set of open file
descriptors. Each file descriptor in the child refers to the same open
file description (see open(2)) as the corresponding file descriptor in the
parent. This means that the two descriptors share open file status
flags, current file offset, and signal-driven I/O attributes (see the
description of F_SETOWN and F_SETSIG in fcntl(2)).
Since the open file descriptors are inherited, it is probably a good idea to
flush them before forking.
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
Prints the install script from a given package file or from a package
in the pacman repo.
Original-work-by: Giulio "giulivo" Fidente <giulivo.navigante@gmail.com>
Improvements-by: Xavier Chantry <shiningxc@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
Signed-off-by: Dan McGee <dan@archlinux.org>
The pkgdelta script can be used instead.
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
Signed-off-by: Allan McRae <allan@archlinux.org>
Signed-off-by: Dan McGee <dan@archlinux.org>
The from_md5 and to_md5 fields were a nice extra safety, which would avoid
trying to apply deltas on corrupted package files. However, they are not
strictly necessary, since xdelta should be able to detect that on its own.
The main problem is that it is impossible to compute these informations from
the delta only. So repo-add would not be able to compute the delta entry
based on just the delta file.
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
Don't prompt the user for unignore of IgnorePkg/IgnoreGroup packages,
except for packages explicitly listed for sync by the user. This
eliminates many unnecessary prompts when IgnorePkg/IgnoreGroup is
used.
Signed-off-by: Bryan Ischo <bryan@ischo.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
Enabled a new prompt to ask the user if they'd like to remove
unresolvable packages from the transaction rather than failing it.
Many pactest tests that used to fail now return success codes, because
pacman now issues a prompt allowing the user to cancel rather than
failing many transactions, and the pactest scripts always choose to
cancel with no error rather than failing. The only net effect is that
the return status of pacman is now 0 in cases where it used to be
nonzero.
Signed-off-by: Bryan Ischo <bryan@ischo.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
Commit 1e656c0a introduced the changing of pkgrel to 1 when the
pkgver was updated in SCM PKGBUILDs. However, the output in the
"Making package:" was wrong. Attempting to fix that created
another bug (FS#13416). Interestingly, pkgver was only ever
being updated in the fakeroot stage which caused this problem.
Now both pkgver and pkgrel are updated after the first
devel_check and devel_update. Enjoy the really long explaination
for a two line fix...
Signed-off-by: Allan McRae <allan@archlinux.org>
Signed-off-by: Dan McGee <dan@archlinux.org>
Rework slightly db_write_entry so that $pkgfile is no longer referenced
from the temporary dir. This means $pkgfile can be a relative path and does
not need to be converted with realpath anymore.
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
REPO_DB_FILE does not need to be an absolute path anymore so no need to
call realpath.
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
This change reorganizes the internal code so that packages are
resolved one at a time instead of all at once from a list. This will
allow a future checkin to prompt the user to see if they'd rather
remove unresolvable packages from the transaction and continue, or
fail the transaction. This change does not affect the actual behavior
of libalpm and all tests pass without changes.
Signed-off-by: Bryan Ischo <bryan@ischo.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
The previous translation of 'targets' meant literally 'it targets' and it
sounded awkwardly. The current version is a plural of a 'target'.
Signed-off-by: Jan Stępień <jstepien@users.sourceforge.net>
Signed-off-by: Dan McGee <dan@archlinux.org>
This function was used only once, was basically just one line, and was also
called with an unused argument.
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
Signed-off-by: Dan McGee <dan@archlinux.org>
I initially only wanted to add a -l/--locate option to use locate instead of
find, which should have been easy.
Then I thought I would try to support filename with whitespace while I was
at it, and this was a bit more complex. The safest ways seem to be the
following ones : http://mywiki.wooledge.org/BashFAQ/020
Then I received a lot of suggestions on #bash about how to improve the
script, which I tried to address.
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
[Dan: fix grouping of find arguments]
Signed-off-by: Dan McGee <dan@archlinux.org>
Change configure.ac to use the full path of stat when on darwin/mac.
This is needed for situations when a user installs the GNU/coreutils
and places it in their path before /usr/bin, but the SIZECMD is
already configured for Darwin's version of stat.
Signed-off-by: Kevin Barry <barryk gmail com>
Signed-off-by: Dan McGee <dan@archlinux.org>
eval was ugly and dirty, and bit us here. Instead, use a safer form of
variable declaration to ensure quotes don't foil us in pkgdesc or any other
fields.
This fixes FS#10837.
Signed-off-by: Dan McGee <dan@archlinux.org>
The Linux and BSD versions of strip have the --strip-debug option (as
well as the -S option), however Mac OS X only has -S.
Signed-off-by: Allan McRae <allan@archlinux.org>
Signed-off-by: Dan McGee <dan@archlinux.org>
We don't need to create the directories when local or sync dbs are
registered. For example, if a sync db does not exist, we cannot even do
"pacman -Q" as an user.
Instead, we can create the local db if needed during the db_prepare
operation, and sync dbs on db_update.
Also remove some more useless abstractions in db_update and switch to a much
more efficient way to remove a sync db : rm -rf.
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
These db_open and db_close looked quite useless. And they caused the db
directory to be opened on a simple registering of a database. This is
totally unneeded, this opening can be delayed to when we actually need it.
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
Instead of appending the prefix to each entry name, we can chdir to the
prefix before extracting, and restoring when it is done.
This seems to work better with the strange and special case of FS#12148
where an archive contained the "./" entry.
Signed-off-by: Xavier Chantry <shiningxc@gmail.com>
The goal of this fix was empty string comparisons:
- if [ "$pkgname" != "" ]; then
+ if [ -n "$pkgname" ]; then
Signed-off-by: Dan McGee <dan@archlinux.org>
In a lot of places, we had the following construct:
[ "$foobar" = "0" ]
which is better represented by using the integer tests:
[ $foobar -eq 0 ]
Attempt to unify makepkg to use the latter rather than the former in all
places. From here on out we should ensure anything that is set to 0, 1, etc.
uses the -eq format rather than =.
In addition, fix a few other test anomalies including usage of double
equals.
Signed-off-by: Dan McGee <dan@archlinux.org>
This patch fixes FS#12148 ('unstable' regular file).
I also changed the other archive_entry_set_mode usage in add.c to
archive_entry_set_perm.
Since I cannot find any relevant info in libarchive manual, I quote
Tim Kientzle (the author of libarchive) here, and I say thank you for
his help.
*** Tim Kientzle wrote *************************************
This is the problem in libalpm/util.c:
323 if(S_ISREG(st->st_mode)) {
324 archive_entry_set_mode(entry, 0644);
325 } else if(S_ISDIR(st->st_mode)) {
326 archive_entry_set_mode(entry, 0755);
327 }
Your example unstable.db.tar.gz is not empty. It has
one entry in it, called "./". That entry is marked
as a directory. But, when you call archive_entry_set_mode(),
you are changing the file type! archive_read_extract()
then creates the file /var/unstable as you requested.
(archive_read_extract() will replace an empty directory
with a file.)
You should either set the mode value correctly:
323 if(S_ISREG(st->st_mode)) {
324 archive_entry_set_mode(entry, IFREG | 0644);
325 } else if(S_ISDIR(st->st_mode)) {
326 archive_entry_set_mode(entry, IFDIR | 0755);
327 }
Or use archive_entry_set_perm(), which does not change
the file type:
323 if(S_ISREG(st->st_mode)) {
324 archive_entry_set_perm(entry, 0644);
325 } else if(S_ISDIR(st->st_mode)) {
326 archive_entry_set_perm(entry, 0755);
327 }
************************************************************
Signed-off-by: Nagy Gabor <ngaba@bibl.u-szeged.hu>
Signed-off-by: Dan McGee <dan@archlinux.org>
When using the optional package() function or split packages, the
entire packaging step is rerun instead of just final package
creation step.
Signed-off-by: Allan McRae <allan@archlinux.org>
Signed-off-by: Dan McGee <dan@archlinux.org>
* am/splitpkg:
makepkg: implement creation of split packages
makepkg: Optional argument for run_package and create_package
makepkg: hack around tee in run_package function
makepkg: add functions for backup and restore of package fields
makepkg: add optional package function
makepkg: Add PKGBUILD-split.proto
Generalize run_package to allow the passing of a package name
directing the use of an alternative package function. A similar
adjustment to create_package to prepare split packages.
Signed-off-by: Allan McRae <allan@archlinux.org>
Piping the package function through tee to log the outut also
clears any variables set in the package function. This is a
problem in split packages as package variable overrides are done
in the package function. This is fixed by creating a node which
the output is piped through and duplicated using the tee function.
Signed-off-by: Allan McRae <allan@archlinux.org>
