]> git.saurik.com Git - apt.git/log
apt.git
10 years agodo not revert candidate for protected packages
David Kalnischkies [Sat, 7 Jun 2014 20:46:37 +0000 (22:46 +0200)] 
do not revert candidate for protected packages

In commit 21b3eac8 I promoted the check for installable dependencies to
a pre-install check, which also reverts to a known good candidate (the
installed version) if it fails. This revert was done even for user
requested candidate switches which disabled our Broken detection so that
install requests which are impossible to satisfy do not fail anymore,
but print an (incomplete) solution proposal and then exit successfully.

Closes: 745046
10 years agoItalian program translation update
Milo Casagrande [Sat, 7 Jun 2014 15:17:20 +0000 (17:17 +0200)] 
Italian program translation update

Closes: 750009
10 years agocmdline/apt-helper.cc: use less generic description/short-description
Michael Vogt [Fri, 6 Jun 2014 09:24:44 +0000 (11:24 +0200)] 
cmdline/apt-helper.cc: use less generic description/short-description

10 years agouse 'native' instead of 'amd64' as pkg arch
David Kalnischkies [Fri, 30 May 2014 21:15:28 +0000 (23:15 +0200)] 
use 'native' instead of 'amd64' as pkg arch

Git-Dch: Ignore

10 years agoupdate symbols files
David Kalnischkies [Fri, 30 May 2014 21:10:30 +0000 (23:10 +0200)] 
update symbols files

Git-Dch: ignore

10 years agoshow current/total testcase statistic also in concise mode
David Kalnischkies [Fri, 30 May 2014 17:18:25 +0000 (19:18 +0200)] 
show current/total testcase statistic also in concise mode

Git-Dch: Ignore

10 years agosupport parsing EDSP requests Architecture{,s} stanza
David Kalnischkies [Fri, 30 May 2014 16:01:47 +0000 (18:01 +0200)] 
support parsing EDSP requests Architecture{,s} stanza

Adds also a small testcase for EDSP

Git-Dch: Ignore

10 years agowithout a filename we can't stat pipes
David Kalnischkies [Fri, 30 May 2014 15:14:13 +0000 (17:14 +0200)] 
without a filename we can't stat pipes

EDSP code uses pipes opened via an FD as sources and later for those
files modification times and filesize are read - but never really used
again. The result we get from FileFd is probably wrong, but as we don't
use it anyway, we just don't fallback if we have nothing to fallback to

Git-Dch: Ignore

10 years agoprint error stack on failure of 'apt' solver
David Kalnischkies [Fri, 30 May 2014 14:52:19 +0000 (16:52 +0200)] 
print error stack on failure of 'apt' solver

Git-Dch: ignore

10 years agocheck exit status of external solvers
David Kalnischkies [Fri, 30 May 2014 13:58:30 +0000 (15:58 +0200)] 
check exit status of external solvers

Solvers are supposed to exit successfully even if they haven't found a
solution, but a solver which fails drastically (like e.g. segfaults)
should be detected and dealt with accordingly instead of ignored.

10 years agoif Resolver fails, do not continue even if not broken
David Kalnischkies [Fri, 30 May 2014 11:42:36 +0000 (13:42 +0200)] 
if Resolver fails, do not continue even if not broken

This can happen if the request is already a well-formed request all by
itself (e.g. the package has no dependencies), but the resolver found
a reason to not accept it as solution. Our edsp 'dump' solver e.g.
shouldn't be able to trigger install, which it does otherwise.

10 years agoMerge EDSP 0.5 w/ multi-arch support for external solvers
David Kalnischkies [Fri, 30 May 2014 11:13:03 +0000 (13:13 +0200)] 
Merge EDSP 0.5 w/ multi-arch support for external solvers

"I am going to merge it tomorrow…"

10 years agoMerge remote-tracking branch 'mvo/debian/sid' into debian/sid
Michael Vogt [Thu, 29 May 2014 10:23:51 +0000 (12:23 +0200)] 
Merge remote-tracking branch 'mvo/debian/sid' into debian/sid

10 years ago(try to) fix travis-ci build failures
David Kalnischkies [Mon, 12 May 2014 19:25:43 +0000 (21:25 +0200)] 
(try to) fix travis-ci build failures

dpkg on Ubuntu 12.04 does not seem to support parsing arch-specific
dependencies, so we try to detect if we face such a dpkg in the test.

In the other test the order depends on libdb, which changes per arch, so
we just run it through our sorting binary and be happy (hopefully).

Git-Dch: Ignore

10 years agoFix warning about uninitialized variable
Michael Vogt [Wed, 28 May 2014 08:00:52 +0000 (10:00 +0200)] 
Fix warning about uninitialized variable

Reported-By: clang++ -Werror
10 years agoFix uninitialized value
Michael Vogt [Wed, 28 May 2014 07:24:58 +0000 (09:24 +0200)] 
Fix uninitialized value

Reported-By: scan-build
10 years agouse free() instead of delete[] in debSrcRecordParser::~debSrcRecordParser
Michael Vogt [Tue, 27 May 2014 21:42:10 +0000 (23:42 +0200)] 
use free() instead of delete[] in debSrcRecordParser::~debSrcRecordParser

The Buffer was allocated using strndup() so we need to free it using
free() instead of delete[]

10 years agoDo not try to cast a pkgDepCache::Policy to a pkgCache
Michael Vogt [Tue, 27 May 2014 15:49:53 +0000 (17:49 +0200)] 
Do not try to cast a pkgDepCache::Policy to a pkgCache

Fix incorrect cast in pkgDepCache::Policy::GetCandidateVer()

Reported-By: clang -fsanitize=address -fno-omit-frame-pointer
10 years agouse free() instead of delete() when realloc is used
Michael Vogt [Tue, 27 May 2014 14:25:43 +0000 (16:25 +0200)] 
use free() instead of delete() when realloc is used

ContentsExtract::~ContentsExtract() needs to use free() because
Data got allocated via realloc()

Reported-By: clang -fsanitize=address -fno-omit-frame-pointer
10 years agoMerge remote-tracking branch 'upstream/debian/sid' into debian/sid
Michael Vogt [Tue, 27 May 2014 12:09:48 +0000 (14:09 +0200)] 
Merge remote-tracking branch 'upstream/debian/sid' into debian/sid

10 years agoupdate Doxyfile to 1.8.7 to silence deprecation warnings
David Kalnischkies [Thu, 22 May 2014 16:41:52 +0000 (18:41 +0200)] 
update Doxyfile to 1.8.7 to silence deprecation warnings

Git-Dch: ignore

10 years agofix screen width detection for apt/apt-get lists
Sebastian Schmidt [Sat, 17 May 2014 04:57:47 +0000 (06:57 +0200)] 
fix screen width detection for apt/apt-get lists

3163087b moved SigWinch(int) from apt-get.cc to private-output.cc
without moving #include <sys/ioctl.h>, making SigWinch a nop.

Closes: 748430, 747942
10 years agofix tight loop detection and temporary removes
David Kalnischkies [Sat, 17 May 2014 10:37:13 +0000 (12:37 +0200)] 
fix tight loop detection and temporary removes

As outlined in #748355 apt segfaulted if it encountered a loop between a
package pre-depending on a package conflicting with the previous as it
ended up in an endless loop trying to unpack 'the other package'.

In this specific case as an essential package is involved a lot of force
needs to be applied, but can also be caused by 'normal' tight loops and
highlights a problem in how we handle breaks which we want to avoid.

The fix comes in multiple entangled changes:
1. All Smart* calls are guarded with loop detection. Some already had it,
   some had parts of it, some did it incorrect, and some didn't even try.
2. temporary removes to avoid a loop (which is done if a loop is
   detected) prevent the unpack of this looping package (we tried to unpack
   it to avoid the conflict/breaks, but due to a loop we couldn't, so we
   remove/deconfigure it instead which means we can't unpack it now)
3. handle conflicts and breaks very similar instead of duplicating most
   of the code. The only remaining difference is, as it should:
   deconfigure is enough for breaks, for conflicts we need the big hammer

10 years agoconsistently fail if Smart* packagemanager actions fail
David Kalnischkies [Fri, 16 May 2014 15:28:11 +0000 (17:28 +0200)] 
consistently fail if Smart* packagemanager actions fail

These failure conditions come with an error message attached and the
conditions aren't workaroundable (otherwise this would have been done
instead of returning failure), so not erroring out here means that we
execute dpkg later on with a known not-working ordering adding insult
(our own error messages at the end) to injury (dpkg failure).

10 years agomark as Automatic/Downloadable pure as gcc suggests
David Kalnischkies [Thu, 8 May 2014 17:03:33 +0000 (19:03 +0200)] 
mark as Automatic/Downloadable pure as gcc suggests

Git-Dch: Ignore
Reported-By: gcc
10 years agomake test independent from the actual Install-Size
David Kalnischkies [Thu, 8 May 2014 16:52:51 +0000 (18:52 +0200)] 
make test independent from the actual Install-Size

Git-Dch: Ignore

10 years agoadd an additional test for arch specific conflicts
David Kalnischkies [Sat, 10 May 2014 12:21:10 +0000 (14:21 +0200)] 
add an additional test for arch specific conflicts

In bugreport #747261 I confirmed with this testcase that apt actually
supports the requested architecture-specific conflicts already since
2012 with commit cef094c2ec8214b2783a2ac3aa70cf835381eae1.

The old test only does simulations which are handy to check apt,
this one builds 'real' packages to see if dpkg agrees with us.

Git-Dch: Ignore

10 years agoNever parse Version/Architecture tags in a Translation-$lang file
Michael Vogt [Thu, 15 May 2014 12:37:33 +0000 (14:37 +0200)] 
Never parse Version/Architecture tags in a Translation-$lang file

Version/Architecture information in a Translation-$lang file is
not allowed, so don't try to parse it. This is a fix for a bugreport
where a Translation-en file contained the content of the regular
Packages file (probably due to local FS corruption). This lead to
strange error messages on file download.

Thanks to Thomas Reusch for the report.

10 years agoMerge remote-tracking branch 'mvo/feature/apt-ftparchive-srccache2' into debian/sid
Michael Vogt [Thu, 8 May 2014 09:51:09 +0000 (11:51 +0200)] 
Merge remote-tracking branch 'mvo/feature/apt-ftparchive-srccache2' into debian/sid

10 years agofix regression from commit 215b0faf
Michael Vogt [Thu, 8 May 2014 09:46:29 +0000 (11:46 +0200)] 
fix regression from commit 215b0faf

10 years agoshow progress in run-tests
Michael Vogt [Wed, 7 May 2014 19:03:45 +0000 (21:03 +0200)] 
show progress in run-tests

10 years agoMerge remote-tracking branch 'mvo/feature/apt-ftparchive-srccache2' into debian/sid
Michael Vogt [Wed, 7 May 2014 16:10:01 +0000 (18:10 +0200)] 
Merge remote-tracking branch 'mvo/feature/apt-ftparchive-srccache2' into debian/sid

10 years agoMerge remote-tracking branch 'donkult/debian/sid' into debian/sid
Michael Vogt [Wed, 7 May 2014 16:09:11 +0000 (18:09 +0200)] 
Merge remote-tracking branch 'donkult/debian/sid' into debian/sid

10 years agoprepare release 1.0.3 1.0.3
Michael Vogt [Mon, 5 May 2014 14:26:12 +0000 (16:26 +0200)] 
prepare release 1.0.3

10 years agoRevert "add bash completion for the "apt" command"
Michael Vogt [Mon, 5 May 2014 14:24:52 +0000 (16:24 +0200)] 
Revert "add bash completion for the "apt" command"

This reverts commit 697c9314c8ba24f3e393b5de11a3fad7adae4bfc.

Conflicts:
debian/rules

10 years agoMerge remote-tracking branch 'upstream/debian/sid' into debian/sid
Michael Vogt [Mon, 5 May 2014 14:03:03 +0000 (16:03 +0200)] 
Merge remote-tracking branch 'upstream/debian/sid' into debian/sid

10 years agoEDSP: add APT-Release field to Package stanzas
Stefano Zacchiroli [Thu, 5 Sep 2013 08:54:12 +0000 (10:54 +0200)] 
EDSP: add APT-Release field to Package stanzas

10 years agoEDSP: add Source field to Package stanzas
Stefano Zacchiroli [Thu, 5 Sep 2013 14:36:44 +0000 (16:36 +0200)] 
EDSP: add Source field to Package stanzas

10 years agoEDSP: add Architecture(s) multi-arch fields to the Request stanza
Stefano Zacchiroli [Wed, 4 Sep 2013 15:13:49 +0000 (17:13 +0200)] 
EDSP: add Architecture(s) multi-arch fields to the Request stanza

10 years agoEDSP: bump protocol version to 0.5
Stefano Zacchiroli [Wed, 4 Sep 2013 15:06:11 +0000 (17:06 +0200)] 
EDSP: bump protocol version to 0.5

10 years agoEDSP doc: fix typo in Request stanza description
Stefano Zacchiroli [Wed, 4 Sep 2013 15:04:09 +0000 (17:04 +0200)] 
EDSP doc: fix typo in Request stanza description

10 years agoOnly do openpty() if both stdin/stdout are terminals
Michael Vogt [Wed, 30 Apr 2014 15:04:29 +0000 (17:04 +0200)] 
Only do openpty() if both stdin/stdout are terminals

Closes: 746434
10 years agoinitialize Verify in second pkgAcqIndex constructor
David Kalnischkies [Mon, 28 Apr 2014 15:36:54 +0000 (17:36 +0200)] 
initialize Verify in second pkgAcqIndex constructor

gcc reports in testcase ./test-bug-596498-trusted-unsigned-repo:
apt-pkg/acquire-item.cc:1059:7: runtime error: load of value 234, which
is not a valid value for type 'bool'

This happens as the bool Verify is initialized only in one of the two
constructors of the pkgAcqIndex class. It isn't a problem through as the
verification controlled by this flag is optional and used to fail early
on garbage files (like network portal pages) instead of later on in the
hashsum verification or while parsing (the then untrusted) file.

Reported-By: gcc-4.9 -fsanitize=undefined
10 years agoFix missing ScreenWidth check in apt.cc
Michael Vogt [Mon, 28 Apr 2014 11:41:33 +0000 (13:41 +0200)] 
Fix missing ScreenWidth check in apt.cc

10 years agodebian/apt-doc.docs: remove README.MultiArch
Michael Vogt [Mon, 28 Apr 2014 10:57:01 +0000 (12:57 +0200)] 
debian/apt-doc.docs: remove README.MultiArch

10 years agol10n: vi.po: Update one new string
Trần Ngọc Quân [Mon, 28 Apr 2014 02:27:44 +0000 (09:27 +0700)] 
l10n: vi.po: Update one new string

Signed-off-by: Trần Ngọc Quân <vnwildman@gmail.com>
10 years agoallow vendors to install configuration files
David Kalnischkies [Sat, 26 Apr 2014 11:31:41 +0000 (13:31 +0200)] 
allow vendors to install configuration files

Vendors like ubuntu need to change some options, so giving them
a way to do this less painfully avoids reducing differences.

10 years agofix FileFd::Size bitswap on big-endian architectures
Adam Conrad [Sat, 26 Apr 2014 08:24:40 +0000 (10:24 +0200)] 
fix FileFd::Size bitswap on big-endian architectures

gzip only gives us 32bit of size, storing it in a 64bit container and
doing a 32bit flip on it has therefore unintended results.
So we just go with a exact size container and let the flipping be handled
by eglibc provided le32toh removing our #ifdef machinery.

Closes: 745866
10 years agoenforce LFS for partial files in https range requests
David Kalnischkies [Thu, 24 Apr 2014 08:49:41 +0000 (10:49 +0200)] 
enforce LFS for partial files in https range requests

10 years agoremove unused duplicated acqprogress.{cc,h} files
David Kalnischkies [Thu, 24 Apr 2014 08:33:57 +0000 (10:33 +0200)] 
remove unused duplicated acqprogress.{cc,h} files

Those files are in apt-private and used from there.

Git-Dch: Ignore

10 years agobuild http request in a stringstream
David Kalnischkies [Thu, 24 Apr 2014 08:20:45 +0000 (10:20 +0200)] 
build http request in a stringstream

beside reducing code a bit, it avoids oddball problems while building
the string and doesn't trigger static analyse warnings.

10 years agoremove outdated README.MultiArch
David Kalnischkies [Tue, 22 Apr 2014 17:02:53 +0000 (19:02 +0200)] 
remove outdated README.MultiArch

Debian wheezy shipped MultiArch to the masses and the predictions
remained true in sofar as little changes in apt itself and many
other frontends were needed compared to the fallout if done differently.

The info included is this file is therefore no longer current and adds
no useful information anymore, so we can drop it for good.

10 years agoadd a README for vendor information
David Kalnischkies [Tue, 22 Apr 2014 16:59:22 +0000 (18:59 +0200)] 
add a README for vendor information

10 years agoadd vendor information for BlankOn
Mahyuddin Ramli [Tue, 22 Apr 2014 15:32:37 +0000 (17:32 +0200)] 
add vendor information for BlankOn

Modified by commiter to not publicily export the codename (as the
manpages do not use it that way) and removing the included additional
derives logic as it was not working (the link always exists at that
point) and isn't needed as we do the special casing for debian mainly
because it would shallow all distributions otherwise.
(similar, but not that strong for ubuntu)

Closes: 743595
10 years agohandle pkgnames shorter than modifiers
David Kalnischkies [Mon, 21 Apr 2014 11:26:55 +0000 (13:26 +0200)] 
handle pkgnames shorter than modifiers

The bugreport highlights the problem with an empty package name. We fix
this by 'ignoring' these so that it behaves just like "apt-get install".
The deeper problem is that modifier strings can be longer than a package
name in which case the comparison doesn't make sense, so don't compare
then. Was not noticed so far as all modifiers are of length 1, so the
only package name shorter than this is in fact the empty package name.

Closes: 744940
10 years agoproperly undo CD-ROM mount in all error cases
John Ogness [Mon, 21 Apr 2014 09:54:34 +0000 (11:54 +0200)] 
properly undo CD-ROM mount in all error cases

In bug #740673 various issues in the CD-ROM handling code were
identified, while most the issues ended up being fixed in another way,
the unmounting of the CD-ROM in error cases was not tackled so far.

(The patch was modified by the commiter to apply)

10 years agoprovide support for debian/apt.conf.$(lsb_release -i -s) vendor specific config files
Michael Vogt [Fri, 25 Apr 2014 16:18:28 +0000 (18:18 +0200)] 
provide support for debian/apt.conf.$(lsb_release -i -s) vendor specific config files

10 years agoadd bash completion for the "apt" command
Michael Vogt [Fri, 25 Apr 2014 13:16:46 +0000 (15:16 +0200)] 
add bash completion for the "apt" command

10 years agoreduce delta from ubuntu
Michael Vogt [Fri, 25 Apr 2014 12:41:35 +0000 (14:41 +0200)] 
reduce delta from ubuntu

10 years agoprepare release 1.0.2 1.0.2
Michael Vogt [Fri, 25 Apr 2014 11:16:18 +0000 (13:16 +0200)] 
prepare release 1.0.2

10 years agoMerge remote-tracking branch 'upstream/debian/sid' into debian/sid
Michael Vogt [Tue, 22 Apr 2014 14:21:05 +0000 (16:21 +0200)] 
Merge remote-tracking branch 'upstream/debian/sid' into debian/sid

10 years agoConsistently use Dpkg::Progress* in documentation
James McCoy [Mon, 21 Apr 2014 20:35:28 +0000 (16:35 -0400)] 
Consistently use Dpkg::Progress* in documentation

Closes: 745452
10 years agoUpdated Thai program translation (closes: #745120)
Theppitak Karoonboonyanan [Tue, 22 Apr 2014 13:21:38 +0000 (15:21 +0200)] 
Updated Thai program translation (closes: #745120)

10 years agoMerge remote-tracking branch 'donkult/debian/sid' into debian/sid
Michael Vogt [Tue, 22 Apr 2014 13:19:17 +0000 (15:19 +0200)] 
Merge remote-tracking branch 'donkult/debian/sid' into debian/sid

10 years agoFix option name DPkg::Progress-Fancy in apt.8 manpage (LP: #1310506)
Michael Vogt [Tue, 22 Apr 2014 13:14:03 +0000 (15:14 +0200)] 
Fix option name DPkg::Progress-Fancy in apt.8 manpage (LP: #1310506)

10 years agoapt-private/acqprogress.cc: fix output when ctrl-c is hit during apt update (LP:...
Michael Vogt [Tue, 22 Apr 2014 13:10:19 +0000 (15:10 +0200)] 
apt-private/acqprogress.cc: fix output when ctrl-c is hit during apt update (LP: #1310548, closes: #744297)

10 years agosupport dist-upgrade options in full-upgrade
David Kalnischkies [Wed, 16 Apr 2014 20:47:25 +0000 (22:47 +0200)] 
support dist-upgrade options in full-upgrade

dist-upgrade is supposed to be an alias for full-upgrade in apt, but
dist-upgrade was the only command recognized of the two in the option
and flags recognition code.

10 years agoextract travis installs from build-depends automatically
David Kalnischkies [Wed, 16 Apr 2014 20:06:22 +0000 (22:06 +0200)] 
extract travis installs from build-depends automatically

I forgot to add libgtest-dev to the list of packages to install on
travis, so this slightly hacky oneliner might prevent us from having
the same problem again if we happen to change dependencies again.

Git-Dch: Ignore

10 years agouse Google C++ Testing Framework for libapt tests
David Kalnischkies [Wed, 16 Apr 2014 15:09:37 +0000 (17:09 +0200)] 
use Google C++ Testing Framework for libapt tests

My commit 45df0ad2 from 26. Nov 2009 had a little remark:
"The commit also includes a very very simple testapp."
This was never intended to be permanent, but as usually…

The commit adds the needed make magic to compile gtest statically
as it is required and links it against a small runner. All previous
testcase binaries are reimplemented in gtest and combined in this
runner. While most code is a 1:1 translation some had to be rewritten
like compareversion_test.cc, but the coverage remains the same.

10 years agoclear HitEof flag in FileFd::Seek
David Kalnischkies [Tue, 15 Apr 2014 08:21:52 +0000 (10:21 +0200)] 
clear HitEof flag in FileFd::Seek

fseek and co do this to their eof-flags and it is more logic this way as
we will usually seek away from the end (e.g. to re-read the file).

The commit also improves the testcase further and adds a test for the
binary compressor codepath (as gz, bzip2 and xz are handled by
libraries) via the use of 'rev' as a 'compressor'.

10 years agoapt-private/acqprogress.cc: reset color in apt update
Michael Vogt [Wed, 16 Apr 2014 11:15:55 +0000 (13:15 +0200)] 
apt-private/acqprogress.cc: reset color in apt update

10 years agoforce fancy progressbar redraw on window size change
David Kalnischkies [Mon, 14 Apr 2014 15:12:09 +0000 (17:12 +0200)] 
force fancy progressbar redraw on window size change

We always reacted on the size change, but the bar is only redraw if the
precentage changes, which can take quiet a while in big upgrades, so
with a bit of refactoring we can now call for a redraw immediate to fix
this.

This refactor also helps in avoiding obscure pitfalls clangs static
analyser was complaining about (namely failure of ioctl resulting in
garbage values in the struct).

10 years agocompile with absolute paths to allow lcov use
David Kalnischkies [Sun, 13 Apr 2014 19:40:35 +0000 (21:40 +0200)] 
compile with absolute paths to allow lcov use

Instructing gcc (or clang) to prepare for capturing coverage data is
easy: Just build with: CXXFLAGS=--coverage

The hard part is that our buildsystem uses relative paths and so
confuses the hell out of lcov as it assumes this way that all our *.cc
files are in the same directory… by changing to absolute paths in the
compile rules we solve this problem.

Still not perfect as it refers to build/include files for most headers
and our forking/threading code isn't properly captured, but good enough
to see red reports for now:

CXXFLAGS=--coverage make
make test
./test/integration/run-tests -q
lcov --no-external --directory . --capture --output-file apt.info
genhtml --output-directory ./coverage/ apt.info

Git-Dch: Ignore

10 years agol10n: vi.po (624t): Update translation
Trần Ngọc Quân [Fri, 11 Apr 2014 18:19:40 +0000 (20:19 +0200)] 
l10n: vi.po (624t): Update translation

Signed-off-by: Michael Vogt <mvo@debian.org>
10 years agoMerge remote-tracking branch 'mvo/feature/helpful-apt-list' into debian/sid
Michael Vogt [Fri, 11 Apr 2014 12:40:08 +0000 (14:40 +0200)] 
Merge remote-tracking branch 'mvo/feature/helpful-apt-list' into debian/sid

Conflicts:
test/integration/test-apt-cli-list

10 years agoMerge remote-tracking branch 'mvo/bugfix/apt-list-rc-pkgs' into debian/sid
Michael Vogt [Fri, 11 Apr 2014 12:39:28 +0000 (14:39 +0200)] 
Merge remote-tracking branch 'mvo/bugfix/apt-list-rc-pkgs' into debian/sid

10 years agoconsider priorities only for downloadable pkgs in resolver
David Kalnischkies [Fri, 11 Apr 2014 11:33:31 +0000 (13:33 +0200)] 
consider priorities only for downloadable pkgs in resolver

A package which can't be downloaded anymore is very likely dropped from
a release and can therefore no longer be 'standard' (or similar). We
therefore do not grant points for them anymore and demote them to
prio:extra instead which helps other packages breaking them away even if
they have a lower priority.

The testcase was initially created by Michael Vogt and just amended.

10 years agoextend FileFd test behond basic permission tests
David Kalnischkies [Fri, 11 Apr 2014 09:29:31 +0000 (11:29 +0200)] 
extend FileFd test behond basic permission tests

We now do Open, Write and Read (the later multiple ways) for each
permission and each compressor we have configured to cover more cases
and especially ensure that compressors do not change our premissions.

This test is also to be credited for discovering the skippos-fix.

Git-Dch: Ignore

10 years agodeal with umask only if we really need to for mkstemp
David Kalnischkies [Fri, 11 Apr 2014 09:22:10 +0000 (11:22 +0200)] 
deal with umask only if we really need to for mkstemp

As the comment actually says: open() does the umask dance by itself, so
we don't need to do it for it. We have to do it after mkstemp in Atomic
though, so move it into the if.

Also removes the "micro-optimisation" "FilePermissions == 600" as it
doesn't trigger at the moment anyway as 600 != 0600.

10 years agodon't double-count seeks in FileFd::Skip for bzip/xz
David Kalnischkies [Fri, 11 Apr 2014 09:18:58 +0000 (11:18 +0200)] 
don't double-count seeks in FileFd::Skip for bzip/xz

FileFd::Read already deals with the increase of the skipposition so that
we as the caller in FileFd::Skip really shouldn't increase it, too.

10 years agodo not create an (additional) empty compressor
David Kalnischkies [Fri, 11 Apr 2014 09:16:04 +0000 (11:16 +0200)] 
do not create an (additional) empty compressor

FileFd code knows how to deal with such a compressor, so it isn't a
problem, but it is absolutely not needed as we already have an
(matching) identity compressor with '.' earlier in the list.

Git-Dch: Ignore

10 years agobe able to run abicheck from any directory
David Kalnischkies [Sun, 23 Mar 2014 13:09:12 +0000 (14:09 +0100)] 
be able to run abicheck from any directory

Git-Dch: Ignore

10 years agouse wildcard to get files in our library makefiles
David Kalnischkies [Sun, 23 Mar 2014 12:17:24 +0000 (13:17 +0100)] 
use wildcard to get files in our library makefiles

The explicit listing is a pain every time you want to add a file to the
list and serves no propose as we list all files there anyway, so this is
not only easier but also documents this fact.

Git-Dch: Ignore

10 years agofix test-failure in adt
Michael Vogt [Thu, 10 Apr 2014 11:53:28 +0000 (13:53 +0200)] 
fix test-failure in adt

10 years agoprepare 1.0.1 release 1.0.1
Michael Vogt [Thu, 10 Apr 2014 07:49:25 +0000 (09:49 +0200)] 
prepare 1.0.1 release

10 years agoapt: Minor typo in 'apt' man page (closes: #743657)
Josef Vitu [Thu, 10 Apr 2014 07:33:14 +0000 (09:33 +0200)] 
apt: Minor typo in 'apt' man page (closes: #743657)

10 years agoMerge remote-tracking branch 'mvo/bugfix/lp1304657-perms' into debian/sid
Michael Vogt [Thu, 10 Apr 2014 07:15:03 +0000 (09:15 +0200)] 
Merge remote-tracking branch 'mvo/bugfix/lp1304657-perms' into debian/sid

10 years agoMerge remote-tracking branch 'mvo/bugfix/apt-search-case' into debian/sid
Michael Vogt [Thu, 10 Apr 2014 07:14:57 +0000 (09:14 +0200)] 
Merge remote-tracking branch 'mvo/bugfix/apt-search-case' into debian/sid

10 years agoimprove umask/fchmod code readability
Michael Vogt [Thu, 10 Apr 2014 07:11:57 +0000 (09:11 +0200)] 
improve umask/fchmod code readability

10 years agoRename FileFd::Open() Perms to AccessMode
Michael Vogt [Wed, 9 Apr 2014 08:24:47 +0000 (10:24 +0200)] 
Rename FileFd::Open() Perms to AccessMode

Bug lp:#1304657 was caused by confusion around the name Perms.
The new name AccessMode should make it clear that its not the
literal file permissions but instead the AccessMode passed to
open() (i.e. the umask needs to be applied)

10 years agoFix insecure file permissions when using FileFd with OpenMode::Atomic
Michael Vogt [Wed, 9 Apr 2014 08:12:10 +0000 (10:12 +0200)] 
Fix insecure file permissions when using FileFd with OpenMode::Atomic

Commit 7335eebea6dd43581d4650a8818b06383ab89901 introduced a bug
that caused FileFd to create insecure permissions when FileFd::Atomic
is used. This commit fixes the permissions and adds a test.

The bug is most likely caused by the confusing "Perm" parameter
that is passed to Open() - its not the file permissions but intead
the "mode" part of open/creat.

10 years agoNotice the user about "apt list -a" when only a single hit if found
Michael Vogt [Wed, 9 Apr 2014 14:23:14 +0000 (16:23 +0200)] 
Notice the user about "apt list -a" when only a single hit if found

If the user is using "apt list pattern" and there is only a single
hit, notice about "--all-versions" as this is what the user may
be interessted in

10 years agoFix possible race when stunnel/aptwebserver create their PID files
Michael Vogt [Wed, 9 Apr 2014 13:28:00 +0000 (15:28 +0200)] 
Fix possible race when stunnel/aptwebserver create their PID files

This patch should fix spurious test failures in jenkins or travis
that are caused by a race condition in the {stunnel,aptwebserver}.pid
file creation

10 years agofix apt list output for pkgs in dpkg ^rc state
Michael Vogt [Tue, 8 Apr 2014 07:04:15 +0000 (09:04 +0200)] 
fix apt list output for pkgs in dpkg ^rc state

Packages in the "deinstall ok config-file" have no candidate or
instaleld version. So they must be special cased in the apt list
generation.

10 years agoMerge branch 'bugfix/apt-search-case' into debian/sid
Michael Vogt [Tue, 8 Apr 2014 06:31:14 +0000 (08:31 +0200)] 
Merge branch 'bugfix/apt-search-case' into debian/sid

10 years agomake apt search case-insensitive by default
Michael Vogt [Mon, 7 Apr 2014 14:40:12 +0000 (16:40 +0200)] 
make apt search case-insensitive by default

10 years agoAdd versioned openjdk-6-jdk breaks
Julian Andres Klode [Mon, 7 Apr 2014 09:11:06 +0000 (11:11 +0200)] 
Add versioned openjdk-6-jdk breaks

This helps if people did unclean upgrades from squeeze, namely to
jessie directly.

10 years agoVersion the Breaks/Replaces for sun-java{5,6}-jdk (LP: #1302736)
Julian Andres Klode [Mon, 7 Apr 2014 09:01:07 +0000 (11:01 +0200)] 
Version the Breaks/Replaces for sun-java{5,6}-jdk (LP: #1302736)

This is a *hack* to work around unofficial packages for Java 7
and 8 that wrongly provide the Java 5 and 6 packages.

Closes: #743616
10 years agoensure "--db" also works with the new srcpkgdb
Michael Vogt [Mon, 7 Apr 2014 07:41:20 +0000 (09:41 +0200)] 
ensure "--db" also works with the new srcpkgdb

10 years agorefactor _InitQuery()
Michael Vogt [Fri, 4 Apr 2014 15:21:40 +0000 (17:21 +0200)] 
refactor _InitQuery()