summaryrefslogtreecommitdiff
path: root/apt-pkg/algorithms.cc
AgeCommit message (Collapse)Author
2011-05-02doesn't execute autoremove marker setting if an external solver is calledDavid Kalnischkies
and instead rely on the Autoremove tagging to show us what could be done. (apt-internal-solver doesn't support this currently as it doesn't load the auto-information into the cache)
2011-05-02refactor: move solver execution into his own EDSP methodDavid Kalnischkies
2011-05-02handle Dir::Bin::Solvers as a list of directories and find theDavid Kalnischkies
solver in this list of directories
2011-04-16cppcheck: (style) Variable 'State' is assigned a value that is never usedDavid Kalnischkies
2011-04-02send the scenario through a pipe to the solver and get the solution backDavid Kalnischkies
The solution is NOT interpreted so far.
2011-04-01Read and apply install/remove requests correctlyDavid Kalnischkies
2011-03-31rename edspwriter to straight edsp in toplevel as it does more thanDavid Kalnischkies
just writing stuff… it also reads and can work for both: - APT talking to an external solver - an external solver (understanding EDSP) talking to APT
2011-03-31rename the 'universe' to 'scenario' to reflect the naming in the draftDavid Kalnischkies
2011-03-30be able to write solutions, tooDavid Kalnischkies
2011-03-30add a first round of stuff needed for talking between APT and solversDavid Kalnischkies
based on a very early draft for EDSP by Stefano APT can now write a scenario as well as load most stuff from it.
2011-03-08Remove the "pseudopackage" handling of Architecture: all packages forDavid Kalnischkies
Multi-Arch; instead, Arch: all packages only satisfy dependencies for the native arch, except where the Arch: all package is declared Multi-Arch: foreign. (Closes: #613584) This has the sideeffect that arch:all packages internally show up as coming from the native arch - so packages with the architecture "all" doesn't exist any longer in the pkgcache
2011-02-10merged from lp:~mvo/apt/mvoMichael Vogt
2011-02-10* merged lp:~evfool/apt/fix641673:Michael Vogt
* merged lp:~evfool/apt/fix418552: - Grammar fix for bug LP: #418552, thanks to Robert Roth
2011-02-07* apt-pkg/algorithms.cc:David Kalnischkies
- mark pseudo packages of installed all packages as configured in the simulation as we don't call configure for these packages
2010-11-29* apt-pkg/algorithms.cc:David Kalnischkies
- mark all installed packages first without auto installation in a dist-upgrade to prefer upgrading packages instead of installing new packages in versioned or-groups (Closes: #605394)
2010-11-10* apt-pkg/algorithms.cc:David Kalnischkies
- if the package was explicitly marked as ToRemove don't consider it as a candidate for FixByInstall
2010-09-15merged lp:~mvo/apt/conflicts-on-virtuals to better deal withMichael Vogt
conflicts/breaks against virtual packages (LP: #614993)
2010-09-03remove duplicated test, move test for conflicts higher up, better debug outputMichael Vogt
2010-08-31apt-pkg/algorithms.cc: show debug output only if debug option is givenMichael Vogt
2010-08-31apt-pkg/algorithms.cc: if a package was added to the kill list via a ↵Michael Vogt
DpkgBreaks, make sure to break in the loop (just like for a conflict)
2010-08-31apt-pkg/algorithms.cc: in the case of conflicts/breaks against a Provide ↵Michael Vogt
Start.AllTarget() returns all non-canidates as well, we want to filter them out
2010-08-14do not unconditional print the new debugoutput for "FixByInstall"David Kalnischkies
2010-08-13* apt-pkg/algorithms.cc:David Kalnischkies
- let the problem resolver install packages to fix or-groups as a needed remove nuked another or-member (helps for #591882) The resolver sometimes need to break or-groups because he need to remove the previous satisfier as a result of a breaks/conflicts. In such broken or-groups he now tries to install a non-installed non-deleted member of the or-group to statisfy it again. This move is dangerous as it introduces the possibility of a bunch of "new" broken breaks/conflicts on the new statisfier or its dependencies and we don't know about them beforehand. We also don't have the infrastructure to undo the decision later, so we just do it and pray! For now a test option is provided to disable this behavior on the fly to test if the situation is worse: pkgProblemResolver::FixByInstall
2010-08-10* apt-pkg/algorithms.cc:David Kalnischkies
- change the debug outputs to display also arch of the package and version dependencies information
2010-06-09* merge the remaining Ubuntu change:Michael Vogt
- on gpg verification failure warn and restore the last known good state - on failure display the IP of the server (useful for servers that use round robin DNS) - support Original-Maintainer in RewritePackageOrder - enable cdrom autodetection via libudev by default - show messsage about Vcs in use when apt-get source is run for packages maintained in a Vcs - better support transitional packages with mark auto-installed. when the transitional package is in "oldlibs" the new package is not marked auto installed (same is true for section metapackages) - provide new "deb mirror://archive.foo/mirrors.list sid main" method expects a list of mirrors (generated on the server e.g. via geoip) and will use that, including cycle on failure - write apport crash file on package failure (disabled by default on debian until apport is available) - support mirror failure reporting (disabled by default on debian)
2010-04-07* apt-pkg/contrib/fileutl.cc:David Kalnischkies
- add a parent-guarded "mkdir -p" as CreateDirectory() * apt-pkg/acquire.{cc,h}: - add a delayed constructor with Setup() for success reporting - check for and create directories in Setup if needed instead of error out unfriendly in the Constructor (Closes: #523920, #525783) - optional handle a lock file in Setup() * cmdline/apt-get.cc: - remove the lock file handling and let Acquire take care of it instead
2010-03-01fix compiler error if compiled with -std=c++0xDavid Kalnischkies
2010-02-12Add yet another pseudo package which isn't as pseudo as the others:David Kalnischkies
Arch all packages are now represented by arch depending packages which all depend on a package with the same name and the special arch "all". This packages has NO dependencies, but beside this the same information. It is the only package which has a size, the arch depending ones all have a zero size. While the arch depending pseudo packages are used for dependency resolution the arch "all" package is used for downloading and ordering of the package.
2009-07-30merged from debianMichael Vogt
2009-07-21The 'not dead yet' releaseMichael Vogt
* add hook for MarkInstall and MarkDelete (closes: #470035) * add the various foldmarkers in apt-pkg & cmdline (no code change) * versions with a pin of -1 shouldn't be a candidate (Closes: #355237) * prefer mmap as memory allocator in MMap instead of a static char array which can (at least in theory) grow dynamic * eliminate (hopefully all) segfaults in pkgcachegen.cc and mmap.cc which can arise if cache doesn't fit into the mmap (Closes: #535218) * display warnings instead of errors if the parts dirs doesn't exist * honor the dpkg hold state in new Marker hooks (closes: #64141)
2009-07-21* apt-pkg/acquire.cc:Michael Vogt
- make the max pipeline depth of the acquire queue configurable via Acquire::Max-Pipeline-Depth * apt-pkg/deb/dpkgpm.cc: - add Dpkg::UseIoNice boolean option to run dpkg with ionice -c3 (off by default) - send "dpkg-exec" message on the status fd when dpkg is run - provide DPkg::Chroot-Directory config option (useful for testing) - fix potential hang when in a backgroud process group * apt-pkg/algorithms.cc: - consider recommends when making the scores for the problem resolver * apt-pkg/acquire-worker.cc: - show error details of failed methods * apt-pkg/contrib/fileutl.cc: - if a process aborts with signal, show signal number * methods/http.cc: - ignore SIGPIPE, we deal with EPIPE from write in HttpMethod::ServerDie() (LP: #385144) * apt-pkg/indexcopy.cc: - support having CDs with no Packages file (just a Packages.gz) by not forcing a verification on non-existing files (LP: #255545) - remove the gettext from a string that consists entirely of variables (LP: #56792) * apt-pkg/cacheiterators.h: - add missing checks for Owner == 0 in end() * apt-pkg/indexrecords.cc: - fix some i18n issues * apt-pkg/contrib/strutl.h: - add new strprintf() function to make i18n strings easier - fix compiler warning * apt-pkg/deb/debsystem.cc: - make strings i18n able * fix problematic use of tolower() when calculating the version hash by using locale independant tolower_ascii() function. Thanks to M. Vefa Bicakci (LP: #80248) * build fixes for g++-4.4 * cmdline/apt-mark: - add "showauto" option to show automatically installed packages * document --install-recommends and --no-install-recommends (thanks to Dereck Wonnacott, LP: #126180) * Updated cron script to support backups by hardlinks and verbose levels. All features turned off by default. * Added more error handlings. Closes: #438803, #462734, #454989, * Refactored condition structure to make download and upgrade performed if only previous steps succeeded. Closes: #341970 * Documented all cron script related configuration items in configure-index. * apt-ftparchive might write corrupt Release files (LP: #46439) * Apply --important option to apt-cache depends (LP: #16947)
2009-06-30* add hook for MarkInstall and MarkDelete (closes: #470035)David Kalnischkies
* honor the dpkg hold state in new Marker hooks (closes: #64141) Combine the proposed AutoInstOk and IsAutoInstallOk to more general hooks for MarkInstall (and another one for MarkDelete) with the same parameters as the call these hooks should check.
2009-06-30add the various foldmarkers in apt-pkg & cmdline (no code change)David Kalnischkies
2009-06-30merge the AutoInstOk patch from debian-experimentalMichael Vogt
2009-06-29* support IsAutoInstallOk in the resolver tooMichael Vogt
* honor the dpkg hold state in IsAutoInstallOk (closes: #64141)
2009-06-29* add Debug::pkgDepCache::Marker with more detailed debug output Michael Vogt
* add Debug::pkgProblemResolver::ShowScores and make the scores adjustable
2009-06-23* apt-pkg/algorithms.cc:Michael Vogt
- consider recommends when making the scores for the problem resolver
2008-11-24merged from debian-sidMichael Vogt
2008-10-28* apt-pkg/acquire-item.cc:Michael Vogt
- Added fallback to uncompressed 'Packages' if neither 'bz2' nor 'gz' available. (Closes: #409284) * apt-pkg/algorithm.cc: - Strip username and password from source URL in error message. (Closes: #425150)
2008-10-26Forgot to remove regex.h include in previous commit. Now doneEugene V. Lyubimkin
2008-10-26Switched from regexp to using existing URI class to strip user/password.Eugene V. Lyubimkin
2008-10-25Corrected determining if regexp matched.Eugene V. Lyubimkin
2008-10-25Strip user/password from URL in error message.Eugene V. Lyubimkin
2008-10-03Fix some typos in docs and translations (thanks to timeless, closes: 368665)Luca Bruno
2008-10-01* apt-pkg/packagemanager.cc, apt-pkg/deb/dpkgpm.cc:Michael Vogt
- move the state file writting into the Go() implementation of dpkgpm (closes: #498799) * apt-pkg/algorithms.cc: - fix simulation performance drop (thanks to Ferenc Wagner for reporting the issue)
2008-02-21* apt-pkg/algorithms.cc: Michael Vogt
- add APT::Update::Post-Invoke-Success script slot (LP: #188127)
2008-02-06merged from mvoMichael Vogt
2008-02-06 - Make the breaks handling use the kill list. This means, that aMichael Vogt
Breaks: Pkg (<< version) may put Pkg onto the remove list.
2008-02-01* apt-pkg/algorithms.cc: Michael Vogt
- add APT::Update::Post-Invoke-Success script slot
2008-01-16* Merged apt-authentication-reliabilty branch. This meansMichael Vogt
that apt will refuse to update and use the old lists if the authentication of a repository that used to be authenticated fails. See https://wiki.ubuntu.com/AptAuthenticationReliability for more details.