Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | depgraph: avoid atom hash collisions in dep_check | Zac Medico | 2011-02-13 | 1 | -7/+8 |
| | | | | | | | | Atoms are stored in the graph as (atom, id(atom)) tuples since each atom is considered to be a unique entity. For example, atoms that appear identical may behave differently in USE matching, depending on their unevaluated form. Also, specially generated virtual atoms may appear identical while having different _orig_atom attributes. | ||||
* | depgraph: return virtual edges from select_atoms | Zac Medico | 2011-02-13 | 1 | -36/+42 |
| | |||||
* | emerge.1: cross-ref --newuse and --changed-use | Zac Medico | 2011-02-13 | 1 | -1/+4 |
| | | | | This will fix bug #354437. | ||||
* | depgraph: don't distort edges for virtuals | Zac Medico | 2011-02-13 | 1 | -9/+19 |
| | | | | | | | | In _add_pkg_dep_string(), _add_pkg() was called with generated Dependency instances that created fake/distorted edges in the graph for new-style virtuals (similar to those that were fixed in commit 5b16e0d800f34dc90cdd8c505172ad16b97f0af4). Now there won't be any such distorted edges. | ||||
* | depgraph: fix distorted display of virtuals | Zac Medico | 2011-02-13 | 1 | -0/+13 |
| | | | | | | | | As mentioned in bug #353933, comment #4, special new-style virtual atoms generated/distorted by _expand_new_virtuals() can prevent extract_affecting_use() from working properly. This is fixed by saving the original atoms so that the depgraph can map them back into place at the appropriate step in dependency evaluation. | ||||
* | depgraph: avoid reinstall of identical binpkg | Zac Medico | 2011-02-13 | 1 | -0/+15 |
| | | | | This will fix bug #354441. | ||||
* | emerge: support ../ path arguments | Zac Medico | 2011-02-08 | 1 | -1/+2 |
| | | | | This will fix bug #354141. | ||||
* | get_dep_chain: add "set" node typev2.1.9.36 | Zac Medico | 2011-02-04 | 1 | -2/+5 |
| | |||||
* | depgraph: make downgrades trigger complete graph | Zac Medico | 2011-02-04 | 1 | -2/+21 |
| | | | | | It's common for downgrades to break dependencies, as in bug #353613, so automatically trigger complete graph mode when a downgrade is found. | ||||
* | depgraph: don't drop deps in _add_dep | Zac Medico | 2011-02-04 | 1 | -29/+0 |
| | | | | | | This will fix bug #353613. We remove a hunk of code that was meant to limit recursion, but is obsolete an buggy. The recursion limiting is now handled by _add_pkg and the _ignored_deps stack. | ||||
* | REQUIRED_USE: display unsatisfied part | Zac Medico | 2011-02-03 | 1 | -1/+14 |
| | | | | This will fix bug #353234. | ||||
* | action_uninstall: allow repo with wildcard atom | Zac Medico | 2011-02-03 | 1 | -1/+1 |
| | |||||
* | depgraph: use catsplit where appropriate | Zac Medico | 2011-02-02 | 1 | -2/+2 |
| | |||||
* | depgraph: remove obsolete gpg exception handlers | Zac Medico | 2011-02-02 | 1 | -14/+0 |
| | |||||
* | depgraph: skip backtrack for fatal errors | Zac Medico | 2011-02-02 | 1 | -0/+10 |
| | |||||
* | depgraph: rm obsolete AmbguousPackageName handler | Zac Medico | 2011-02-02 | 1 | -21/+2 |
| | |||||
* | REQUIRED_USE: no autounmask, so skip backtracking | Zac Medico | 2011-02-02 | 1 | -3/+8 |
| | |||||
* | REQUIRED_USE: handle unsatisfied like normal dep | Zac Medico | 2011-02-02 | 1 | -17/+6 |
| | |||||
* | Fix grammar in a comment. | Zac Medico | 2011-02-02 | 1 | -1/+1 |
| | |||||
* | REQUIRED_USE: add blank line after display | Zac Medico | 2011-02-02 | 1 | -0/+1 |
| | |||||
* | Make || choices independent of REQUIRED_USE. | Zac Medico | 2011-02-02 | 1 | -36/+35 |
| | |||||
* | _get_dep_chain: allow SetArg for first node | Zac Medico | 2011-02-02 | 1 | -14/+15 |
| | |||||
* | Show "missed update" message for REQUIRED_USE. | Zac Medico | 2011-02-02 | 1 | -7/+32 |
| | | | | This will fix bug #353443. | ||||
* | EbuildFetcher: use default SIGINT/TERM handlers | Zac Medico | 2011-02-02 | 1 | -0/+6 |
| | | | | This seems to fix bug #353239. | ||||
* | depgraph: display 'searching for similar names' | Zac Medico | 2011-02-02 | 1 | -3/+11 |
| | | | | It can take few seconds to search, so display a status message. | ||||
* | Print near matches for cmd line atoms without matching ebuild | Sebastian Luther | 2011-02-02 | 1 | -0/+33 |
| | |||||
* | EbuildBuild: spawn nofetch for fetchonly | Zac Medico | 2011-02-02 | 1 | -5/+12 |
| | |||||
* | Allow relative paths (starting with ./) on the cmd line | Sebastian Luther | 2011-02-02 | 1 | -0/+7 |
| | | | | bug 352208 | ||||
* | slot collision suggestions: Avoid showing redundant suggestions. | Sebastian Luther | 2011-02-02 | 1 | -14/+71 |
| | | | | http://forums.gentoo.org/viewtopic-t-862257.html?sid=9ba8646a15044fe024a41154df351c6a | ||||
* | depgraph: fix get_dep_chain unicode handling | Zac Medico | 2011-02-02 | 1 | -7/+8 |
| | |||||
* | autounmask USE change comment: ensure that the first parent requires the change | Sebastian Luther | 2011-02-02 | 1 | -16/+29 |
| | | | | Fixes bug 350515. | ||||
* | Show use conditionals in old style dep chains | Sebastian Luther | 2011-02-02 | 1 | -126/+121 |
| | |||||
* | Update Portage to support multiple binhosts. | David James | 2011-02-02 | 1 | -2/+2 |
| | | | | | | | | | | | BUG=chromium-os:11295 TEST=Build x86-mario with multiple binhosts (board binhost, preflight binhost) and verify that binaries are picked up from both binhosts. Also verify that if one of the binhosts fails, the other binhost is used correctly. Review URL: http://codereview.chromium.org/6329022 Change-Id: I095f28c82e3effbc25105af6ac89e42c335e6381 | ||||
* | Wildcard support for -c/-C | Sebastian Luther | 2011-02-02 | 1 | -1/+24 |
| | |||||
* | fdopen: read from processes in unbuffered modev2.1.9.35 | Zac Medico | 2011-01-21 | 1 | -2/+2 |
| | | | | | It doesn't make sense to use buffers in cases like this, and is potentially error prone (as in bug #287648 / python issue 5380). | ||||
* | SpawnProcess: read proc with unbuffered fdopen | Zac Medico | 2011-01-21 | 1 | -2/+4 |
| | | | | | This enables pty support in python3, by using unbuffered fdopen to avoid http://bugs.python.org/issue5380. | ||||
* | BinpkgFetcher: ust pty like EbuildFetcher | Zac Medico | 2011-01-19 | 1 | -1/+16 |
| | |||||
* | depgraph: avoid duplicate ebuild visibility check | Zac Medico | 2011-01-18 | 1 | -1/+1 |
| | |||||
* | depgraph: avoid updates more | Zac Medico | 2011-01-18 | 1 | -9/+13 |
| | | | | | The fix from bug #351828 broke some of the update avoidance behavior from bug #275945, and this patch should restore it. | ||||
* | Fix some poor installed/masked || choices. | Zac Medico | 2011-01-18 | 1 | -12/+7 |
| | | | | This will fix bug #351828. It's fallout from bug #350285. | ||||
* | doebuild_environent: handle A/AA initialization | Zac Medico | 2011-01-18 | 2 | -14/+1 |
| | | | | | This allows elimination of duplicate code, some of which was introduced by the fix for bug #351505. | ||||
* | Fix broken pkg.use.enabled ref from last commit.v2.1.9.33 | Zac Medico | 2011-01-17 | 1 | -1/+1 |
| | |||||
* | EbuildBuild: Ensure $A is defined in pkg_nofetch. | Zac Medico | 2011-01-17 | 1 | -1/+8 |
| | | | | This will fix bug #351505. | ||||
* | Scheduler: remove uninstall task from pkg_queuev2.1.9.32 | Zac Medico | 2011-01-17 | 1 | -0/+4 |
| | |||||
* | Update copyright headers for touched files. | Zac Medico | 2011-01-16 | 1 | -1/+1 |
| | |||||
* | Tweak PollScheduler signal handling. | Zac Medico | 2011-01-16 | 2 | -33/+37 |
| | |||||
* | Remove duplicate Scheduler constructor call. | Zac Medico | 2011-01-16 | 1 | -3/+0 |
| | |||||
* | Add PollScheduler.terminate() for interruption. | Zac Medico | 2011-01-16 | 5 | -27/+129 |
| | | | | | This allows PollScheduler instances to do basic cleanup and terminate gracefully when SIGINT or SIGTERM signals are received. | ||||
* | When killed by signal, return 128 + signum. | Zac Medico | 2011-01-16 | 1 | -2/+2 |
| | | | | | This is the same convention that bash uses for returncodes of processes that are killed by signals. | ||||
* | AsynchronousLock: use process by default | Zac Medico | 2011-01-16 | 1 | -1/+9 |
| | | | | | | | The default behavior is to use a process instead of a thread, since there is currently no way to interrupt a thread that is waiting for a lock (notably, SIGINT doesn't work because python delivers all signals to the main thread). |