| Age | Commit message (Collapse) | Author | |
|---|---|---|---|
| 2014-02-01 | style | Marc Espie | |
| 2014-02-01 | sort out the code to figuring out kept packages. | Marc Espie | |
| Fix a (formerly) silent bug: kept packages *must* be completed somewhere. (older solve_depends would create dummy plists, new one errors out on them!) | |||
| 2014-02-01 | let solve_depends work as soon as we have update_info | Marc Espie | |
| 2014-02-01 | handling of critical section was wrong. Interrupting an update once we're | Marc Espie | |
| "committed" is a bad idea. Make it possible by ^C^C still, but try to get out only when the db is in a safe state. | |||
| 2014-02-01 | don't bother caching, it's no longer useful | Marc Espie | |
| 2014-02-01 | use conflict_list directly from handle. | Marc Espie | |
| (only user left of plist->conflict_list is PkgCfl::register thru OldLibs stub_list) | |||
| 2014-02-01 | no need to test for plist, we get the conflict_list since we have update_info | Marc Espie | |
| 2014-02-01 | handles are more adequate places to stash conflict lists, since they can | Marc Espie | |
| be computed directly from update_info (without the full plist) | |||
| 2014-02-01 | style | Marc Espie | |
| 2014-01-31 | @md5 annotations died at the end of 2008, so it's reasonable to scrape the | Marc Espie | |
| code now. | |||
| 2014-01-31 | no longer a need for the solver, which is fortunate as this creates | Marc Espie | |
| a cycle (in perl -current at least). found by careful usage of Devel::Leak::Object and going back in time to 2007... | |||
| 2014-01-31 | avoid creating backrefs | Marc Espie | |
| 2014-01-31 | handle conflict with partial in a less hackish ways, by passing an | Marc Espie | |
| extra "with_partial" parameter to specs. This also removes quite a lot extraneous objects. | |||
| 2014-01-31 | minor fix: don't go to cache if genuine install | Marc Espie | |
| 2014-01-30 | wipe locations a bit more thoroughly | Marc Espie | |
| 2014-01-30 | avoid storing dead keys. | Marc Espie | |
| quirks is exceptional, since it needs to be cached negatively. | |||
| 2014-01-30 | don't save empty errors | Marc Espie | |
| 2014-01-30 | indent fix | Marc Espie | |
| 2014-01-30 | most PkgSpec are default conflicts, so not storing empty constraints... | Marc Espie | |
| 2014-01-30 | not really a need to store state in the location, when we can ask | Marc Espie | |
| our repository for it. | |||
| 2014-01-30 | minor bug-fix, make sure partial is handled properly. | Marc Espie | |
| also use delete instead of undef so that they entry vanishes... | |||
| 2014-01-30 | don't create empty entries | Marc Espie | |
| 2014-01-30 | finish killing old annotation | Marc Espie | |
| 2014-01-25 | more ArcCheck nits: don't allow a file to be u-r/g-r without an explicit | Marc Espie | |
| @mode annotation. This mostly fixes packaging bugs (like documentation only readable by root and such nonsense), and also force *actual* mode settings for security reasons to be apparent in the packing-list. | |||
| 2014-01-25 | somehow, bytes got garbled. Be less cute and seek twice, that works. | Marc Espie | |
| sorry about that. | |||
| 2014-01-24 | subject -z hints (fuzzy update) to quirks' tweak_search. | Marc Espie | |
| Reminded by sthen@ This solves the 5.4 -> 5.5 delete/reinstall gap for changed pkgnames. | |||
| 2014-01-23 | tweak the way we create and sign packages significantly: | Marc Espie | |
| all modern compression formats support multi-stream archives. Store the packing-list as a separate stream, and note where the first stream ends when reading back the packing list. When pkg_sign sees a split-stream archive, it does not need to unpack/repack the whole archive: it can just sign the packing-list, close the stream, and copy the next stream(s) from the source archive verbatim. This is perfectly transparent to pkg_add, but abuses IO::Uncompress::Gunzip internals slightly (to be solved with cpan). also adds explicit option -C for generating a SHA256 list of all packages. and okay naddy@ | |||
| 2014-01-23 | oops, pkg_sign is unhappy without -o. | Marc Espie | |
| Minimal fix, in order not to create too much conflict with pending work... (problem noticed by naddy@) | |||
| 2014-01-23 | more sophisticated warning: builds with empty SIGNING_PARAMETERS | Marc Espie | |
| run with -Dunsigned, so they shouldn't show UNSIGNED PACKAGES while handling dependencies. *but* if we install a mix of signed and unsigned packages, ALWAYS show the line. | |||
| 2014-01-23 | new interface to be able to manipulate output better | Marc Espie | |
| 2014-01-23 | explain signatures | Marc Espie | |
| 2014-01-23 | clean-up synopsis | Marc Espie | |
| 2014-01-23 | check parameters better. | Marc Espie | |
| create -o directory if necessary | |||
| 2014-01-23 | separate pkg_sign documentation. Option description becomes MUCH saner. | Marc Espie | |
| 2014-01-23 | don't create cheap gzip archives if non signed, there's a better way. | Marc Espie | |
| 2014-01-23 | zap the sign-only part (moved to PkgSign.pm). | Marc Espie | |
| reuse the signer part (move to Signer.pm) | |||
| 2014-01-23 | reorg code: make "signing only" a separate command, as it has nothing in | Marc Espie | |
| common with pkg_create. Move the signing part in its own module, so that PkgCreate can reuse it for "signing while we build" mode. naddy@ agrees | |||
| 2014-01-22 | Fix the ambigous text in afterboot. For a newbie, the text really | juanfra | |
| said: "download the packages from a mirror and install these with pkg_add". Also add a mention to pkg.conf in pkg_add.1. Help and suggestions from jmc@ and sthen@. OK from both. | |||
| 2014-01-21 | obvious .Pa fixes; found with mandocdb(8) | Ingo Schwarze | |
| 2014-01-20 | for the SHA256 file, output checksums in base64; ok espie@ | Christian Weisgerber | |
| 2014-01-18 | use the standard notation of <set>XX.tgz for the install sets | Alexander Hall | |
| from frantisek holop, thanks! "Go ahead" espie@ | |||
| 2014-01-18 | involuntary sneak preview, revert | Marc Espie | |
| 2014-01-18 | ...and sort SHA256 properly | Marc Espie | |
| 2014-01-18 | caching a bit more tricky... problem noticed by naddy@ | Marc Espie | |
| 2014-01-17 | do check_digest per plist. | Marc Espie | |
| show UNSIGNED packages unconditionally. | |||
| 2014-01-17 | make ArcCheck less confusing, don't archive stuff that WON'T survive | Marc Espie | |
| extraction anyways (those metadatas ARE in the packing-list anyways). | |||
| 2014-01-17 | "recognize" extended header thingies (we don't handle them, we just | Marc Espie | |
| spew more useful error messages) | |||
| 2014-01-17 | use internal gunzip, now that the fh leak is closed. | Marc Espie | |
| 2014-01-17 | oops, keeping _current around means I still have a live fh after deref. | Marc Espie | |
| So remove it too. Accordingly, the code that resyncs archives on reopen needs us tracking the name itself. | |||
| 2014-01-17 | tweak documentation to match what's going on | Marc Espie | |
 |
index : src | |
| OpenBSD base system |
| summaryrefslogtreecommitdiff |