summaryrefslogtreecommitdiff
path: root/usr.sbin/pkg_add/OpenBSD
AgeCommit message (Collapse)Author
2009-12-27check for loops and merge sets, in case we end up with real funky depends.Marc Espie
also, reset package->{before} before solving depends, since we may solve them several times.
2009-12-27solver is no longer needed at that point, so kill it. this shaves a bit ofMarc Espie
memory.
2009-12-27don't ask/report the same exec issue twiceMarc Espie
2009-12-27sortMarc Espie
2009-12-27fix partial stems, as noticed by numerous peopleMarc Espie
2009-12-26qualify (Direct Dependencies)Marc Espie
2009-12-26Better, more global way of dealing with forward dependencies, and toMarc Espie
ask just once per-set for exec/unexec. Todo: forward-deps could be merged, but this might lead to dependency loops, we have to detect them and act first Todo: rewrite the other part of forward dependency handling (dep adjustment)
2009-12-26zap message for .libs, be smart about partial-*, it's easy to install/updateMarc Espie
actually
2009-12-26protect against partial-* packing-listsMarc Espie
2009-12-24make the final progress bar vanish unless verbose >= 2, as requested byMarc Espie
Theo.
2009-12-24avoid infinite loop, noticed by Rivo Nurges.Marc Espie
2009-12-22progressmeter always, unless -x is givenTheo de Raadt
ok espie
2009-12-21if we're not in a "pretend" mode, synchronization operations on vstat occurMarc Espie
after the files have been committed to the file system, so we can forget about them... I don't know where I lost that change, restores pkg_add memory usage to very small values...
2009-12-21consistent spellingMarc Espie
2009-12-21much simpler SIG{INFO} handling: create a status object in state andMarc Espie
update it when needed. Install a SIG{INFO} handler at start that will do what's needed.
2009-12-20change verbosity handling: be silent by default, turn on the first level if -v.Marc Espie
Add more levels, specifically, lots of info regarding individual files only visible if -vvvvv. Make -n display the same stuff as normal (need -n -v to display more stuff). Introduce -s, "size-only", turns on -n and does skip any extraction. Sprinkle handlers for SIG_INFO, not really nice yet. Note that -n is non functional for complex updates, because of caching issues I have to fix.
2009-12-19create a $state->todo method to account for sets to install.Marc Espie
have tracker keep track of sets not done, and ask it directly for how many there's still to do. Provide an offset into next for cases where we know we already did a set but haven't accounted for it yet. -> more accurate number of sets still to do.
2009-12-17be slightly more paranoid about shell stuff we run.Marc Espie
2009-12-17move the ntogo actual code in the progressmeter, so that -x shuts it off.Marc Espie
2009-12-17add a callback to file creation, so that we get progress meter for hugeMarc Espie
files in packages. Should affect emacs, chrome and the likes.
2009-12-17display progressmeter at 0% at start of install/deinstall.Marc Espie
(better synch for packages that start with a huge binary)
2009-12-14since some packages in updatesets don't need updates, be explicit about it.Marc Espie
This is more complex and simpler at the same time...
2009-12-14when removing sets, remove cant markers as wellMarc Espie
2009-12-14combined updates show the packages that can't be updatedMarc Espie
2009-12-14fuss about install msgs: new method state->ntogo, use it to annotate moreMarc Espie
messages with global progress information. Also, when not finding a suitable update, report it if set is not a simple update. Helps in figuring out cases like gstreamer...
2009-12-14temporary fix for kept packages until we move them awayMarc Espie
2009-12-13okay, so old libs are not 'kept', keep track of updates correctly insteadMarc Espie
2009-12-13clean up code, no functional changeMarc Espie
2009-12-13simplify set printingMarc Espie
2009-12-12update will ask about conflicts, so we need that in updateinfo...Marc Espie
2009-12-12autovivifiedMarc Espie
2009-12-12fix the fam->libgamin issue. Look for update candidates in self beforeMarc Espie
asking the repository. This solves this kind of conflicts. also, register packages we keep, this makes for simpler scenarios (todo: split updateset into further categories)
2009-12-12remove *everything* from to_update, including hintsMarc Espie
2009-12-12tie conflict lists to packing lists and use themMarc Espie
2009-12-11record internal updateset dependencies for new packages, and respect themMarc Espie
while installing (need doing something similar for old packages)
2009-12-07don't use $set->print, consistent messages.Marc Espie
2009-12-07simplify tracker a bit now we know what we're doing.Marc Espie
kill code that never got used
2009-12-05use eval {} everywhere around quirks.Marc Espie
also, remove quirks from the list of packages to update, since it deserves special treatment...
2009-12-05since we don't redisplay missing libraries message, tell why we can't installMarc Espie
in general terms...
2009-12-05better quirks interface: pass the search object that was built, so we canMarc Espie
adjust it (eventually, pass the whole list ?)
2009-12-05change the internal representation of Search::Stem to allow the additionMarc Espie
of supplementary stems (useful for quirks)
2009-12-05oopsMarc Espie
2009-12-05cosmeticMarc Espie
2009-12-05make quirks silent. Track # of updates found, instead of newer size, sinceMarc Espie
we may end up updating sets with newer == 0
2009-12-05fix off-by-one, noticed by frantisek holopMarc Espie
2009-12-04create new updatesets when neededMarc Espie
2009-12-04set-up for quirks, just don't do it yet.Marc Espie
2009-12-04simpler unified lookup for libs inside set/installed packages.Marc Espie
2009-12-04internal API: is_known, for pkgnames that are known by the tracker,Marc Espie
so that we decide whether to create new updatesets. (not efficient, but isolate from rest of code before Tracker refactor)
2009-12-03zap useless debug codeMarc Espie