summaryrefslogtreecommitdiff
path: root/usr.sbin/pkg_add/OpenBSD
AgeCommit message (Collapse)Author
2009-10-15allow solve to find dependencies in current UpdateSetMarc Espie
2009-10-15sync updatesets with tracker when we create them...Marc Espie
2009-10-15separate the dependencies in several lists for registration, so thatMarc Espie
newer can have several entries.
2009-10-15remove parameter we no longer useMarc Espie
2009-10-15the to_install/to_update stuff is now in the tracker object.Marc Espie
2009-10-15move the installed check in Tracker, where they belongMarc Espie
2009-10-15pkg_add will need to track installation progress globally, so that we don'tMarc Espie
solve dependencies with stuff that we want to update later... create that information, to be used...
2009-10-15reorg codeMarc Espie
2009-10-15missed oneMarc Espie
2009-10-15let handle compute pkgnames correctly, simplifies code a bitMarc Espie
2009-10-14move OpenBSD::Handle into its own file. I guess I need to disentangle thisMarc Espie
code before I go further....
2009-10-14add a few comments, finish killing kitchensink.Marc Espie
2009-10-14use choose_locationMarc Espie
2009-10-14make signature checking more quiet: discard openssl's STDERR, keep a tallyMarc Espie
of signed packages. - display a message only if signed packages have been signed. - show IN BIG CAPS unsigned package numbers if a mixture of signed/unsigned has been seen.
2009-10-14add a "print" method to ProgressMeter and StubProgress to allow me to cleanMarc Espie
up the current mess (e.g., print things out to stderr without needing to explicitly clear the current line if needed). create a choose_location method directly in state, that will replace all instances of looking for a package eventually (hopefully).
2009-10-13deal with UpdateSets earlier: build an updateset, and ask Update to completeMarc Espie
it. Tweak the control for updates to move to main program. Eventually, we will run updates on the fly (there's still some merging code and debugging to go)
2009-10-13you have to hate those stupid tools:Marc Espie
if you echo nlist |ftp some_url the site *won't* give you an error code, it will just spew out a stupid line like: Failed to change directory. Can't change directory to pub/OpenBSD/zoinx: No such file or directory and that on stdout !!! Detect this in list grabbing, and display a decent error message for the user. Avoids some duh moments where you won't find shit because you mistyped the url...
2009-10-12more handle -> newer.Marc Espie
it is going to get more complicated sooner....
2009-10-12more simple uses of set->newerMarc Espie
2009-10-12looking through installed packages is cool, but remember to filter outMarc Espie
older stuff in the updateset...
2009-10-11repair tagging of user-specified packagenames, so that manual-install ↵Marc Espie
actually occurs some of the time...
2009-10-11maybe I'll change the code later (two XXX), but allows flavor after stem,Marc Espie
so that e.g. pkg_add star-- pkg_add star--static are now not ambiguous.
2009-10-11need to save more stuff about updateset for complicated upgrade scenarios.Marc Espie
2009-10-11fix arguments orderMarc Espie
2009-10-11don't copy digital sig over when it makes no sense...Marc Espie
2009-10-11store localbase in solver, in preparation to multiple handles.Marc Espie
2009-10-11overlay extract/delete/install messages, for shorter reports in normal modeMarc Espie
2009-10-11store installed packages in a hash, since we have simple keys for that.Marc Espie
provide newer_names/older_names methods, and a shorter print version of an updateset (to display during update since it's no longer quite pkg_add -r)
2009-10-10have modifiers return self so we can chain themMarc Espie
2009-08-24URI-decode the filenames in the listing used for HTTP pkg_add updates.Stuart Henderson
Fixes a problem with pkg_add -u where package names are encoded in the listing (e.g. those containing + like gtk+2-2.16.5 on standards-compliant servers). Problem noted and fix tested by ajacoutot@ against a server running lighttpd. No regressions against Apache. ok espie@
2009-07-03allow pkg_mklocatedb -p to work again; from espie@Christian Weisgerber
2009-06-25pkgpath are only used for updates, so matching should only be pkgpath vsMarc Espie
default info, and not maximal matching, otherwise too many things will match. problem noticed by marco@ With input from sthen@ and tests from kili@
2009-06-10fix dewey comparison for suffixes (no early return)Marc Espie
2009-06-06fix direct pkgpath caseMarc Espie
2009-06-06quick path for pkg cfl: check whether the new package conflicts withMarc Espie
installed stuff first. If that's the case, and we're not in replacement mode, we don't really need to scan all the installed packages.
2009-06-06my $_, except for the server code, which could still run on older perlMarc Espie
2009-06-04Canonicalise package names when comparing them against the package database.wcmaier
This unconfuses pkg_* when passed paths that, while valid, don't exactly match /var/db/pkg. For example, the following now works (note the double slash): $ pkg_info /var//db/pkg/somepackage Hasty tab-completers the world over shall rejoice. Tweaks and ok espie@.
2009-06-03rewind DATA, so that the next scp: url will have code to connect to...Marc Espie
2009-05-11remove an extraneous slash from the requested URL; some strict proxiesStuart Henderson
(and maybe http servers) have a problem with it. ok espie@
2009-04-24oopsMarc Espie
2009-04-22fix creating packages with long names in plist; from espie@Christian Weisgerber
2009-04-20restore tagged user packages to functionality (for now)Marc Espie
2009-04-19remove class FatPackageLocation. Instead make PackageLocation a hierarchyMarc Espie
so that I can handle installed packages without playing ping-pong with the PackageRepository hierarchy (mostly)
2009-04-19synch with my new style search/location changesMarc Espie
It's likely there will be some fallout, but it's getting a bit too large to keep around. This does kill a few very old oddities as well.
2009-04-18don't copy DigitalSig to partial packages, it makes no senseMarc Espie
2009-04-14initial implementation of package signatures, based on x509 certificates andMarc Espie
smime detached signatures.
2009-03-07tweak code to use same comparison operations as PackageNames.Marc Espie
so the "special code" to handle weird suffixes and epoch names (v*) finally comes into operation. regress tests still pass, I still expect some breakage...
2009-03-07introduce versionspecs as a derivative object from version.Marc Espie
Allows an operator such as <=. also overrides the compare operator for the pnum part, because specifically, in pkgspecs, if the pnum is not specified, it stands for "any pnum will do"... yes, that's hairy...
2009-03-05call may_check_digest after extractionMarc Espie
2009-03-05more stuff to check digests during extraction, if wanted.Marc Espie
Fix read/write of digitalsignature stuff. don't integrate manual-installs, since it's highly variable. create correct category for digitalsignature, so it gets written out.