Age | Commit message (Collapse) | Author | |
---|---|---|---|
2005-08-17 | fix a few typos. | Marc Espie | |
link to RequiredBy(3p) now that it exists. | |||
2005-08-17 | drop paragraph about ftp-proxy. | Marc Espie | |
jmc@ insisted... and yeah, it's a bit confusing. | |||
2005-08-17 | document new method | Marc Espie | |
2005-08-17 | better message when we can't resolve a stem, taken from pkg_delete. | Marc Espie | |
2005-08-17 | reuse compute_closure | Marc Espie | |
2005-08-17 | we have a much more accurate image of stuff in -n mode, so we can | Marc Espie | |
ditch this test, pkg_delete -n should no longer loop indefinitely. | |||
2005-08-17 | clean-up, move the compute_closure code to RequiredBy, since it's | Marc Espie | |
quite generic. | |||
2005-08-17 | add newlines in hostapd_fatal() calls, fix some typos. | Reyk Floeter | |
thanks to Brian Poole. | |||
2005-08-17 | works better with the actual file. | Marc Espie | |
2005-08-17 | when looking for updates, select the best version: we know what to do | Marc Espie | |
with p suffixes in a safe way. | |||
2005-08-17 | if we can't read the packing-list, it's because it's a bad pkgname. | Marc Espie | |
reported by niklas@ | |||
2005-08-17 | ftp-proxy misbehavior is of the past. | Marc Espie | |
2005-08-17 | save architecture as well for ExtraInfo, so that pkg_add -u won't try | Marc Espie | |
to add packages for the wrong architecture. Noticed by Tim Kornau. | |||
2005-08-17 | document RequiredBy module. | Marc Espie | |
2005-08-17 | restore $main::not that is needed by OpenBSD::RequiredBy, | Marc Espie | |
put an # XXX comment so I won't be tempted to clean it again. | |||
2005-08-17 | rename erase to more appropriate forget | Marc Espie | |
2005-08-17 | fill out description for most classes | Marc Espie | |
2005-08-17 | finish fleshing out the class hierarchy | Marc Espie | |
2005-08-17 | fix class hierarchy indentations | Marc Espie | |
2005-08-17 | document Getopt. | Marc Espie | |
2005-08-17 | document OpenBSD::PackageName | Marc Espie | |
2005-08-16 | big kludge that allows updating mozilla-thunderbird from 3.7 -> current. | Marc Espie | |
okay pvalchev@ | |||
2005-08-16 | unsafe operations -> potentially unsafe operations | Marc Espie | |
2005-08-16 | make sure libs get registered if we take a shortcut and don't reinstall | Marc Espie | |
the package, or if we call add_bogus_libs directly. | |||
2005-08-16 | avoid computing signature twice! | Marc Espie | |
slightly more verbose message. | |||
2005-08-16 | check special names too. | Marc Espie | |
Fixes a nasty bug: you could have /var/db/pkg mounted read-only, and still delete a package and keep the packing information around. Also makes pkg_add -r estimation of size more accurate, even though it usually doesn't matter, except for packages with huge packing-lists. | |||
2005-08-16 | clean-up TODO list of stuff that was actually done. | Marc Espie | |
2005-08-16 | in case we're pretending to install, there's no borked package to create. | Marc Espie | |
2005-08-16 | put a signature check way earlier, so that user doesn't get confused | Marc Espie | |
seeing update messages for stuff that won't happen anyways. | |||
2005-08-16 | remove trailing space; | Jason McIntyre | |
2005-08-16 | In some cases, pkg_add -r will remove some extra packages while performing | Marc Espie | |
replacement on dependencies. When we finally reach the package that triggered the dependencies, the package it's supposed to replace is no longer there, so do a quick sanity check to only replace packages that have not already been replaced... | |||
2005-08-16 | update dependency explanation, noticed by Gerardo Santana. | Marc Espie | |
2005-08-16 | when you ask to update a package, you generally want to update its | Marc Espie | |
dependencies as well... so do the stem -> pkgname conversion early, and then compute the closure of dependencies on the list of packages we asked for. | |||
2005-08-16 | use a hash for compiled stems, to avoid spurious duplicates. | Marc Espie | |
adjust find_updatenames accordingly. | |||
2005-08-16 | don't mix nlist with ls | Marc Espie | |
2005-08-16 | document the current state of pkg_add -u | Marc Espie | |
2005-08-16 | add code that figures out the new package names after an update. | Marc Espie | |
This is done very carefully, so that pkg_add -u only displays the names to use, and so that it does not interfere with existing code. -F pkgpath recommended for speed. | |||
2005-08-16 | support code for pkg_add -u: | Marc Espie | |
- distant_available(), using the distant listing functions. - stem compiler, to match lots of stems | |||
2005-08-16 | be more specific about `every precaution' in pkg_add, so that people won't | Marc Espie | |
be scared about -F update. Document the nasty issues I've encountered with FTP. | |||
2005-08-16 | make the code that enforces the connection limit a separate function, so | Marc Espie | |
that the distant listing code can use it too. Since the distant listing code creates transient connections, it does not actually need to be tracked, just to ensure that it has a slot ! | |||
2005-08-16 | fix a few minor details: _list() needs to be called as a method. | Marc Espie | |
use nlist *.tgz instead of ls to gain speed. http does not need any command, it just retrieves the directory index. (code not used yet, but so blatantly incorrect) | |||
2005-08-16 | fix getopt line, noticed by Gerardo Santana | Marc Espie | |
2005-08-15 | Fix possible race condition in signal handler. | Esben Norby | |
Report and fix from Micheal Knudsen. Prep, test and OK claudio@ | |||
2005-08-14 | if there's no identifiable version in the spec, and in the package name, | Marc Espie | |
then it matches when it is equal. | |||
2005-08-14 | no reason not to record special files sizes along with md5. | Marc Espie | |
2005-08-14 | catched->caught | Miod Vallat | |
2005-08-14 | zap extra line | Marc Espie | |
2005-08-14 | fix error recovery in replacing case. | Marc Espie | |
- build the list of done items during extract only, since later all files have a tempname. - wrap pre-install in an exception handler, since there already is stuff to save in a partial package... | |||
2005-08-13 | symlinks and hardlinks do not have an md5 | Marc Espie | |
2005-08-13 | don't set $file->{cwd} until *after* we're sure it's defined. | Marc Espie | |
Allows pkg_add to properly report truncated archives, instead of weird can't call method check_name() on an unblessed reference messages. problem reported by brad@, somewhat difficult to reproduce. as usual, the most obvious bugs take the longest time... |