From 088aa884d4369bff024c76035173ffe4f2ae455d Mon Sep 17 00:00:00 2001 From: Thierry Vignaud Date: Wed, 12 Sep 2007 16:17:47 +0000 Subject: update --- ChangeLog | 787 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 787 insertions(+) (limited to 'ChangeLog') diff --git a/ChangeLog b/ChangeLog index 150295e..cbb52a3 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,3 +1,790 @@ +2007-09-12 15:11 Pixel + + * : better indicate the test case in urpmi + +2007-09-12 15:04 Pixel + + * : update doc + +2007-09-12 15:04 Pixel + + * : update doc + +2007-09-12 15:01 Pixel + + * : update doc + +2007-09-12 15:00 Pixel + + * : create _remove_all_rejected_from() out of disable_selected() + +2007-09-12 14:57 Pixel + + * : create _remove_rejected_from() + +2007-09-12 14:47 Pixel + + * : update doc + +2007-09-12 14:44 Pixel + + * : factorize code into _set_rejected_from() + +2007-09-12 14:38 Pixel + + * : replacing {backtrack}{closure} with {closure} since it looks + like it was a typo + +2007-09-12 14:27 Pixel + + * : adapt doc to is really done (may kill {backtrack}{closure} + later) + +2007-09-12 13:16 Pixel + + * : update doc + +2007-09-12 13:12 Pixel + + * : - fix typo in previous commit: pass $urpm to set_selected() for + debug message + - {from} options to resolve_rejected_ and set_selected is now a + pkg object + instead of a fullname string + +2007-09-12 13:08 Pixel + + * : add debug message + +2007-09-12 13:07 Pixel + + * : simplify + +2007-09-12 13:06 Pixel + + * : handle {size} in set_rejected() + +2007-09-12 12:59 Pixel + + * : - add ability to merge {closure} fields in set_rejected (from + $options{why}) + - {from} is a fullname here + - {why} is a hash + - use set_rejected() to factorize code + +2007-09-12 12:12 Pixel + + * : cleanup + +2007-09-12 12:11 Pixel + + * : fix typo in previous commit + +2007-09-12 12:03 Pixel + + * : create set_rejected() out of resolve_rejected_() + +2007-09-12 11:03 Pixel + + * : dep field "choices" is here for debug purpose, rename it + _choices + +2007-09-12 11:00 Pixel + + * : create _dep_to_name() and use it + +2007-09-12 10:50 Pixel + + * : drop unused option "closure_as_removed" + +2007-09-12 10:14 Pixel + + * : backtrack_selected() was always returning empty list, so + backtrack_selected_psel_keep() should return empty list. but it's + simply to + modify callers to take into account the return value is () + +2007-09-12 10:07 Pixel + + * : move var declaration where it's used + +2007-09-12 10:06 Pixel + + * : move function created in previous commit + +2007-09-12 10:04 Pixel + + * : create _choose_required() out of + resolve_requested__no_suggests_() + +2007-09-12 09:50 Pixel + + * : this is good to have in NEWS + +2007-09-11 16:27 Pixel + + * : document side-effects of each functions + +2007-09-11 15:41 Pixel + + * : add some doc + +2007-09-11 15:35 Pixel + + * : rename disable_selected_unrequested_dependencies() into + disable_selected_and_unrequested_dependencies() + +2007-09-11 15:09 Pixel + + * : don't keep a hash when we really want its values + +2007-09-11 15:03 Pixel + + * : - create strict_arch_check() out of find_required_package() + - and factorize some more code into $may_add_to_packages + +2007-09-11 14:46 Pixel + + * : create backtrack_selected_psel_keep() out of + backtrack_selected() and use it + +2007-09-11 14:12 Pixel + + * : drop keep_unrequested_dependencies option (it is unused) + +2007-09-11 14:10 Pixel + + * : drop keep_requested_flag by creating + resolve_requested__no_suggests_() + +2007-09-11 14:04 Pixel + + * : simplify + +2007-09-11 14:00 Pixel + + * : define var where it's needed + +2007-09-11 13:59 Pixel + + * : rename var, $dep will only be used as an element from + @properties + +2007-09-11 13:58 Pixel + + * : rename var, $dep will only be used as an element from + @properties + +2007-09-11 13:52 Pixel + + * : add doc + +2007-09-11 13:37 Pixel + + * : add some doc + +2007-09-11 13:37 Pixel + + * : simplify + +2007-09-11 13:32 Pixel + + * : simplify + +2007-09-11 13:24 Pixel + + * : simplify: make find_candidate_packages() the wrapper around + find_candidate_packages_() instead of the other way round + +2007-09-11 13:21 Pixel + + * : create find_candidate_packages_(), wrapper around + find_candidate_packages() + +2007-09-11 13:16 Pixel + + * : drop callback_backtrack, unused for a long time + +2007-09-11 11:54 Pixel + + * : fix old typo in comment + +2007-09-11 11:51 Pixel + + * : add a comment + +2007-09-11 10:54 Pixel + + * : fix old typo (rejected field is unused anyway afaik) + +2007-09-11 10:00 Pixel + + * : fix typo in previous typo fix + +2007-09-11 10:00 Pixel + + * : fix typo in previous commit :-/ + +2007-09-11 09:58 Pixel + + * : - speedup is_arch_compat (x7) by keeping the platform in a + cache + +2007-09-11 09:55 Pixel + + * : add missing free() + +2007-09-11 08:44 Pixel + + * : move doc where it belongs + +2007-09-11 07:28 Pixel + + * : add comment + +2007-09-10 16:58 Pixel + + * : simplify prototype of find_candidate_packages() + +2007-09-10 16:56 Pixel + + * : rename find_chosen_packages() into find_required_package() + (better name) + +2007-09-10 16:49 Pixel + + * : - pass the prefered choices to {callback_choices}: this allows + urpmi to select + all the prefered packages according to installed locales + +2007-09-10 15:48 Pixel + + * : - do not propose packages for non installed locales + +2007-09-10 15:48 Pixel + + * : - do not propose packages for non installed locales + - cleanup code + +2007-09-10 15:46 Pixel + + * : fix typo in previous commit + +2007-09-10 15:16 Pixel + + * : create _find_chosen_packages__sort() out of + find_chosen_packages() + +2007-09-08 20:29 Pixel + + * : more detailed doc + +2007-09-08 09:24 Pixel + + * : typo in comment + +2007-09-08 09:16 Pixel + + * : rename lists with "closure" in the name to things like + "xxx_todo" or + "all_xxx", i find it more readable, and it helps making the + difference with + {closure} field + +2007-09-08 09:10 Pixel + + * : rename local var for clarity + +2007-09-08 09:07 Pixel + + * : rename local var + +2007-09-08 09:05 Pixel + + * : document $state fields + +2007-09-08 08:10 Pixel + + * : rename a local variable for clarity + +2007-09-08 08:09 Pixel + + * : create _handle_conflicts() out of + resolve_requested__no_suggests() + +2007-09-08 07:39 Pixel + + * : fix typo in older commit + +2007-09-08 07:38 Pixel + + * : use function calls instead of method calls (it allows + perl_checker checks when + temporarilly renaming URPM/Resolve.pm package name to URPM) + +2007-09-07 21:56 Pixel + + * : create _handle_diff_provides() out of + resolve_requested__no_suggests() + +2007-09-07 20:49 Pixel + + * : create _compute_diff_provides() and + _compute_diff_provides_one() out of + resolve_requested__no_suggests() + +2007-09-07 20:31 Pixel + + * : factorize regexps into functions + +2007-09-07 19:45 Pixel + + * : make it more clear + +2007-09-07 19:35 Pixel + + * : cleanup + +2007-09-07 19:32 Pixel + + * : create _ids_to_fullnames() and _ids_to_names() + +2007-09-07 19:19 Pixel + + * : fix typo in recent commit + +2007-09-07 19:18 Pixel + + * : make it clear that resolve_rejected_ is modifying @properties + from + resolve_requested() + +2007-09-07 18:39 Pixel + + * : revert previous commit: compute_deps() is still used by mkcd + +2007-09-07 17:55 Pixel + + * : remove unused compute_deps() \o/ + +2007-09-07 17:49 Pixel + + * : simplify: nopromoteepoch is always 1 (no caller ever define it) + +2007-09-07 17:43 Pixel + + * : cleanup: nopromoteepoch is 1 by default in + find_candidate_packages + +2007-09-07 15:44 Pixel + + * : factorize code in functions whatrequires() and + whatrequires_id() + +2007-09-07 15:30 Pixel + + * : simplify: whatrequires is a hash with provides_nosense as key, + don't need to + use ->provides then remove the sense + +2007-09-07 15:19 Pixel + + * : hoist a debug message so that it occurs for other uses of + with_db_unsatisfied_requires() + +2007-09-07 15:06 Pixel + + * : re-use same code for conflict from installed package as is used + for conflict + from selected package. this fixes test_gh() from urpmi + split-transactions--promote test case + +2007-09-07 14:44 Pixel + + * : move code from resolve_requested__no_suggests() into new + function _handle_provides_overlap() + +2007-09-07 14:30 Pixel + + * : prepare to factorize code handling property becoming + unavailable. + + it really seems {why}{conflicts} is only used for debugging + purpose, so this + change won't break. + +2007-09-07 14:24 Pixel + + * : one more debug message + +2007-09-07 11:26 Pixel + + * : small cleanup + +2007-09-07 11:26 Pixel + + * : add some debug messages + +2007-09-05 15:19 Pixel + + * : - fix displaying @sorted whereas some elements have been + removed from it + - more complete error message + +2007-09-05 09:37 Pixel + + * : add debug code + +2007-09-03 15:21 Pixel + + * : 1.80, bugfix release + +2007-09-03 15:21 Pixel + + * : - fix bug in sort_graph (used by build_transaction_set) + +2007-09-03 15:20 Pixel + + * : 1.79 + +2007-09-03 12:59 Pixel + + * : add a test + +2007-09-03 08:08 Pixel + + * : - fix bug in sort_graph (used by build_transaction_set) + +2007-09-03 08:07 Pixel + + * : 1.78 + +2007-08-31 15:54 Pixel + + * : fix dead-loop in build_transaction_set (#33020) + + and ensure the resulting graph is correct by checking it + +2007-08-31 07:02 Pixel + + * : add a comment + +2007-08-29 09:33 Pixel + + * : 1.77 + +2007-08-29 09:33 Pixel + + * : - disable "dropping tags from rpm header" until we can safely + use it + +2007-08-29 09:32 Pixel + + * : 1.76 + +2007-08-28 14:12 Pixel + + * : ensure #31969 doesn't occur anymore. if somethings goes wrong: + - display error message + - then default to one big transaction + + also display error message "using one big transaction" in the old + check (which + i don't really know what it does) + +2007-08-28 13:54 Pixel + + * : - build_transaction_set: new sort algorithm which allow + returning sets of + circular dependent packages, taking into account obsoleted + packages + (fixes #31969). It may still fail in presence of conflicts + + a better fix would be to make ->resolve_requested__no_suggests + handle + obsolete. ie: + - a requires b : bb or b + - bb requires c-1 + - b requires c-2 + - b obsoletes bb + => with a, bb and c-1 installed, "urpmi c" should upgrade bb into + b instead of + removing a and bb. + +2007-08-28 11:03 Pixel + + * : - spec2srcheader: workaround parseSpec returning a header where + ->arch is set + to %{_target_cpu} whereas we really want a header similar to + .src.rpm + (see #32824) + +2007-08-24 12:01 Pixel + + * : - fix split_length > 1 + (eg: "urpmi --split-length 2 a b c" will only install 2 pkgs) + (this bug has not been reported, just discovered it while reading + the code) + + - cleanup + +2007-08-24 11:50 Pixel + + * : - more debug messages + - little rewrite + +2007-08-23 13:22 Pixel + + * : - allow running transaction with justdb option + +2007-08-12 20:45 Pixel + + * : 1.75 + +2007-08-12 20:44 Pixel + + * : - fix dropping tags from rpm header. + it hasn't work since MDK8.1 and rpm 4.0. + it may break urpmi!! but potentially allows a much smaller + hdlist.cz :) + +2007-08-12 11:32 Pixel + + * : 1.74 + +2007-08-12 11:29 Pixel + + * : - sort choices per media, then per version + + nb: + + on 2004-12-13, rgs replaced "$a->id <=> $b->id" with + "$b->compare_pkg($a) || $a->id <=> $b->id" + + Return the list of chosen packages sorted by descending + version (bug #12645). + + the end result is the sorting by media was dropped :-/ + +2007-08-12 10:15 Pixel + + * : cleanup (easier to read) + +2007-08-11 12:21 Pixel + + * : 1.73 + +2007-08-11 12:21 Pixel + + * : - allow running transaction with replagekgs option + +2007-08-10 18:01 Pixel + + * : 1.72 + +2007-08-10 17:50 Pixel + + * : - modify parse_hdlist so that partial hdlist reading can be + used + (needed when some stuff is already done in the callback) + +2007-08-10 17:35 Pixel + + * : add documentation + +2007-08-10 17:34 Pixel + + * : there is no use retrying after a headerRead failure since + headerRead may have + consumed part of the input, and so next headerRead will read in + the middle of + a header, and so always fail. The only real solution is to ensure + rpmlib don't + timeout too quickly (and fd->rd_timeoutsecs is rpmlib internal + only, so we + can't hint it from here. in any case 1sec timeout is really too + low) + +2007-08-09 15:25 Pixel + + * : 1.71 + +2007-08-09 15:24 Pixel + + * : - compilation fixes on rpm < 4.4.8 + +2007-08-09 14:58 Pixel + + * : 1.70 + +2007-08-09 14:58 Pixel + + * : compilation fixes on rpm < 4.4.8 + +2007-08-09 13:24 Pixel + + * : 1.69 + +2007-08-09 13:24 Pixel + + * : - "suggests" are no more handled as "requires" + - resolve_requested support "suggests": a newly suggested package + is installed + as if required (can be disabled with option no_suggests) + + nb: URPM.xs change is quite complex since suggests are mostly + seen as requires + inside rpmlib. + +2007-08-08 18:18 Pixel + + * : factorize some code into with_db_unsatisfied_requires: + + #- this function is "suggests vs requires" safe: + #- 'whatrequires' will give both requires & suggests, but + unsatisfied_requires + #- will check $p->requires and so filter out suggests + +2007-08-08 17:38 Pixel + + * : "nopromoteepoch => 1" is the default option in + ->unsatisfied_requires + +2007-08-03 14:43 Pixel + + * : 1.68 + +2007-08-03 14:42 Pixel + + * : - add $trans->Element_version and $trans->Element_release + +2007-07-05 13:36 nanardon + + * : - 0.67 + +2007-07-04 22:26 nanardon + + * : - Urpm_read_config_files properly return the value to perl + - kill useless var dcl + +2007-07-04 22:21 nanardon + + * : - fix parseSpec argument + +2007-07-04 16:43 nanardon + + * : - add osscore() and archscore() function to evaluate computer + compatiblity to an arbitrary value + +2007-07-04 15:53 nanardon + + * : - add Pkg_is_platform_compat() and Urpm_platformscore() coming + with rpm 4.4.8 + +2007-07-02 15:14 Pixel + + * : 1.66 + +2007-07-02 15:11 Pixel + + * : - fix --auto-select skipping some packages because of other + packages providing + a more recent version, but no obsolete between those packages. + the fix is to revert commit from Aug 2002: + "fixed propable old package (according provides) requested by + request_packages_to_upgrade." + + hopefully this change won't break too much... + + for the record, a few issues: + - skipping java-1.5.0-gcj-1.5.0.0-14.7mdv2008.0.i586 since + java-1.7.0-icedtea-1.7.0.0-1.3mdv2008.0.i586 provides a more + recent version of jre (1.7.0 vs 1.5.0) + - skipping emacs-common-22.1-2mdv2008.0.i586 since + gnus-emacs-5.10.8-1mdv2007.0.noarch provides a more recent vers + whereas java-1.5.0-gcj-1.5.0.0-14.6mdv and emacs-common-22.1-1mdv + are installed + +2007-07-02 15:05 Pixel + + * : help perl_checker + +2007-07-02 14:38 Pixel + + * : help debugging the strange behaviour of this code + +2007-06-21 14:13 nanardon + + * : - 1.42 + +2007-06-19 14:44 nanardon + + * : - make is_arch_compat rpm 4.4.8 compliant + +2007-06-15 20:00 Pixel + + * : document the fact that build_synthesis returns true on success + +2007-06-12 14:03 Pixel + + * : fix release date + +2007-06-12 13:57 Pixel + + * : 1.64 + +2007-06-12 13:56 Pixel + + * : - hack on $pkg->is_arch_compat to make it return true for + noarch packages + when using rpm 4.4.8 (#31314) + +2007-05-09 16:00 Pixel + + * : - new release, 1.63 + - add $trans->Element_fullname + +2007-05-03 14:51 Pixel + + * : 1.61 has not been released + +2007-05-03 14:49 Pixel + + * : 1.62 + +2007-05-03 14:49 Pixel + + * : - pass the virtual package name as a parameter to + {callback_choices} in + ->resolve_requested + +2007-04-27 12:37 Pixel + + * : re-sync after the big svn loss + +2007-04-27 12:37 Pixel + + * : re-sync after the big svn loss + +2007-04-27 12:31 Pixel + + * : - 1.61 + - add $trans->NElements and $trans->Element_name + to be able to display name of uninstalled package in + callback_uninst + - fix b--obsoletes-->a and c--conflicts-->a prompting for + upgrading a + (need a fix in urpmi which rely on the $state->{rejected} + to upgrade (-U) b instead of installing (-i) it) + +2007-04-27 12:30 Pixel + + * : re-sync after the big svn loss + +2007-04-24 19:11 Pixel + + * : re-sync after the big svn loss + 2006-12-04 10:28 Pixel * URPM.xs: correctly set "\0" -- cgit v1.2.1