The unified diff between revisions [8846a6b8..] and [ebdccab0..] is displayed below. It can also be downloaded as a raw diff.

This diff has been restricted to the following files: 'ChangeLog'

#
#
# patch "ChangeLog"
#  from [cf7af27d617c006696a08efcd68a9bd9fd401a7a]
#    to [cef9fe8e359d630eb8f61133684772293748eade]
#
============================================================
--- ChangeLog	cf7af27d617c006696a08efcd68a9bd9fd401a7a
+++ ChangeLog	cef9fe8e359d630eb8f61133684772293748eade
@@ -1,3 +1,304 @@
+2006-02-27  Matt Johnston  <matt@ucc.asn.au>
+
+	* std_hooks.lua (ignore_file, dir_matches): add dir_matches helper
+	to avoid including ignored dirs such as '.svn', since dirs now
+	actually exist.
+
+2006-02-26  Nathaniel Smith  <njs@pobox.com>
+
+	* roster_merge.cc: Update notes on tests.
+
+2006-02-25  Nathaniel Smith  <njs@pobox.com>
+
+	* work.cc (attach_node): Apparently the tests depend on update
+	clobbering existing non-versioned files... and we don't have a
+	good solution _anyway_, so just go back to clobbering them.
+
+2006-02-25  Nathaniel Smith  <njs@pobox.com>
+
+	* cset.cc (root_dir_test, invalid_csets_test): Don't test that
+	root dir stuff fails.  Test that it works.
+
+2006-02-25  Nathaniel Smith  <njs@pobox.com>
+
+	* roster.cc (detach_node): Add missing invariant.
+
+2006-02-25  Nathaniel Smith  <njs@pobox.com>
+
+	* roster_merge.cc (roster_merge): Fix strange merge-created
+	problem...
+
+2006-02-25  Nathaniel Smith  <njs@pobox.com>
+
+	* work.cc (attach_node): This code should use path_exists, not
+	file_exists.
+
+2006-02-24  Timothy Brownawell  <tbrownaw@gmail.com>
+
+	* netcmd.cc: Ignore protocol version field on usher_cmd packets. It
+	should now be possible to use an usher to redirect connections based
+	on netsync version.
+	* netcmd.hh: Remove unused netcmd::get_version() and
+	unused/unimplemented netcmd::netcmd(u8 version).
+
+2006-02-24  Richard Levitte  <richard@levitte.org>
+
+	* commands.cc (ls_changed): I was a bit overly paranoid about the
+	possibilities with C++ and defined an explicit functor for the set
+	instead of relying on the automatic generation of less<file_path>.
+	Derek Scherger made me realise I was a bit overzealous, and this
+	change removes the explicit functor.
+
+2006-02-23  Matthew Gregan  <kinetik@orcon.net.nz>
+
+	* enumerator.hh: Another GCC 4.1 compile fix.
+
+	* tests/t_log_selectors.at: UnXFAIL.
+
+	* commands.cc (complete): Refactor selector completion to enable
+	return of single or sets of completed revisions.
+	(CMD(log)): Do something sensible with selectors that return
+	multiple revisions.
+
+	* tests/t_log_selectors.at: XFAILed test for bug #15877.
+
+	* testsuite.at: Add it.
+
+	* app_state.cc, app_state.hh, commands.cc, monotone.cc,
+	options.hh: Revert failed UI experiment: reenable logging merges
+	by default and rename --merges back to --no-merges.
+
+	* contrib/color-logs.{conf,sh}, contrib/monotone-notify.pl,
+	contrib/monotone.el, tests/t_log_nofiles_nomerges.at,
+	testsuite.at: Handle --no-merges.
+
+	* monotone.texi: Document --no-merges.
+
+2006-02-23  Matt Johnston  <matt@ucc.asn.au>
+
+	* enumerator.{cc,hh}: avoid transferring deltas on both sides of merge
+	revisions, and prefer deltas to data when both are available.
+	See
+	https://savannah.nongnu.org/bugs/?func=detailitem&item_id=15846
+
+2006-02-21  Nathaniel Smith  <njs@pobox.com>
+
+	* work.cc (detach_node): This time for sure!
+
+2006-02-21  Nathaniel Smith  <njs@pobox.com>
+
+	* work.cc (detach_node): Oops, x != y != !(x == y).
+
+2006-02-21  Nathaniel Smith  <njs@pobox.com>
+
+	* roster_merge.cc: Fixup after merge.
+
+2006-02-21  Nathaniel Smith  <njs@pobox.com>
+
+	* work.cc (detach_node): Check if we are passed the root dir, and
+	error out if so.
+
+2006-02-21  Matt Johnston  <matt@ucc.asn.au>
+
+	* commands.cc (pid_file): newline-terminate the pid
+
+2006-02-21  Richard Levitte  <richard@levitte.org>
+
+	* contrib/usher.cc (server::set_hosts): Erasing the list node that
+	we're iterating on, then trying to go to the next node doesn't
+	work.  Save the iterator, then increment it before erasing the
+	node using the saved value.  No more segfaults.
+
+2006-02-21  Matthew Gregan  <kinetik@orcon.net.nz>
+
+	* sanity.hh: Work around roster_merge.cc compilation failure with
+	GCC 3.4.
+
+2006-02-20  Richard Levitte  <richard@levitte.org>
+
+	* monotone.cc (cpp_main), options.hh, ui.cc (redirect_log_to),
+	ui.hh (struct user_interface): Add --log option, to redirect the
+	log lines to a file.
+
+	* monotone.texi (OPTIONS): Document it.
+
+	* po/sv.po: Translate help for this option to Swedish.
+
+	* testsuite.at, tests/t_log_to_file.at: Test it.
+
+2006-02-20  Matt Johnston  <matt@ucc.asn.au>
+
+	* database.cc (remove_version): get rid of dangling deltas,
+	don't try to put data or deltas if they already exist.
+	* tests/t_db_kill_rev_locally_2.at: un-XFAIL
+
+2006-02-20  Matthew Gregan  <kinetik@orcon.net.nz>
+
+	* testsuite.at: Add an ADD_FILE variant that allows use of
+	alternate databases.
+
+	* tests/t_netsync_permissions.at: Missed some cases of the
+	database locking race.
+
+2006-02-19  Nathaniel Smith  <njs@pobox.com>
+
+	* ChangeLog: Fixup after xxdiff lossage.
+
+2006-02-19  Nathaniel Smith  <njs@pobox.com>
+
+	* roster_merge.cc (make_lifecycle_objs): Fix test bug.
+
+2006-02-19  Matthew Gregan  <kinetik@orcon.net.nz>
+
+	* testsuite.at: Add a REVERT_TO variant that allows use of
+	alternate databases.
+
+	* tests/t_netsync_permissions.at: Attempt to avoid a database
+	locking race in this test that is causing spurious failures by
+	using the new REVERT_TO variant to cause revert to be performed
+	using the "client" database.
+
+2006-02-19  Nathaniel Smith  <njs@pobox.com>
+
+	* roster.cc (shallow_equal): Publically expose.
+	* roster.cc, roster_merge.cc: Various compile fixes.
+
+2006-02-19  Nathaniel Smith  <njs@pobox.com>
+
+	* roster_merge.{hh,cc}: Make terminology more consistent.
+	"marking_map" type -> "markings" name, "marking_t" type ->
+	"marking" name.
+
+2006-02-19  Nathaniel Smith  <njs@pobox.com>
+
+	* roster.{hh,cc} (testing_node_id_source): Make this node source
+	available to unit tests in other files.
+	* roster_merge.cc (test_roster_merge_node_lifecycle): New test.
+	Still quite ugly.
+
+2006-02-18  Nathaniel Smith  <njs@pobox.com>
+
+	* roster_merge.cc (roster_merge): Remove obsolete FIXME.
+
+2005-10-19  Matthew A. Nicholson  <matt@matt-land.com>
+
+	* contrib/monotone.bash_completion: Update for 0.25.
+
+2006-02-19  Matthew Gregan  <kinetik@orcon.net.nz>
+
+	* cset.hh (struct editable_tree): Add commit() member function to
+	editable_tree.
+
+	* cset.cc (cset::apply_to): Call editable_tree::commit() after
+	applying any other changes.
+
+	* roster.hh, roster.cc: Empty implementation of
+	editable_roster_base::commit().
+
+	* work.hh, work.cc: Implementation of
+	editable_working_tree::commit() that ensures all detached nodes
+	have been reattached.
+
+	* work.hh (struct editable_working_tree): Add map for tracking
+	path name mappings across node detach operations.
+
+	* work.cc (editable_working_tree::detach_node): Insert path name
+	mappings into map.
+	(editable_working_tree::drop_detached_node,
+	editable_working_tree::attach_node): Report add/drop/rename
+	operations during workspace updates.
+
+	* lua.cc: Use the safer luaL_check* rather than lua_to* in
+	monotone_*_for_lua functions.
+
+2006-02-18  Markus Schiltknecht  <markus@bluegap.ch>
+
+	* tests/t_cvsimport_branch.at, testsuite.at: New XFAIL test for
+	cvs_import branch reconstruction.
+
+2006-02-18  Matthew Gregan  <kinetik@orcon.net.nz>
+
+	* tests/t_log_nofiles_merges.at: Add test for the log options
+	--no-files and --merges.
+
+	* testsuite.at: Add t_log_nofiles_merges.at.
+
+2006-02-13  Nathaniel Smith  <njs@pobox.com>
+
+	* roster_merge.cc (log_conflicts): Tweak string.
+	Add list of tests needed.
+
+2006-02-13  Nathaniel Smith  <njs@pobox.com>
+
+	* roster_merge.cc (is_clean): Simplify.
+	(add_roster_merge_tests):
+	* unit_tests.cc (init_unit_test_suite):
+	* unit_tests.hh (add_roster_merge_tests): Add unit test
+	boilerplate.
+
+2006-02-18  Matthew Gregan  <kinetik@orcon.net.nz>
+
+	* tests/t_db_kill_rev_locally_2.at: Add an XFAIL test for a
+	kill_rev_locally bug reported by Daniel Carosone.
+
+	* testsuite.at: Add t_db_kill_rev_locally_2.at.
+
+	* sqlite/parse.h: Regenerated parse.h from pristine SQLite 3.3.4
+	source.  The version committed in the 3.3.4 import had a bunch of
+	duplicate entries.
+
+	* commands.cc (CMD(identify)): This isn't really a "workspace"
+	command--stick it under "debug" for lack of a better place.
+
+	* commands.cc (CMD(refresh_inodeprints)): Check for a valid
+	workspace rather than failing with an invariant when run outside
+	of a workspace.
+
+	* tests/t_revert_new_project.at: Add an XFAIL test for a bug where
+	reverting a file added in a new project will leave the workspace
+	in a bad state until MT/work is removed manually.
+
+	* testsuite.at: Add t_revert_new_project.at.
+
+	* app_state.cc, app_state.hh, commands.cc, monotone.cc,
+	options.hh: Add '--no-files' option to log to allow users to
+	exclude the list of files changed in each revision from the log
+	output.
+
+	* monotone.texi: Document '--no-files', and '--next' and '--diffs'
+	while there.
+
+2006-02-17  Matthew Gregan  <kinetik@orcon.net.nz>
+
+	* lua.cc, lua.hh, monotone.texi, std_hooks.lua, test_hooks.lua,
+	testsuite.at: Remove unused non_blocking_rng_ok hook.
+
+	* sqlite/*: Import SQLite 3.3.4.
+
+2006-02-16  Patrick Mauritz  <oxygene@studentenbude.ath.cx>
+
+	* netsync.cc (handle_new_connection): Netxx::Address.get_name()
+	returns NULL every now and then. if so, continue with "" instead
+
+	* sqlite/parse.c: move #line under all #include directives so
+	the compiler can't be confused by it.
+
+2006-02-14  Richard Levitte  <richard@levitte.org>
+
+	* Makefile.am (htmldir): Add variables so monotone.html is created
+	and installed automatically.
+	This is prompted by debian/monotone.html, which indicates
+	monotone.html should be available.
+
+	* netsync.cc (serve_connections): Correct spelling.
+
+2006-02-13  Matthew Gregan  <kinetik@orcon.net.nz>
+
+	* sanity.cc (sanity::dump_buffer): Fix a SEGV when we're in an
+	error unwind and about to ask the user to mail us the crash
+	log--we must use FL() rather than F() here, since by the time this
+	is called we can't rely on the i18n infrastructure being alive.
+
 2006-02-12  Nathaniel Smith  <njs@pobox.com>

 	* netsync.cc (serve_connections): Revert garbage that I