Converted old GUI1 starting positions tool menu to GUI2
* Added a column displaying existing starting locations' coordinates to
the player numbers list
* May have made the player number entries easier to translate
* Removed overly long infodump about hotkeys that's already part of the
tooltip on the editor palette anyway.
* Ran into a lot of frustration involving adding or substracting one
from various things like map coordinates and team numbers because the
zero-based C arrays assumption doesn't seem to work here.
...are applied immediately when closing settings
Formerly they'd only have an effect when automatic map view updates were
enabled, or when opening the settings dialog again, or when toggling
presets.
...when custom lighting isn't in use
Sliders are normally disabled when custom lighting is, when the dialog
is running. A missing function call solves the pre_show inconsistency.
..."recursion prevention" bullshit I devised back in 2008
A static counter variable wasn't being reset, ever, so a number of
version comparisons during a single session would make the code stop
working.
The basic idea is idiotic and I should probably rewrite this code ASAP.
Found a better way to avoid copying singular iterators.
Gabba, tschmitz please review the code and if it's correct it should be
save to revert 2011-10-23T17:48:59Z!koraq@xs4all.nl.
This applies a slightly modified version of the patch linked in
bug #18399. The changes are coding-style related.
The patch was intended to fix compilation with boost 1.47, but that
boost version is not yet in Debian so didn't test that part. Since it
does fix compilation with clang 2.9 Iapplied it, without waiting for
boost 1.47 to hit Debian.
(fix for bug #16111)
There obviously existed an [endlevel] tag in the savegame at some time
but it must have been lost. Added read and write method
to struct end_level_data.
(fix for bug #16821)
The problem was that this variable gets usually set to false during
the start of the side turns. If check_victory() ends the level between
the side turns, this initialization is missed so linger mode thinks it
has already been ended.
Other call cases of linger() should call the function with
end_turn_== false already. That is, I think they already do, since if
not these would have fallen into this bug category as well.
Gabba, tschitz please review whether the code is correct. It fixes some
abortions of Wesnoth when using the gcc debug containers due to
undefined behaviour.
...to test if the whiteboard is a good replacement."
by gabba, 2010-08-15T04:17:49Z!gabrielmorin@gmail.com
1. The whiteboard is obviously still way not ready to be a replacement.
2. It has been requested by several users (and by me and Gambit).
3. For real multiplayer, the whiteboard is too complicated in
comparison to delay shroud.
4. We need time and testing before 1.10 to remove possible problems
interfering with this old code which can have been introduced in the
time since it was disabled.
