Most changes according to reports and fixes posted by Wesbane on the forums, but I've tweaked most of his code and added a few things. Most notable change is that I'm not accounting for the player's possible undead units, because I intend to remove the Undead Bat.
Regression from e4aa7c0d9504b5286ece73941421768948024774. It was the
drawing widget code that was in the wrong, since styled_widget has never
explicitly guaranteed that it won't shrink.
Fixes a WML error in shop dialogs in the UMC campaign Bad Moon Rising with
a low screen resolution.
I also removed the TODO. Adding dummy add-ons to the add-on manager is now
significantly more difficult because fetching the add-on list from the
server is now a responsibility of the add-on manager.
I don't see the point of such a change. More important, it breaks all
dialogs which search a listbox with find_widget<listbox>() and don't have
a separate code path for --new-widgets, such as the new add-on manager.
In particular, testing the new add-on manager without code changes was
impossible because it was disabled without --new-widgets and failed to
find the add-on list with it.
Excludes level 0 units from being affected, removes the unnecessary prerecruit,prerecall part from an event name, adds handling of leveling and removes a [kill] that could possibly cause problems for other die events.
This should ensure that accessing the page via https will cause the style to also be referenced via https,
thus preventing security warnings from browsers.
Note that this will not take effect on the live copy until it is regenerated.
`std::remove_if()` is not guaranteed to work by swapping the
to-be-removed items with others.
Thanks to @AI0867 for showing the error in my ways^H code.
The check to see whether the timeout command is capable of sending
SIGKILL was broken. This fixes the check, which in turn will ensure
that tests are always terminated. This will resolve travis deciding
that the entire build has timed out.
After predicting fight result with Monte Carlo simulation, it was possible
for the `summary` array and the `slowed` variable to disagree about the
probability of the combatant to be slowed, which resulted in a crash if
another fight was simulated in Monte Carlo mode for the same unit
aftwerwards.
