This adds the Exxx and Fxxx blocks to the usable range for NewGRF
local strings. TTDPatch uses these ranges for internal strings, but as
we don't support any of them anyway, it is "free" real estate for us.

…rectly

…S (#10487)

…489)

Also prevent random_deviation to be below 0.

…igits + 1 for sign

…1 for termination, enough to fit min and max int

When a road vehicle is already running on a multi level crossing, and a train shows up ahead, don't make the road vehicle crash on the side of the train.

…everywhere else

 * If loading fails, it usually returns SL_REINIT which doesn't trigger check
 * If savegame has NewGRFs, it complains NewGRFs are not allowed in intro game

… (#10583)

…ite group mapping.

Previously this was checked after loading ids, or repeatedly checked for
each item.

…(#10622)

This can happen if NewGRFs are missing so that engine or road type definitions are wrong.

Wrong tile used in ownership checks

In the case where the object is on an unowned canal tile and
the new canal tile is owned

…m (#10657)

…ble vehicles at the start date. (#10673)

…#10686)

…10690)

Additionally use TC_FORCE to prevent additional colours in the shaded text.

… handling.

Don't allocate 128KB on stack, but rather on the heap.

…(#10751)

…haracters. (#10775)

GCC warns about redundant std::move while clang warns when they are missing, so we silence the less harmful one

When rail toolbar or rail waypoint build windows closed

This allows selection of the correct face in truetype fonts containing
multiple faces.

Manual cherry-pick from albertodemichelis/squirrel@23a0620

This replaces the separate window class and number properties, and
allows the window system to close dropdowns automatically.

…916)

Basically, we haven't been a good neighbour. Turns out you shouldn't
actually call FcFini when you are done, as some library might still
want to use FontConfig. And they use a shared instance for their
administration.

The idea is that you call FcInit once, and use FcConfigReference
after that to get an instance, you can release. This entry is
ref-counted, and things happen automatically based on that.

At least, I think.

Basically, we drop RelativeMode completely, and use the same trick
as used by the Windows driver: read all motion events till the last
one, and use that as value.

The function is only called with fix_at=true, so don't support
the other cases.

No backend uses it anymore, so also no longer any need to support
it.

Wayland doesn't support mouse warping, X11 only for native
systems (so not for remote desktop, WSLg, etc), and emscripten
neither without complications. All these cannot offer a
mouse-lock.

…ast attempt (#10928)

Caused by incorrect order of operations when buying a train engine with refit and attaching free wagons.

…ution screens (#10934)

…0932)

Co-authored-by: Jonathan G Rennison <[email protected]>

…y (#10942)

…val (#10874)

Solution is to not focus any tooltips, so that the dropdown doesn't lose focus. Tooltips don't accept any input so this does not change their behaviour.

…on. (#10954)

Since dropdowns self-close, the detection of re-clicking a dropdown
button no longer worked, as the dropdown is already closed.

Instead set (and then test) a flag on the parent widget to indicate that
the dropdown closed. This method avoids looping windows on every click.

…#10979)

On Windows in fullscreen you cannot reach the top with
the cursor for the halve of the height of your toolbar.

Additionally, on Win10 in fullscreen you can see the actual toolbar.

…uild workflow

This reverts commit ee6e30e.

…s default. (#11000)

Clicking and releasing on the query toolbar icon is meant to select the land-info tool.

This did not work as during closing a window, OnFocusLost() is called, which then closes the window again. These two calls toggled the land-info tool one and off in the same action.

Resolve by not calling Window::Close in OnFocusLost() if the window is already closing.

…11044)

Since dropdown menus now get closed if they lose focus, 'instant close' dropdowns (i.e. the toolbar dropdowns) should no longer execute their action to avoid unintended actions.

Don't invent a packet type in the log message if we can't/don't
read a packet type at all.
Fix packet type not being logged when appropriate.

This fixes various use after free scenarios in error handling paths