This fixes tons of warnings with clang from a recent xcode version on
macOS (and possibly other systems) complaining that an instantiation
of _singleton is required but no definition is available.
When widget::draw() is called it asks the ThemeEngine to redraw the background
first and then the widget gets redrawn in drawWidget(). The ThemeEngine uses
an extended rect to restore the background to include bevel and shadow effects.
However if this extended rect overlaps with other widgets, since those other
widgets are not redrawn, a part of those will be missing. See for example
bug #6394: GUI: List View save page drawns over font.
In case we get overlap we might need to change the way widgets are drawn so
that all widgets intersecting the area where the backgroud is restored are
redrawn. This commit simply seperate the bevel and shadow effects, and uses
the shadow offset only to extend the bottom and right sides of the rectangle
(while the bevel offset is still used to extend all four sides). This
results in a smaller extended rectangle (if the shadow offset is bigger than
the bevel offset, which is the case of the list view) and thus decrease the
risk of the issue happening. The particular cases described in bug #6394
are all fixed with this change.
PR#921 changes the behavior of the client that if the GUI fails to be
rendered the previously applied settings in the misc category are
restored. Error messages were altered according to the changes.
Bug: #9717 GUI: Indirectly changing 'GUI Language' can produce
inconsistent behaviour when changing some options.
The issue was with the focus or mouse widget as it may not be a
direct child of the dialog (it may for example be a child of the tab
widget in the options dialog) and removing a widget was not
resetting the mouse of focus widget if that widget was not a
direct child.
This is achieved by adding a list of GuiObject to delete to GuiManager
and doing the deletion in GuiManager::runLoop.
The main purpose of this is to avoid the deletion of ButtonWidget object
while their ButtonWidget::sendCommand function is being called. For
example the sendCommand of the Apply button of the OptionsDialog
may cause a rebuild (if the GUI language was changed) which tries to
delete the widgets inside the OptionsDialog, including the Apply button.
The theme label in the Misc tab will not change to the correct theme
when current language and theme is changed and 'apply' pressed.
loadNewTheme() does not do a rebuild of all widgets, including the
theme label, like it is explicitly done in the 'language section'.
The problem is that rebuild() uses the currently applied settings to
rebuild all widgets. Although a new theme was selected by the user the
label will be overwritten with the name of the still active theme.
By rearranging the logic a complete rebuild of the GUI is done and
updates the widgets correctly.
getThemeId() returned "builtin" for valid filenames because FSNode only
searches for the theme filename, like "scummmodern.zip" in the current
directory. listUsableThemes() searches SearchMan default directories
for theme files.
The width of each tab is now computed from its title, independently of
the other tabs. This increases the number of tabs that fit on the
screen.
This rewrite also fixes a bug where if the window size increased while
_firstVisibleTab > 0, some tabs would become inaccessible when the
scroll buttons were hidden.
The layout key Globals.TabWidget.Tab.Width is now treated as minimal
tab width. This is set so that the tabs fit reasonably well in lowres
layouts.
At the same time, this reduces the lowres scroll buttons heights to fit.
This patch makes the Nintento DS hacks in TabWidget obsolete.
(Hopefully! I'm not able to test.)
- control options tab is moved past graphics options tab
- control tab is not the default anymore
- changed label of "Mouse Speed" option to "Pointer Speed"
- changed Pointer Speed infotip description to make it more clear
Fixes erratic speeds in analog pointer motion
Implemented option to set analog/keyboard pointer speed
and control the analog joystick deadzone. The deadzone option appears
only if the build supports analog joystick (via JOY_ANALOG define)
GUIO_NOSPEECHVOLUME can be used for games that allow toggling of
speech but do not provide the ability for users to control speech
volume.
GUIO_LINKMUSICTOSFX and GUI_LINKSPEECHTOSFX can be used for games
that allow control of music, sfx, and speech in combinations, like
games that provide control of digital audio separately from MIDI,
or games that only control all three audio types through a single
volume control.
This fix is not totally clean as it spuriously uses the
serverLabelPosition variable in one case of the preprocessor ifdef
configuration, but with the current structure, this is a simple hack to
fix.
A better solution would be to rewrite some of these functions to remove
some of the preprocessor usage if possible.
Out-of-bounds values are always indicative of a bug somewhere else,
but at least not crashing here allows the user to recover by
interacting with the slider control. The error will still be
obvious because the associated text field will display the
original weird value.
Unlike other options in the OptionsDialog, the theme change was
done when selecting a new theme and not when clicking on OK or
Apply. This commit makes it consistent with other options
The issue could occur when adding or removing widgets to a tab, and then
not switching to a different tab before the destructor or reflowLayout() were
called. In such a case the firstWidget of the current widget in the _tabs list
could be out of date. Accessing this first widget from the destructor or from
reflowLayout() could then cause a crash, or random issues caused to access
to free'ed memory. In theory this could also lead to a memory leak, although
I don't think this could occur in our current code.
Usually we add several tabs to a TabWidget and then switch back to the first
tab after building all the tabs. So in such a case the issue would not occur.
But because we are deleting and reconstructing the clear buttons for the
MIDI and Path tabs of the options dialog from reflowLayout(), if the current
tab is the Path tab, it would be kept as active tab after adding and removing
widget to it and the issue would occur.
This fixes bug #9618.
A value of 0 is valid for tab ids, so the correct initialisation at this
stage is -1. However only one constructor properly initialized all the
tab ids to -1 in its initialisation list, but it was then changed to 0 in
init(). I have added the missing ones to the other constructors and
removed the incorrect ones in init(). But maybe all tab ids should be
initialised in init() rather than in the constructors initialisation lists.
The cloud manager registers itself as an event source as a mean to be polled
periodically by the GUI or engine code. The periodical polling is used to
update the OSD icon indicating background sync activity.
Also move the cloud icon from ConnectionManager to CloudManager,
allowing to decouple icon handling from network connections updates.
The connect button is enabled/disabled whenever the text in the
code widget is changed and it's validity assessed. But the initial
state was not correct.
- The checkbox state was not initialized when opening the dialog.
- The visibility state was initialized from ConfMan too early resuling
in the value being incorrect when multiple file browsers are used.
StorageWizardDialog now runs LocalWebserver in "minimal mode" for
security reasons. In this mode server uses only those handlers which
state to support it.
There are two handlers which support minimal mode: IndexPageHandler
(which handles `code` requests needed by StorageWizardDialog) and
ResourceHandler (which provides inner resources like `style.css` or
`logo.png` from `wwwroot.zip` archive).
Commit adds kFeatureClipboardSupport. hasTextInClipboard() and
getTextFromClipboard().
OSystem_SDL has this feature if SDL2 is used.
EditableWidget and StorageWizardDialog use g_system to access clipboard
now.
It pastes clipboard contents as code into 8 fields of that dialog.
(Clipboard support works with SDL2 only.)
"Open URL" and "Paste" buttons are placed in the left column under the
picture (because there is no room for 4 buttons in the bottom row).
Commit also adds "dropbox.bmp", which is just a square 115x115 picture.
Such pictures are would be used as Storages logos in that dialog.
In lowres there is no left column, so all 4 buttons are in the same row.
None of them are visible, because they are overflowed. Container has to
be added to continue working on them.
Added it into hasFeature() of all engines which returned `true` in
simpleSaveNames() before.
As mentioned in #788, SCI is not always using simple names, so it
doesn't have such feature now.
If it was completely clipped out (empty rectangle), it was drawing the
whole text ("empty means no clipping"), so I had to detect such cases
and change textArea to one small pixel.
It now hides code fields not just when built with SDL_Net, but also when
LocalWebserver's using default port.
So that's why NETWORKING_LOCALWEBSERVER_ENABLE_PORT_OVERRIDE is defined
in localwebserver.h now.
It's enabled only when NETWORKING_LOCALWEBSERVER_ENABLE_PORT_OVERRIDE is
defined.
It's not defined, because override means we have to reconfigure our
redirect links somehow to use the override port.
Now it respects outer code's decision to hide or move some widgets
around. Outer code must be CommandReceiver which is set as
ScrollContainer's target.
In most cases that's the right one to check. USE_CLOUD is defined when
either USE_LIBCURL or USE_SDL_NET are, which means if there is no curl,
USE_CLOUD still could be defined and linking errors would appear.
It now less empty, because if there is no download in progress, user
sees the RemoteBrowser instead of empty dialog. The cancel button is now
in the left bottom corner.
CloudManager's shortcuts are added too.
The idea is to keep FolderDownload request within Storage, and provide
necessary means to access it. The download is started and cancelled
through the DownloadDialog.
No wait when "Go up" is pressed. These contents could be invalid,
though. In order to refresh contents, one has to go up one more time and
then get back inside (in root folder - just press "Go up" to refresh
it).
The "comparison between signed and unsigned integer expressions" one.
Note that in UploadRequests size() and pos() are acutally signed,
because they could return -1. This commit implies that Requests are
working with such Streams which doesn't.
Now the code contains its own crc16 in it, plus the way checksum is
calculated has changed.
Some online tool calls this exact way of calculating crc16
"CRC16_CCITT_FALSE".
