Platform code should not generate key events with GLFW_REPEAT.
GLFW_PRESS is translated into GLFW_REPEAT by shared code based on the
key state cache.
This confused the automatic key release logic into not generating an
event with GLFW_RELEASE for a key being repeated when the window lost
input focus.
(cherry picked from commit 3f5dfeaf29)
We were previously storing the pointer position only when on the main
window, so when the user clicked on a fallback decoration it would use
the last position of the cursor on the main window, instead of the
position in the decoration surface.
Fixes part of #1991.
(cherry picked from commit 855d338a65)
The OpenBSD ports tree assigns its own soname version numbers, so the
hardcoded sonames GLFW uses to load libraries on non-macOS Unices are
often incorrect. Instead OpenBSD recommends that run-time loading
should leave out the version numbers entirely. The OpenBSD ld.so then
finds the correct library.
This upstreams the ports tree fixes for Xcursor and EGL, and adds the
corresponding fix for all other run-time loaded library sonames.
Tested on OpenBSD 7.0.
This issue was initially reported on IRC.
(cherry picked from commit 7d060ba4f1)
Corrects the protocol violation when creating an xdg_surface from a
wl_surface that already has a buffer due to EGL buffer swaps.
This commit is based on PR #1731 by @ghost, but adapted and altered:
- The XDG surface and role are now only created when a window is shown
to prevent application lists from showing command-line applications
with off-screen-only windows
- The special case of Wayland+EGL buffer swap is now in the EGL code
to mirror how X11 is handled
- Adaption to run-time platform selection and separate credits file
Fixes#1492Closes#1731
(cherry picked from commit 094aa6d3c7)
If a window lost input focus while a key was held down, the key repeat
mechanism would resume once the window regained focus.
(cherry picked from commit e24fe4b189)
This extensions allows GLFW to instruct the driver to ignore the alpha
bits, even in formats which contain them. This makes it possible to use
the alpha bits as extra storage, without it affecting the end result
getting displayed to the user.
Fixes#1434Fixes#1803
(cherry picked from commit 6281f498c8)
In the case the key repeat timerfd was interrupted before read(), the
cursor timerfd wasn’t read at all even when it could.
Related to #1711
(cherry picked from commit 68879081cb)
Only apply the content scale to the initial size of the window if
content scale retrieval succeeded.
Related to #1615.
(cherry picked from commit 53d86c64d7)
The monitor handle could have become invalid just before the call to
GetDpiForMonitor. It was possible for both window and monitor content
scale queries.
This ensures both that an appropriate error is emitted and that the
retrieved values are zero on error.
Fixes#1615
(cherry picked from commit fbfd7e65c8)
The conversion of window icon image data involves unsigned char color
values being promoted to int and then shifted to the left by 24. For
32-bit ints this is just far enough to trigger undefined behavior.
It worked by accident because of how current compilers translate this
piece of code.
This was caught by @slimsag while working on [Zig bindings for GLFW][1],
and diagnosed together with @Andoryuuta, as described [in an
article][2]. Zig has UBSan enabled by default, which caught this
undefined behavior.
[1]: https://github.com/hexops/mach-glfw
[2]: https://devlog.hexops.com/2021/perfecting-glfw-for-zig-and-finding-undefined-behavior#finding-lurking-undefined-behavior-in-6-year-old-glfw-code
Thanks to Maato, martinhath, dcousens, drfuchs and Validark for helping
to refine the solution.
This commit message was rewritten by @elmindreda to hopefully reflect
the conclusions of the pull request thread.
Related to hexops/mach#20
Closes#1986
(cherry picked from commit 9cd4d2fa20)
Joysticks already connected when GLFW was initalized did not get gamepad
mappings applied to them.
Regression introduced by 74a8ba26c3.
This change was backported without taking into account that 3.3.x does
not have on-demand joystick init.
Fixes#1996
Looking into the definition of kIOMainPortDefault, the following
description can be found:
When specifying a main port to IOKit functions, the NULL argument
indicates "use the default". This is a synonym for NULL, if you'd
rather use a named constant.
Thus, we do not have to utilize an external symbol for the identifier
of the default main IOKit port, but MACH_PORT_NULL suffice. This
simplifies compatibility between macOS versions as the symbol was
renamed with macOS 12.0.
Fixes#1985Closes#1994
(cherry picked from commit 544790666b)
This adds a workaround for kIOMasterPortDefault having been deprecated
in favor of kIOMainPortDefault in macOS 12.0.
Closes#1980
(cherry picked from commit f75c251dec)
The performance counter API is guaranteed to succeed on Windows XP and
later so there is no need for a fallback.
This removes our last dependency on winmm.
(cherry picked from commit b6834bf2a1)
The native access functions for context handles did not verify that the
context had been created with the same API the function was for.
This makes these functions emit GLFW_NO_WINDOW_CONTEXT on API mismatch.
(cherry picked from commit cca9008db2)
There is a suppression interval (0.25 seconds by default) after a call
to CGWarpMouseCursorPosition, during which local hardware events
(keyboard and mouse) are ignored. GLFW already calls
CGEventSourceSetLocalEventsSuppressionInterval with a value of 0.0, but
it doesn't help in this case, there is still a short delay before the
cursor can be moved. Moving the CGAssociateMouseAndMouseCursorPosition
call after the cursor position has been restored, fixes the issue.
Closes#1962
(cherry picked from commit 157ebb80aa)
The Wayland protocol spec[1] states that set_cursor must be called
with the serial number of the enter event. However, GLFW is passing in
the serial number of the latest received event, which does not meet the
protocol spec.
[1] https://wayland.freedesktop.org/docs/html/apa.html#protocol-spec-wl_pointer
As a result, set_cursor calls were simply ignored by the compositor.
This fix complies with the protocol more closely by specifically caching
the enter event serial, and using it for all set_cursor calls.
Fixes#1706Closes#1899
(cherry picked from commit e7758c506d)
According to the libxkbcommon documentation[1], xkb_keymap_key_repeats
requires keymap and keycode as input:
int xkb_keymap_key_repeats( struct xkb_keymap * keymap,
xkb_keycode_t key)
However, in inputChar in wl_input.c we are passing in xkb_keysym_t,
which was a type mismatch.
This results in some keys not repeating when they should and vice versa.
[1] https://xkbcommon.org/doc/current/group__components.html#ga9d7f998efeca98b3afc7c257bbac90a8Closes#1908.
(cherry picked from commit 216d5e8402)
The _glfwPlatformSetWindowFloating function would return without freeing
the state array if the window was already in the requested state.
(cherry picked from commit 071d7c0f46)
The outer glfwUpdateGamepadMappings function is now bypassed when
parsing the default gamepad mappings. The data in mappings.h does not
contain any comments or line breaks. Also this is done before joystick
support has been initialized, so there is no need to look for matching
devices.
Finally, the array of default mappings is pre-allocated. This has no
measurable performance impact but does generate a lot of calls, which
won't be nice for a user provided custom allocator to deal with.
(cherry picked from commit 201400b974)
This fixes gamepad mappings being included in GLFW binaries that would
then never use them because they were for a different platform.
(cherry picked from commit 0c90e1a33f)
This avoids generating a macro redefinition warning when the
GL_SILENCE_DEPRECATION macro has been defined for the whole of a larger
project.
Closes#1840.
(cherry picked from commit 1d88a2e3d6)
The contents scale of the hosted CAMetalLayer created for MoltenVK was
updated only after the GLFW content scale and framebuffer size events
were emitted, causing the layer to get out of sync with the monitor the
window was on.
(cherry picked from commit 076bfd55be)
Because there are controllers in the wild using the same hardware ID
despite having different numbers of buttons and axes, an error message
was triggered that was only expected for a corrupted mapping database.
This removes the error for now, in preparation for better error handling
for gamepad mappings overall.
Fixes#1763.
(cherry picked from commit 6876cf8d7e)
On FreeBSD O_CLOEXEC is only available when _POSIX_C_SOURCE >= 200809.
O_CLOEXEC is in turn required by the epollshim header.
Issue reported on IRC.
(cherry picked from commit a89fcd20d8)
The EGL backend ignored the state of GLFW_DOUBLEBUFFER and always
created a double-buffered EGL window.
This sets the EGL_RENDER_BUFFER attribute at EGL window creation
appropriately.
Fixes#1843.
(cherry picked from commit 114776a246)
There is no need to enumerate framebuffer configs that will just be
filtered out later by the GLFW_DOUBLEBUFFER hard constraint.
(cherry picked from commit fd79b02840)
This skips the buffer swap after the initial glClear performed during
window creation, if the window is single-buffered. This call confused
apitrace into thinking the window was double-buffered.
Fixes#1873.
(cherry picked from commit 184377b493)
This adds the state part of 04f21abb52
(Make GLFW_DOUBLEBUFFER a window attribute), but without the associated
API change, because this is needed by later bug fixes that will go into
the stable branch.
The /clang: suffix passed to Clang-CL was accidentally also passed to
the regular standalone Clang, which caused compilation to fail. We now
pass /W3 to Clang-CL, which it interprets as -Wall.
The _CRT_SECURE_NO_WARNINGS macro is now defined for both Clang and
Clang-CL.
The /entry: flag passed to link.exe is now also passed to lld-link,
letting the windows subsystem tests and examples link.
Fixes#1807.
Closes#1824.
Closes#1874.
(cherry picked from commit 061a0263a9)
The GPU driver only looks in the executable for the symbol requesting
the high-performance GPU, so enabling them when buidling GLFW as a DLL
will have no effect.
(cherry picked from commit 52ba8c7f07)
This removes the dependency on the (unspecified) ordering of geometry
and mode events in wl_output.
Based on feedback from @linkmauve and @caramelli.
Related to #1792.
(cherry picked from commit b925a54ef1)
GLFW_SCALE_TO_MONITOR had no effect on Windows 8.1 up to and including
Windows 10 version 1607 (Anniversary Update), despite those having
support for per-monitor DPI.
That done was to avoid handling systems that have non-client scaling,
introduced in Windows 10 version 1607, without reliable overriding of
the new window size, introduced in Windows 10 version 1703 (Creators
Update). Both are needed to keep the content area at a fixed size for
windows that have GLFW_SCALE_TO_MONITOR disabled.
This change enables window rescaling on Windows 8.1 and all later
versions but disables non-client scaling for unscaled windows on Windows
10 version 1607. Versions after 1607 are unaffected.
Fixes#1511.
(cherry picked from commit 729c9988d0)
The install name was incorrectly set to a relative path.
This change leaves the install name of the installed dylib as
@rpath/soname. Those who wish to override this can set the
CMAKE_INSTALL_NAME_DIR variable.
Closes#1504.
(cherry picked from commit 384ff74a46)
On Apple Silicon, IOKit is deprecated and there will be no
matching io_service that we can query for name. Luckilly,
NSScreen got an API to fetch the display name in 10.15.
This is a blocker to get glfw running on Apple Silicon.
Fixes#1809.
Closes#1833.
(cherry picked from commit 2bc52ca82e)
NSApp setActivationPolicy: was being called too soon when the app was
not bundled and launched from the command line.
This fix is based on #1802 by @richardwilkes.
Fixes#1648.
Closes#1802.
Adapted to 3.3-stable from 8b11867464.
On Wayland we implement Client-Side Decorations if the compositors do not
implement SSDs. In that case, the destructors of the surfaces were called
in the wrong order, leading to a dereference of an already freed object.
We need to first destroy the subsurface before destroying the parent surface.
Related PR on kitty: https://github.com/kovidgoyal/kitty/pull/3066
Related issue on kitty: https://github.com/kovidgoyal/kitty/issues/3051Closes#1798.
(cherry picked from commit 0dc1005c85)
CMake 3.19 adds -xc when the LANGUAGE file property is C, breaking our
workaround for CMake 3.15 and earlier not understanding the .m suffix.
Fixes#1787.
(cherry picked from commit 3327050ca6)
This mirrors the filtering done on X11 and Cocoa. Possibly this should
be done by shared code instead.
Fixes#1610.
(cherry picked from commit 0bccc3852b)
/W3 is the default for new VS projects and the library builds cleanly
with it on VS 2010-2019 so let's try to keep it that way.
(cherry picked from commit 6b78419c9a)
Clang ignores this when C99 is enabled, while GCC does not. Removing
the checks for Clang to hopefully avoid a false sense of security.
(cherry picked from commit 9b140d08d7)
An incorrect value for _WIN32_WINNT_WINBLUE caused GLFW to think Windows
8 was 8.1 or greater. This led to a crash when attempting to call
SetProcessDpiAwareness.
Closes#1775.
(cherry picked from commit e1d71cfa4701250a44a6c02d1bbb43309ac8a46c)
On Windows 7, when GLFW framebuffer transparency and the DWM are enabled
but DWM transparency is disabled (i.e. when the Transparency setting is
disabled under Personalization > Color), the contents of the framebuffer
is combined with the last frame using additive blending instead of
replacing the previous contents.
This commit limits GLFW framebuffer transparency on Windows 7 to when
DWM transparency is enabled, removing the previous workaround of setting
a layered window color key that led to rendering artifacts.
Fixes#1512.
(cherry picked from commit 05dd2fa298)
The window decoration window can go away between calls to XQueryPointer,
for example if the GLFW_DECORATED window attribute was just cleared.
(cherry picked from commit ceb20c7f97)
Background: The IM will filter out key events, instead sending exact
duplicate events that are not filtered. It does not send these for
every event, however, so the duplicate events cannot be relied on for
key input. Instead we need to identify and discard them. Since they
are identical, they have the same timestamp as the originals.
The previous duplicate event detection would consume unrelated key
events if the keys were pressed simultaneously, as it only tracked
a single timestamp.
This fixes that issue for any combination of keys, at the expense of
a 1 KB array per GLFW window.
This fix is a stopgap until explicit IME support is done.
Based on #1472 by @LucaRood.
Fixes#1112.
Fixes#1415.
Fixes#1616.
Fixes#1663.
Closes#1472.
(cherry picked from commit 9a3664b4a9)
Supplementary Plane codepoints from WM_CHAR and WM_SYSCHAR messages were
reported as UTF-16 surrogate pairs.
Related to #1635.
(cherry picked from commit 6ce2070392)
This fixes a race between the key repeat logic and the surface leave
event handler, which could result in repeated keys being reported with
a window of NULL.
Fixes#1704.
(cherry picked from commit c72da994ba)
These are harmless errors but the code was worth fixing just to reduce
confusion and be more explicit. E.g. using a different variable name
for a new variable of a different type in win32_joystick.c.
Closes#1700.
(cherry picked from commit a84a30ab63)
The `atomCount` variable has the type `unsigned long`,
so the `for` loop iterating over it should use the same type.
Closes#1701.
(cherry picked from commit 250b94cd03)
When compiling with `-Wall` and `-pedantic-errors`, gcc complains with
```
warning: ISO C does not support the '%m' gnu_printf format [-Wformat=]
```
because the `%m` conversion specifier is a GNU extension.
Closes#1702.
(cherry picked from commit d4f5074535)
The only two EGL implementations on macOS are Swiftshader and ANGLE.
While Swiftshader supports both `NSView` and `CALayer` as
`EGLNativeWindowType`, ANGLE supports only `CALayer`. Furthermore
Swiftshader's OpenGL ES frontend is deprecated in favor of using ANGLE's
Vulkan backend on top of Swiftshader's Vulkan frontend.
This means that on macOS `EGLNativeWindowType` should be a `CALayer` for
compatibility with ANGLE.
Fixes#1169.
Closes#1680.
(cherry picked from commit 91eebe922d)
glfwSwapBuffers may be called by any thread but NSWindow may not.
Bug introduced by c3ca88055f and reported
by @crujose.
(cherry picked from commit 949275bbed)
A regression introduced by b889aa7841
broke the special handling of numpad keys for the non-XKB fallback path.
The non-functional remains were later removed. This restores the
original behavior.
(cherry picked from commit cab41529da)
This replaces the hardcoded keycode ranges and various kludgy range
checks with the actual ranges reported by Xlib and XKB.
(cherry picked from commit ee45b58647)
This fixes the issue where function keys would be reported as
GLFW_KEY_UNKNOWN if XKB was available and one of the configured keyboard
layouts was Arabic.
This is only part of #1598, because the full patch removed parts of the
fallback path for when XKB is unavailable.
Closes#1598.
(cherry picked from commit a41a58a95e)
The non-root parent window owned by the WM could be destroyed before we
process the ConfigureNotify event using the cached parent handle.
Bug was found by unmapping a decorated window.
This like all uses of the Xlib error handler is not thread safe and
there is nothing we can do about that short of moving to XCB.
Fixes#1633.
(cherry picked from commit e65de2941c)
Some synthetic key messages come with a scancode of zero, causing them
to be translate to GLFW_KEY_UNKNOWN. This fills in the missing scancode
by translating the provided virtual key.
Rather than further complicate a single-use function, its body is merged
into the key message handler.
Fixes#1623.
(cherry picked from commit 789a2bcb39)
The Win+V hotkey brings up a clipboard history IME that consumes the key
release. This adds left and right Super to the modifier keys manually
polled for undetected release during event processing.
Fixes#1622.
(cherry picked from commit 562c17d131)
If the application is not linked against the Vulkan loader and relies on
a third-party loader library or glfwGetInstanceProcAddress, then our
call to dlopen will fail without a custom dyld environment variable.
This adds a fallback of looking in the directory of the main executable,
which matches the bundle structure recommended by the Vulkan SDK, making
that finally work out of the box for fully dynamic loading.
(cherry picked from commit 7da87aaae7)
This should have been initialized to the screen root, not None. This
issue was introduced by fe57e3c292.
Fixes#1620.
(cherry picked from commit 9372ba95fa)
When the WM does not support EWMH or there is no WM running, GLFW falls
back to XSetInputFocus, which will emit BadMatch if the window is not
viewable, which will terminate the program.
Bug spotted on IRC.
(cherry picked from commit aa5e313561)
A window resize action that also resulting in the window being moved did
not emit any window positions events, as the position of real
ConfigureNotify events was ignored. The real events use parent
coordinates instead of root coordinates so this adds parent tracking and
conditional translation.
Fixes#1613.
(cherry picked from commit fe57e3c292)
The `sizeof()` operator has the type `size_t`, so the `for` loop
iterating over it should use the same type.
Closes#1614.
(cherry picked from commit 74a46dfa0c)
This removes the final dependency on CoreVideo, using a display link to
get the refresh rate of monitors where Core Graphics report a refresh
rate of zero. Instead we now query the I/O registry directly, similarly
to what the display link does at creation.
Thanks to @OneSadCookie for pointers to this solution.
(cherry picked from commit 4ec7daf3e9)
The display link query is not specific to any particular display mode
and so only needs to be done once.
The next step is to replace the display link altogether by querying
IOKit directly, which is what the display link does.
(cherry picked from commit b4a8eb9b19)
This fixes glfwMaximizeWindow having no effect on hidden windows by
manually appending the maximization states to the EWMH state property.
(cherry picked from commit 4837b78ffe)
Polling the event queue before NSApp had been allowed to finish
launching, in our case by starting our self-terminating run loop,
triggered an assertion inside NSApplication.
This fix, which makes all event processing functions capable of starting
it, makes that assertion less likely.
A more Cocoa-friendly fix would be to finish launching NSApp during
glfwInit and let people annoyed by the menu bar disabled it with
GLFW_COCOA_MENUBAR. That may not be suitable for 3.3-stable, though.
Fixes#1543.
(cherry picked from commit 6e6805000a)
It would previously conserve the last name it had before leaving the
border, sometimes desynchronising with what it should have been.
(cherry picked from commit ef6189f348)
That way the application only sees the cursor moving when it is inside
of its area, it won’t go back to the top or left side when trying to
resize the window or just hovering the fallback decorations.
(cherry picked from commit a80788c17f)
Previously, any pointer motion in the window decorations when using the
fallback implementation would obtain the wl_cursor again, and do the
attach danse for no benefit.
This will ultimately allow animated cursors to not reset to the first
frame on motion, once these will be implemented.
(cherry picked from commit a9f674e719)
This falls back to calculating the monitor physical size from the
current resolution and the default X11 DPI when the physical size
returned by RandR is zero.
(cherry picked from commit e96dc5d219)
Iconification (miniaturization) of undecorated windows stopped working
unless the window has the NSWindowStyleMaskMiniaturizable style.
(cherry picked from commit 2c519709be)
On macOS a destroyed window remained on screen until the next time
events were processed. This makes the behavior more consistent with
other platforms.
Fixes#1412.
(cherry picked from commit c819f27ce3)
The SDL2 2.0.5+ controller GUID 03000000790000000600000000000000 matches
many devices with different layouts and element counts but with the same
chipset. This issue is still being resolved upstream. In the meantime
this removes those mappings from GLFW to avoid confusion and errors.
SDL upstream issue: https://bugzilla.libsdl.org/show_bug.cgi?id=4545
Related to #1583.
(cherry picked from commit b69fb99031)
This is a temporary fix while we determine if the color key workaround
should be removed completely. See issue thread for discussion.
Related to #1512.
(cherry picked from commit c88ee1c9d3)
KDE sometimes removes the Xft.dpi resource when it would be set to the
X11 default value of 96, causing GLFW to fall back to a value calculated
from the core display sizes in pixels and mm in a desktop environment
that supports Xft.dpi.
This moves to a hardcoded fallback value of 96 on the assumption that
there are more people running KDE with 96 DPI than there are people
running desktop environments that do not support Xft.dpi.
All of this is terrible please send help.
Fixes#1578.
(cherry picked from commit 75294462b3)
This fixes the enabling of window decorations after creation. Instead
of removing the _MOTIF_WM_HINTS property, we now set or unset the
MWM_DECOR_ALL bit of the decorations field.
Fixes#1566.
(cherry picked from commit 5fc4c01302)
Replaces `VERSION_GREATER_EQUAL` with `VERSION_EQUAL OR
VERSION_GREATER`. `VERSION_GREATER_EQUAL` was only added in CMake 3.7.
Fixes#1584.
Closes#1585.
(cherry picked from commit 4d0ae4ffa7)
The `monitorCount` member in the `_GLFWlibrary` struct is of type `int`, so the `for` loop iterating over it should also use the type `int`.
Closes#1572.
(cherry picked from commit 7749aac50d)
This enables compilation as C99 where supported by the compiler.
A workaround with per-compiler hardcoded flags is used for CMake 3.0,
which does not support the C_STANDARD target property.
Fixes#1560.
Closes#1576.
(cherry picked from commit 8f852e0833)
The `name` member in the `_GLFWmonitorWayland` struct is used in two places. It is assigned the value from a variable of type `uint32_t` and is compared to another variable of type `uint32_t`, so `name` should also have the same type.
Closes#1569.
(cherry picked from commit d25248343e)
This replaces some magic numbers with the corresponding winuser.h
provided macros and unifies how the MSB from Get*KeyState is tested.
(cherry picked from commit 3d2540c373)
The `size` member in the `GLFWgammaramp` struct is of type `unsigned int`, so the `for` loop iterating over it should also use the type `unsigned int`.
Closes#1541.
(cherry picked from commit 243b1bc292)
This makes glfwSetWindowSize use the top-left corner as anchor point
instead of the bottom-left corner.
Fixes#1553.
(cherry picked from commit 7f02898264)
This fixes the bug of video modes being discarded if they had
a different resolution but the same area as another mode.
Fixes#1555.
Closes#1556.
(cherry picked from commit 2777f6a754)
The window rect adjustment for content scale broke the initial, correct
maximization performed when creating the window with WS_MAXIMIZE. This
switches to updating the restored rect instead of the current rect.
Fixes#1499.
Closes#1503.
(cherry picked from commit 711b9694a1)
When using GLFW with CMake and installed GLFW binaries, `-DGLFW_DLL` is
passed on Linux, which should not happen.
Closes#1530.
(cherry picked from commit d7fb01750ece465cfaae0493dfe3003da1e5f500)
For users with multiple keyboard layouts configured, glfwGetKeyName
works fine only with the primary layout. Switching layouts results in
changing the group index. This commit querries the current group index
when initializing keyboard input and keeps track of any change to it.
As a result the scancode -> keyname mapping may change while the program
is running (needs to be documented).
Fixes#1462.
Closes#1528.
(cherry picked from commit 36f90800d8)
MSVC 2019 complains that the code at line 1744
(`GetMonitorInfo(window->monitor->win32.handle, &mi);`) can potentially
dereference a null pointer. The compiler is wrong in this case (it has
not spotted that `monitor` and `window->monitor` must be equal), but
I think it makes sense for our non-NULL test to be on the variable we
actually use rather than the one it was set from.
Related to #1491.
(cherry picked from commit 1d62157268)
The EWMH feature detection atoms are now named and loaded the same way
as other X11 atoms. Detection is now performed after all
non-conditional atoms have been loaded. The EWMH detection now has
hopefully more readable comments.
(cherry picked from commit fad9896d38)
This should have been done when the WGL extension members were moved
from the context struct to the library struct.
(cherry picked from commit 22a6c02a4c)
