manderby / nalib Goto Github PK

Having a button without an image is certainly useful during initialization.
At the same time, review the isImage flag if there is a better solution.
Have a look at #32.

No matter whether push or state button, the clickable region is not according to standard.

So far, this has not been tackeled as it seems to be very undocumented and maybe even impossible. Best solution is to create a plain Cocoa Application with an automatically generated xib file and then start NAApplication in the main method.

Would be nice if NALib would be able to become stand-alone.

Do issue #12 first to have a valid 3D rendering environment.

When a window is set to use a different mode than the system settings, it outputs a warning and chooses the one of the system.

Either find a way to get the proper mode and get the proper color or define default colors which are valid in such a case where the two modes mismatch.

After the postStartup, the NSApplication sets its application delegate in naStartApplication to the default. Desired would be to the delegate class of NALib to capture more notifications. This interfered with some other mechanisms though (don't quite remember now).

Please have a look at the following comments removed from naStartApplication to get back in the mood:

  //[NSApp setDelegate:(NA_COCOA_BRIDGE id)naGetUIElementNativePtr(app)];
  
//  SEL selector = @selector(doSomethingWithObject:afterDelay:);
//  IMP newImp = (IMP)swizzledDoSometingWithObjectAfterDelay;
//  Method method = class_getClassMethod([MyClass class], selector);
//  const char * encoding = method_getTypeEncoding(method);
//  class_replaceMethod([MyClass class], selector, newIMP, encoding);

Contains issue #3, #10

Make Mac and Win equal.

These are some comments removed from the code. They might help:

  // OptionButton: NSBezelStyleShadowlessSquare
  // NSBezelStyleRegularSquare : 5 5 5 5
  // NSBezelStyleShadowlessSquare : 3 3 3 3
  // NSBezelStyleSmallSquare : 2 1 2 1

// Push (Text only) (24px height fixed)
// Option (Text / Image) (3px padding on all sides)
// Borderless (Image only) (0px padding on all sides)

Should ned to test NATree first #42.

Reiterate all NAStack benchmarks and delete all for loops which are not necessary.

Add functionality to NAStack
Add possibility to explicitely store pointers.
Add possibility to define if pool initially is filled or empty

Enhance the GUI Hello World application with a playground window containing all possible GUI elements and behaviours.

When adding a command line argument to the testing, the application is expected to execute only the give group. But all groups after the given one are executed as well. Probably duplicate of #14

Run HelloWorldGUI and select the text in one of the labels. Drag the mouse outside of the window. Crash.

When using a space with an alternate background it is erased with the wrong brush color for Labels and other UI elements.

Store it as a flag. Implement the flag together with issue #31
Can we make sure only one enter or cancel button per window exists?

Loading files results in source apparently has no cache, althoug it loads well.

Leap seconds are about to be removed from the UTC standard. Find information and provide a proper implementation.

Do it using the default.

Probably linked to #25.

Very low priority task. Following removed comments might help:

//    [event modifierFlags]; NSEventModifierFlagCapsLock;
//    let isLeftShift = event.modifierFlags.rawValue & UInt(NX_DEVICELSHIFTKEYMASK) != 0
//    let isRightShift = event.modifierFlags.rawValue & UInt(NX_DEVICERSHIFTKEYMASK) != 0

Note that distinguishing might not be possible depending on the system.

In the application Color Machine, the UI is set up using layout helper functions. Add those to NALib.

This is in response to #11 because Apple fucked up.

During development, lots of code was necessary to create the arguments of the procedure call in na_ExecuteCrashProcess.
Try to store the expressions in the NATestData as NAString and try to move sysctl.h and Tlhelp32.h stuff into a separate module.

NAStack done

When opening a file with an NABuffer, the following error is outputted:

Error in na_GetBufferSourceCache: source has no cache

Some sort of caching all of a buffer is required but it should not happen with naCacheBufferRange. The NABuffer should be self-updating.

Also have a look at other occurences in naWriteBufferBuffer, naGetStringUTF8Pointer, naNewStringWithBasenameOfPath, na_AllocPNGChunkFromBuffer

Note to Manderby: The issue can be reproduced by running Karo Graph without CACHING_ALWAYS and opening a file by double clicking.

Done because the current PNG implementation is a little slow. And of course on windows, anything else than PNG is not yet supported.

Fix would be to improve the PNG implementation

See HelloWorldGUI, watch the dot for some seconds, the framerate drops constantly. Potentially due to SetTimer and KillTimer which gets calld by naCallApplicationFunctionInSeconds

Cut, Copy, Paste do not work currently without a XIB.

Example:
Open NALibVisualStudio, Run test with arguments:
NALib NACore NAValueHelper Flags "Toggling int flags"
Probably duplicate of #24

When running benchmarks (for example NAStack Push), a lot of memory is required. This may exceed the capabilities, especially on 32 bit systems. Find a suitable solution to capture this or break early.

Add NIST CODATA 2022 values as soon as they are public.

See https://physics.nist.gov/cuu/Constants/index.html

There is the possiblilty to do things with plain C, see the code sniplet below which is an old comment (completely out of context, but this is just a hint where to start looking).

Idea is to move at least the NAPreferences out of the GUI code.

  // The ((id (*)(id, SEL)) part is a cast of the objc_msgSend function which
  // is requires since a later version of Objective-C
//  ( (id (*)(id, SEL)) objc_msgSend)(objc_getClass("NSApplication"), sel_registerName("sharedApplication"));

Deallocation using naRelease has this built in because of the NARefCount which is checked upon calling naRelease or naRetain. But naDelete does not have the storage capacity to do that.

Would need an additional dummy-space similar to the refcount but which is only available whe NA_DEBUG is 1.

When doing this, think about adding this functionality to all structs which use naClearXXX. Maybe do something with NAPtr

na_GetUINALibEquivalent is currently done using a list traversal which might be slow for very big UIs. It had beed tried to be done using the GWLP_USERDATA of HWND on windows and with the tag of NSView on macOS but that proved to be a maintenance catastrophy.

Optimal solution would be a a hash map. Maybe find something else.

Resizing the window the space is contained in helps to synchronize the vsync or something. Not quite sure why.