Added more notes on the context work

src/lib.rs

@@ -245,14 +245,28 @@ pub enum StyleVar {
 
 // --- Context -----------------------------------------------------------------------------------
 // TODO(4bb4) Do this properly.
+// I already added a simple Context struct that can be created once and used as long as it is not
+// dropped here for initial tests - this is of course neither threadsafe nor otherwise safe to use
+// unless one "does it right", so it's not a real solution.
+//
 // The context should have to be created, and ideally it should be difficult to impossible
 // to do things without having a context. implot-rs makes it so that there is a context and
 // that context has a "frame()" function that returns a Ui, and that Ui is then used to create
 // widgets. Windows are built with a build() function that takes a reference to that Ui as an
 // argument, but also have a begin() function that take a context and put it in their token.
 //
-// I think I'll mirror that here. Adding some simple start and end functions for initial testing
+// I think I'll mirror that here, except that we don't need a frame() function, it's enough
-// though.
+// to create a context once and then keep passing it around. I'll hence need a mutex and
+// a mechansim similar (or equal) to what imgui-rs does for making sure there can only be
+// a single context. Implementation could go roughly like this:
+//
+// - Add a mutex for modifying context things
+// - Make creation and drop functions use that mutex
+// - Change Plot, PlotLine, PlotScatter, PlotBars, PlotText to all require a context.
+//   I think I'll call this PlotUi to mimmick imgui-rs' Ui.
+// - Think about what this means in terms of the stacks and things like is_plot_hovered() -
+//   they should also only work when there is a context available.
+//
 pub struct Context {
     raw: *mut sys::ImPlotContext,
 }