Lifetime/lifecycle of animations and paths

I’m trying to figure out the lifetime/ownership/etc of the various parts of animations. In my case I have an animation I want to turn on and off depending on state. So I’m trying to figure out which variables I need to keep in my state and which I can allocate on the stack and then forget about.

Starting with something like:

lv_anim_t * a = lv_mem_alloc(...);
state->animation = a;

…I’d then set value params (start, end, etc.) that get passed by value anyway. But what about the path? I would normally assume that a struct needs to live as long as whatever takes it, so I’d allocate that on the heap too. However, when I look at the source for lv_anim.c I notice that it’s memcpy-ing the contents under the hood. So could I actually do:

lv_anim_path_t path;
// init, set callback
lv_anim_set_path(a, &path);

…and not worry about use of stack allocated memory for a heap allocated animation?

My second question is about stopping and restarting animations. There’s a start animation call lv_anim_start(), but to stop animations you delete them with lv_anim_del(). The latter function takes the variable and function but not the animation, because the underlying list only stores the first two things anyway. So does this mean I can reuse the animation variable and repeatedly invoke lv_anim_start()/lv_anim_del() without any other reconfiguration in between?

Take look at the implementation of lv_anim_set_path.

It saves the path in lv_anim_t. So finally the question is what happens with lv_anim_t

In lv_anim_start’s a node is inserted to the animations linked list and your lv_anim_t is copied there.

So lv_anim_t can be in the stack because it its data saved by LVGL.

You can also reuse lv_anim_t a after lv_anim_start(). E.g. To apply the same anim to an other object just call lv_anim_set_var() and lv_anim_start() again.