Fix Time.get_unix_time_from_system() not including msecs

[timothyqiu]

Fixes #64059

OS::get_unix_time() is still uint64_t on 3.x. So a dedicated OS::get_subsecond_unix_time() is added.

This new function is intended to be used only by Time so it's not exposed to scripting.

[akien-mga]

I would also document the difference between OS.get_unix_time() and Time.get_unix_time_from_system() because it's confusing that they don't give the same result.

[timothyqiu]

Added a note for OS.get_unix_time().

Time.get_unix_time_from_system() already has a note about it being different from other integer timestamps.

[akien-mga]

Thanks!

[akien-mga]

Cherry-picked for 3.5.1.

[RPicster]

I think this is a pretty bad change to be honest. It's a breaking change and it would have deserved an own function.
We were now searching for a new bug since changing to 3.5.1 until we found out about this.

[timothyqiu]

The documentation of Time.get_unix_time_from_system() is not changed. Although the fraction part is zero, the value returned is also a float in 3.5. So I think this is a bugfix of the implementation rather than a change in interface.

[azagaya]

The documentation of Time.get_unix_time_from_system() is not changed. Although the fraction part is zero, the value returned is also a float in 3.5. So I think this is a bugfix of the implementation rather than a change in interface.

In 3.5 this code:

func _ready():
	var timestamp = Time.get_unix_time_from_system()
	print(timestamp)

returns this:

1666621969

In 3.5.1:

1666622043.819422

Both of them are floats as you mention. As we were converting to string to use inside custom expressions, and then parsing it, that "." was causing bugs that werent in 3.5

[timothyqiu]

When converting a float to string, how to handle the fraction part should always be considered I think.

Since the documentation in 3.5 mentioned that the return value is in seconds and explained it's float for sub-second precision, I believe the bug in 3.5 hide the bug in your code, and fixing the engine bug exposed the bug in your code rather than causing it.

[azagaya]

yeah you are right. Thanks for taking time to answer!