GetMediaItemInfo_Value API does not return the real values

[amagalma]

and the test code:

item = reaper.GetSelectedMediaItem(0,1)
previousitem = reaper.GetSelectedMediaItem(0,0)
secondstart = reaper.GetMediaItemInfo_Value(item, "D_POSITION")
firstend = reaper.GetMediaItemInfo_Value(previousitem, "D_POSITION") + reaper.GetMediaItemInfo_Value(previousitem, "D_LENGTH")
reaper.ShowConsoleMsg("end of first item is "..firstend.."\n")
reaper.ShowConsoleMsg("start of second item is "..secondstart.."\n")
if firstend == secondstart then
  reaper.ShowConsoleMsg("They are equal")
else
  reaper.ShowConsoleMsg("They are not equal")
end

Either, LUA doesn't do its math well, or the Reaper API does not return the real values.

Reaper v5.40 x86 , Windows 7 x64

[juliansader]

ShowConsoleMsg (and the IDE) rounds the numbers to 11 decimals.

Under the hood, however, REAPER's actual time values have something like 15 decimals.

AFAIK, the last decimals are not entirely reliable, and stuff that are supposed to be on exactly the same grid position can sometimes differ by a few quadrillionths of the second.

[amagalma]

This is confusing.. So, how can one know what are the real values since both the console and the IDE round them? And how one can do maths, if the maths you do are applied to different values and not to the ones you thought?...

[schwa]

reaper.ShowConsoleMsg(string.format("%.16f",firstend))
reaper.ShowConsoleMsg(string.format("%.16f",secondstart))

should show that the numbers are different (probably because the start of the second is snapped to sample).

[Xenakios]

Originally Posted by amagalma

how one can do maths, if the maths you do are applied to different values and not to the ones you thought?...

For starters, don't equality compare floating point numbers :

if firstend == secondstart

It's a pretty rare occasion when you can truly expect 2 floats to be equal if they come from some source that is not entirely under your control.

You can compare their absolute difference to some value though :

local a = 0.000000001
local b = 0.000000001007
if math.abs(a-b)<0.00000000001 then print("approx equal") 
else print("not equal") end

It is left up to you to decide what is that acceptable difference...

[amagalma]

Thanks to all of you for the answers! So, it appears it's not a bug, but a whimsy when doing maths with floating numbers. I get it! Thanks to all of you for the lesson!