Shift of Z values in writeLAS output

[floriandeboissieu]

Hi Jean-Romain,
I noticed a 0.01 shift of certain Z values when writing to las/laz, there seem to be a bug there. It seems to be related with some particular Z values, see code below.
Here is how I could reproduce the bug:

LASfile <- system.file("extdata", "Megaplot.laz", package="lidR")
lidar = readLAS(LASfile)
writeLAS(lidar,"/Megaplot.cpy.laz")
365868 bytes written
Column pulseID skipped. It does not meet las specifications.
lidar1=readLAS("/Megaplot.cpy.laz")
any(lidar$Z-lidar1$Z>0)
[1] TRUE
(lidar$Z-lidar1$Z)[lidar$Z-lidar1$Z>0][1:10]
[1] 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01

writeLAS(lidar1,"/Megaplot.cpy.cpy.laz")
365868 bytes written
Column pulseID skipped. It does not meet las specifications.
lidar2=readLAS("/Megaplot.cpy.cpy.laz")
any(lidar2$Z-lidar1$Z>0)
[1] FALSE

[floriandeboissieu]

Easier to copy paste like this maybe

LASfile <- system.file("extdata", "Megaplot.laz", package="lidR")
lidar = readLAS(LASfile)
writeLAS(lidar,"~/Megaplot.cpy.laz")
lidar1=readLAS("~/Megaplot.cpy.laz")
any(lidar$Z-lidar1$Z>0)
(lidar$Z-lidar1$Z)[lidar$Z-lidar1$Z>0][1:10]
writeLAS(lidar1,"~/Megaplot.cpy.cpy.laz")
lidar2=readLAS("~/Megaplot.cpy.cpy.laz")
any(lidar2$Z-lidar1$Z>0)

[Jean-Romain]

Try that one. It's fine on my computer (devel branch). It was a problem of computer precision