ValueError: cannot convert float to NaN to integer

[gisxperts]

osrm responds with integer values in table, leads on windows (python 2.7.10, numpy 1.9.2, download python_osrm last night) to the error message 'ValueError: cannot convert float to NaN to integer ' for line 201.

my quick and dirty workaround was mat[mat == 3579139.4] = -9999, but better would be to have the array always created as float, or check what type of array is created and depending on the outcome assign proper NaN.

[ustroetz]

Hi @gisxperts, you could add a little bit of code showing what you tried to do, so I can try to reproduce it?

[gisxperts]

list_coord = [[21.0566163803209, 42.004088575972],[21.3856064050746, 42.0094518118189],[20.9574645547597, 41.5286973392856],[21.1477394809847, 41.0691482795275],[21.5506463080973, 41.3532256406286]] list_id = ['name1', 'name2', 'name3', 'name4', 'name5'] time_matrix = osrm.table(list_coord, list_id, output='dataframe', host='http://localhost:5000')

[ustroetz]

@mthh could you maybe have a look at this?

[mthh]

@gisxperts @ustroetz my bad, it was only tested on python 3.4. As you suggested @gisxperts, fixing the type of the array at its creation do the job :

>>> platform.python_version()
'2.7.9'
>>> list_coord = [[40, -40],[21.3856064050746, 42.0094518118189],[20.9574645547597, 41.5286973392856],[21.1477394809847, 41.0691482795275],[21.5506463080973, 41.3532256406286]]
>>> list_id = ['name1', 'name2', 'name3', 'name4', 'name5']>>> time_matrix = osrm.table(list_coord, list_id, output='dataframe', host='http://localhost:5000')
>>> time_matrix
       name1  name2  name3  name4  name5
name1    NaN    NaN    NaN    NaN    NaN
name2    NaN    0.0   88.1  149.4  106.3
name3    NaN   88.6    0.0  100.0   65.6
name4    NaN  137.6   99.8    0.0   49.4
name5    NaN   94.6   65.6   48.8    0.0