Search code examples
pythonpython-3.xnumpyarcgisarcpy

Installing python packages without dependencies


I am trying to install a specific version of a python package into a pre-installed python environment. The package is laspy and the version is an old one (1.2.5). The package is supposed to work with Python version 2.7, but I am trying to install it against version 3.5, as I saw here that it should work on Python 3.4.

The real reason why I am doing so is that this specific Python is shipped with ArcGIS Pro, and I need the arcpy module which is present only in this installation.

I've been able to download the laspy package using pip download. This module depends on the module numpy, which is already present in the Python environment. This is causing the pip install to fail with error:

PermissionError: [WinError 5] Accesso negato: 'C:\\Program Files\\ArcGIS\\Pro\\bin\\Python\\Lib\\site-packages\\numpy'

which I kind of understand (it cannot overwrite the already installed numpy).

Here comes my big doubt: would installing laspy with pip and --no-dependencies option "break" my python installation?


Solution

  • So stupid... The error message PermissionError was just because I opened cmd without administrative privileges...

    Just installed laspy with pip install laspy==1.2.5. Hopefully it will work with this 64bit version of Python shipped with ArcGIS Pro (I was actually using it with the python 2.7 shipped with ArcGIS 10.x but it's 32bit and with LAS files it is easy to receive "out of memory" messages...).

    Will edit this answer to give some news on the compatibility.

    UPDATE

    Seems like I was just able to import laspy, but all other submodules of it didn't work...

    e.g. import laspy works 'from laspy.File import file' throws No module named 'laspy.File'.

    I am now switching to a fork (this one), which should be compatible hopefully.