Previously, using rio-tiler and referring to rio-viz, I made a small application that can provide WTMS services for multiple images simultaneously. Some time ago I needed to deploy it on a Windows server, ran into a few problems, and wrote them down here.
rio-tiler depends on rasterio, rasterio depends on gdal, and installing these on Windows is a bit tricky.
Directly installing rio-tiler with pip will get stuck and throw errors when installing gdal and rasterio.
At first, I found the corresponding versions of GDAL and rasterio on Unofficial Windows Binaries for Python Extension Packages. Even after installing them, there were still some issues, such as rasterio throwing an error that pyproj’s CRS definitions could not be found when imported.
Initially I used conda’s default channel. In rio-tiler=2.0.0rc3, the dependency rasterio ≥ 1.1.7 is required, but that version wasn’t available in the default channel. After switching to the conda-forge channel, I found the versions were much more complete.
First, I used miniconda3, which is smaller in size and straightforward to install. Then I created and activated a new environment.
conda create -n rasterio python=3.7conda activate rasterioconda search rasterio -c conda-forgeLoading channels: done# Name Version Build Channelrasterio 0.36.0 py27h43d01bd_1 pkgs/mainrasterio 0.36.0 py35h49e1f75_1 pkgs/mainrasterio 0.36.0 py36hb8ea33a_1 pkgs/mainrasterio 1.0.13 py36h6bd7d87_0 pkgs/mainrasterio 1.0.13 py37h6bd7d87_0 pkgs/mainrasterio 1.0.18 py36h6bd7d87_0 pkgs/mainrasterio 1.0.18 py37h6bd7d87_0 pkgs/mainrasterio 1.0.21 py36h6bd7d87_0 pkgs/mainrasterio 1.0.21 py37h6bd7d87_0 pkgs/mainrasterio 1.0.23 py36hbf02ebe_0 conda-forgerasterio 1.0.24 py36h163c445_0 conda-forgerasterio 1.0.24 py37h163c445_0 conda-forgerasterio 1.0.25 py36h163c445_0 conda-forgerasterio 1.0.25 py37h163c445_0 conda-forgerasterio 1.0.25 py37he350917_1 conda-forgerasterio 1.0.26 py36h163c445_1 conda-forgerasterio 1.0.26 py37h163c445_1 conda-forgerasterio 1.0.27 py36h163c445_0 conda-forgerasterio 1.0.27 py37h163c445_0 conda-forgerasterio 1.0.28 py36h163c445_0 conda-forgerasterio 1.0.28 py36h163c445_1 conda-forgerasterio 1.0.28 py36h2617b1b_2 conda-forgerasterio 1.0.28 py37h163c445_0 conda-forgerasterio 1.0.28 py37h163c445_1 conda-forgerasterio 1.0.28 py37h2617b1b_2 conda-forgerasterio 1.1.0 py36h039b02d_0 pkgs/mainrasterio 1.1.0 py36h2617b1b_0 conda-forgerasterio 1.1.0 py37h039b02d_0 pkgs/mainrasterio 1.1.0 py37h2617b1b_0 conda-forgerasterio 1.1.1 py36h163c445_0 conda-forgerasterio 1.1.1 py36h2617b1b_0 conda-forgerasterio 1.1.1 py37h163c445_0 conda-forgerasterio 1.1.1 py37h2617b1b_0 conda-forgerasterio 1.1.1 py38h163c445_0 conda-forgerasterio 1.1.1 py38h2617b1b_0 conda-forgerasterio 1.1.2 py36h163c445_0 conda-forgerasterio 1.1.2 py37h163c445_0 conda-forgerasterio 1.1.2 py37h2617b1b_0 conda-forgerasterio 1.1.2 py38h163c445_0 conda-forgerasterio 1.1.3 py36h163c445_0 conda-forgerasterio 1.1.3 py36h2617b1b_0 conda-forgerasterio 1.1.3 py37h163c445_0 conda-forgerasterio 1.1.3 py37h2617b1b_0 conda-forgerasterio 1.1.3 py38h163c445_0 conda-forgerasterio 1.1.3 py38h2617b1b_0 conda-forgerasterio 1.1.4 py36h2409764_0 conda-forgerasterio 1.1.4 py36ha22ed69_0 conda-forgerasterio 1.1.4 py37h02db82b_0 conda-forgerasterio 1.1.4 py37h91b820b_0 conda-forgerasterio 1.1.4 py38h151dc71_0 conda-forgerasterio 1.1.4 py38hef609b1_0 conda-forgerasterio 1.1.5 py36h2409764_0 conda-forgerasterio 1.1.5 py36ha22ed69_0 conda-forgerasterio 1.1.5 py36ha22ed69_1 conda-forgerasterio 1.1.5 py37h02db82b_0 conda-forgerasterio 1.1.5 py37h02db82b_1 conda-forgerasterio 1.1.5 py37h91b820b_0 conda-forgerasterio 1.1.5 py38h151dc71_0 conda-forgerasterio 1.1.5 py38h151dc71_1 conda-forgerasterio 1.1.5 py38hef609b1_0 conda-forgerasterio 1.1.6 py36hc1acebe_0 conda-forgerasterio 1.1.6 py36hc1acebe_1 conda-forgerasterio 1.1.6 py37hce843d0_0 conda-forgerasterio 1.1.6 py37hce843d0_1 conda-forgerasterio 1.1.6 py38hf2e4ed7_0 conda-forgerasterio 1.1.6 py38hf2e4ed7_1 conda-forgerasterio 1.1.7 py36hc1acebe_0 conda-forgerasterio 1.1.7 py37hce843d0_0 conda-forgerasterio 1.1.7 py37hce843d0_1 conda-forgerasterio 1.1.7 py38hf2e4ed7_0 conda-forgerasterio 1.1.7 py38hf2e4ed7_1 conda-forgerasterio 1.1.7 py39h11aa1b2_1 conda-forgerasterio 1.1.8 py37hc4b0cd6_0 conda-forgerasterio 1.1.8 py38h5653988_0 conda-forgerasterio 1.1.8 py39hfec4536_0 conda-forgeThen I directly specified the version to install:
conda install rasterio=1.1.8Using conda to install rasterio conveniently also solved the gdal problem, and I didn’t encounter any other issues during installation.
At this point, in most cases you might run into an issue where rio-color fails to install, because it has a C++ dependency and requires:
error: Microsoft Visual C++ 14.0 is required. Get it with “Microsoft Visual C++ Build Tools”: http://landinghub.visualstudio.com/visual-cpp-build-tools
In this case, just follow the prompt and use visual-cpp-build-tools to install the relevant dependencies (check the following two items):
- MSVC v142 – VS 2019 C++ x64/x86 build tools
- Windows 10 SDK (10.0.18362.0)
After installing these dependencies, I ran into another problem: roughly that the executable “rc.exe” could not be found. After some searching, I found that “rc.exe” existed in the following three directories. I chose to add the x64 folder (since the system is x64) to the PATH environment variable, and that fixed the issue.
C:\Program Files (x86)\Windows Kits\10\bin\10.0.18362.0\arm64\C:\Program Files (x86)\Windows Kits\10\bin\10.0.18362.0\x64\C:\Program Files (x86)\Windows Kits\10\bin\10.0.18362.0\x86\At this point, barring any surprises, you should be able to install rio-tiler with the following command:
pip install rio-tiler --pre