This is my recording of the steps I’ve taken to setup my machine for the data analysis I’ve done here at the City. Theses instructions would be most useful to those working with spatial databases, especially if an ESRI product is installed. I am configuring my installations of PostgreSQL, ArcGIS for Desktop, and Python.
Esri has actually put together a pretty decent how-to guide on getting setup with Anaconda on your machine. For this, I went with the 64-bit installation using Python 3.X
Once you install Anaconda, open the Anaconda command prompt and type the following to install the esri Python API:
conda install -c esri arcgis
The “-c esri” directs conda to look for the arcgis package in the esri channel instead of the default/main channel. In my data analysis, I plan to connect to our PostgreSQL database, so I will need the psycopg2 package:
conda install psycopg2
For moving data to/from the shapefile format, I’ll need fiona:
conda install -c conda-forge fiona
Similarly, we direct conda to look for the fiona package in the conda-forge channel instead of the default/main channel.
Currently have: ArcGIS for Desktop, Python 2.7
Setting Up: PostgreSQL, Python 3
If, like me, you are running ArcGIS for Desktop (10.4.1 in my case), you will already have the 32-bit version of python 2.7 installed here:
C:\Python27\ArcGIS10.4\python.exe
I used the interactive installer provided through EnterpriseDB and recommended on the PostgreSQL website, using StackBuilder to easily add the PostGIS extension. Go ahead and setup with the latest version (9.6.2 in my case with the 64-bit Windows OS). Devan Morris over at SF Public Health has some great screenshots (though using an older verison) to guide you through the installation process.
One of the functionalities that I needed to enable includes the Pl/Python language so that I can run python scripts within the PostgreSQL database. Unfortunately, based on information here, it looks like the Windows builds no longer contain support for plpython2 (for Python 2.7), only plpython3 (for Python 3.X).
Therefore, to be able to use python within PostgreSQL, make sure to install Python 3. When deciding between the 32 or 64 bit installation, check with your PostgreSQL install to make you match (in my case, I installed the 64-bit version) and then install from the python website.
However, the trickiest part of this installation is that you will need to install a specific version of python, which I found out via this answer. In my case, using the dependency walker on plpython3.dll (found in C:\Program Files\PostgreSQL\9.6\lib) showed that plpython3.dll was looking for python33.dll. I therefore installed the latest version of Python 3.5 on my machine and updated the path variable. Once this was complete, I was able to go into PgAdmin4, open up my database, right-click “Languages,” then “Create Language” and then select plpython3u and it worked perfectly.
The tricky part of this is to install the Anaconda distribution of python without breaking the original ESRI python stack. I just followed these instructions from the USGS. Make sure you download the 32-bit installation, even if you have a 64-bit machine.
I followed the instructions on this page to setup my machine for github.