Troubleshooting

This section provides an FAQ for common problems that occur when using TC-Python.

Diagnostics script

If you have problems running TC-Python, run the diagnostics script below.

On Linux you can alternatively download the script directly into your current working directory by:

curl -O https://www2.thermocalc.com/downloads/support/diagnostics-py/tc-python-diagnostic-script-2020b.py
"""
Run this script when troubleshooting TC-Python

It is important to run this script EXACTLY the same way as you run your TC-Python script
(In the same IDE, same project, same Python environment, same Jupyter notebook e.t.c)

"""

version = '2020b'

print('Testing TC-Python version: ' + version)
print('Please make sure that the variable "version" above, matches the release that you want to test, if not change it and re-run this script.')

# below this line, nothing needs to be manually updated.

import sys
print('')
print('Python version: (should be at least 3.5 and can NOT be older than 3.0)')
print(str(sys.version_info[0]) + '.' + str(sys.version_info[1]))
if sys.version_info[0] < 3 or sys.version_info[1] < 5:
    print('Wrong version of Python !!!!!')

print('')
print('Python executable path: (gives a hint about the used virtual / conda environment, in case of Anaconda the corresponding \n'
      'environment name can be found by running `conda env list` on the Anaconda command prompt, '
      'TC-Python must be installed into \nEACH separate environment used!)')
print(sys.executable)

import os
print('')
print('Thermo-Calc ' + version + ' installation directory: (must be a valid path to a complete installation of ' + version + ')')
tc_env_variable = 'TC' + version[2:].upper() + '_HOME'
try:
    print(os.environ[tc_env_variable])
except:
    print('No Thermo-calc environment variable for ' + version + ' was found. (' + tc_env_variable + ')')

print('')
print('Url of license server: (if license server is NO-NET, you need a local license file)')
try:
    print(os.environ['LSHOST'])
except:
    print('No Thermo-calc license server url was found. (LSHOST)')


print('')
print('Path to local license file: (only necessary if not using license server)')
try:
    print(os.environ['LSERVRC'])
except:
    print('No path to local license file was found. (LSERVRC)')


import tc_python
numerical_version = version[:-1]
if version[-1] == 'a':
    numerical_version += '.1.*'
elif version[-1] == 'b':
    numerical_version += '.2.*'
print('')
print('TC-Python version: (needs to be ' + numerical_version + ')')
print(tc_python.__version__)


with tc_python.TCPython() as session:
    print('')
    print('Lists the databases: (should be a complete list of the installed databases that you have license for or do not require license)')
    print(session.get_databases())

“No module named tc_python” error on first usage

This problem occurs because your used Python interpreter cannot find the TC-Python package. We expect that you have installed the TC-Python package in your Python system interpreter following the instructions in the TC-Python Quick Install Guide.

Normally the error message “No module named tc_python” is caused by unintentionally configuring a PyCharm project to use a so-called Virtual Environment. This happens unfortunately by default when creating a new PyCharm project with not changing the default settings.

Note

A Virtual Environment is basically a separate and completely independent copy of the system-wide Python interpreter. It does not contain any packages.

On Windows systems we recommend to use the Anaconda Python Distribution as Python interpreter. However, the instructions given here are valid for any operating system and distribution.

Since TC-Python 2018b we do recommend to not use Virtual Environments unless there is a reasonable use case for that.

There are two possible solutions to fix the problem:

  1. The quick fix for your problem is to run

    pip install <path to the TC-Python folder>/TC_Python-<version>-py3-none-any.whl
    

    within the Terminal window of the opened PyCharm project. This Terminal window automatically runs within the Virtual Environment configured for the project (if any). You can see the name of the Virtual Environment at the beginning of each command prompt line (here it is called venv):

    Microsoft Windows [Version 10.0.16299.431]
    (c) 2017 Microsoft Corporation. All rights reserved.
    
    (venv) C:\Users\User\Documents\>
    

    The command will consequently install TC-Python also within that Virtual Environment automatically. The Terminal window can be found at the bottom of the IDE. Note that it might be necessary to enable these buttons first by selecting the menu entry View→Tool Buttons.

  2. The better fix is to change your project to use the system interpreter. This is described in detail in the section Fixing potential issues with the environment in Step 5 of the TC-Python Quick Install Guide.

    It is recommendable to use that approach also for all your future projects.

Both fixes will only change the configuration of the opened project. Further useful information can be found in the section Python Virtual Environments.

“pip install” fails with “Failed to establish a new network connection” or similar

If pip install fails with a network related error (might also be “socket not available”, “retrying after connection broken”, …) it is often due to the computer being behind a proxy-server, this is common in large organizations. Of course also the network connection might be broken.

TC-Python has dependencies to a few other packages:

  • py4j
  • jproperties
  • six (transient dependency of jproperties)
  1. The recommended approach is to simply use pip. It will resolve the dependencies automatically by downloading them from the PyPI-repository server (https://pypi.org). If your computer is located behind a proxy-server, the connection to the repository will fail. In that case it is necessary to configure pip with the detailed configuration of the proxy server:

    pip install -proxy user:password@proxy_ip:port py4j jproperties
    
  2. Another alternative is to download the latest *.whl-file of each dependency from the repository server (https://pypi.org -> Search projects) and to install it manually using:

    pip install py4j-#.#.#-py2.py3-none-any.whl
    ...
    

    The actual actual version number needs to be inserted into the file name. The downside of this approach is that updates to that package have to be fully manual also in the future. Additionally it is also necessary to install all transient dependencies in that way.