This is a user request that was mentioned during a meeting today (presentation of DataLab to a research laboratory).
The use case is the following:
The end-users are working in scientific experimental facilities
The target OS is Microsoft Windows
Specific features will be required: developing specific plugins in DataLab meets those requirements
The easiest way to deploy DataLab is the standalone version (Windows installer)
However, when developing plugins, one has to rely on the Python environment in which DataLab is executed. For the standalone version, that is the frozen executable Python environment built with PyInstaller. The issue is that this specific environment does not allow to install additional packages that would be required by the third-party plugins.
So, we propose to investigate an alternative to PyInstaller: building our own custom WinPython distribution. This would allow to distribute DataLab on Windows with the same advantages as PyInstaller (i.e. ready-to-use standalone DataLab version) and with an access to the Python interpreter/environment, thus allowing to install additional packages.
Warning: this would also be a significant flaw as this distribution could easily be broken by the end-user (whereas the PyInstaller version is bullet proof).
Unless we have another idea to address the initial request?
This is a user request that was mentioned during a meeting today (presentation of DataLab to a research laboratory).
The use case is the following:
However, when developing plugins, one has to rely on the Python environment in which DataLab is executed. For the standalone version, that is the frozen executable Python environment built with
PyInstaller
. The issue is that this specific environment does not allow to install additional packages that would be required by the third-party plugins.So, we propose to investigate an alternative to
PyInstaller
: building our own custom WinPython distribution. This would allow to distribute DataLab on Windows with the same advantages asPyInstaller
(i.e. ready-to-use standalone DataLab version) and with an access to the Python interpreter/environment, thus allowing to install additional packages.Warning: this would also be a significant flaw as this distribution could easily be broken by the end-user (whereas the
PyInstaller
version is bullet proof).Unless we have another idea to address the initial request?