`Error: Python module QuantLib was not found.`

[avhz]

I am trying to use QuantLib in an R package via reticulate, but am running into the following:

.onLoad <- function(libname, pkgname) {
    reticulate::py_require("QuantLib")
    ql <<- reticulate::import("QuantLib", delay_load = TRUE)
}

When I run devtools::load_all(), I get:

> devtools::load_all() 

List of 22
 $ python              : chr "/opt/venv/bin/python"
 $ libpython           : chr "/usr/lib/python3.12/config-3.12-x86_64-linux-gnu/libpython3.12.so"
 $ pythonhome          : chr "/opt/venv:/opt/venv"
 $ pythonpath          : chr "/usr/lib/python312.zip:/usr/lib/python3.12:/usr/lib/python3.12/lib-dynload:/opt/venv/lib/python3.12/site-packages"
 $ prefix              : chr "/opt/venv"
 $ exec_prefix         : chr "/opt/venv"
 $ base_exec_prefix    : chr "/usr"
 $ virtualenv          : chr "/opt/venv"
 $ virtualenv_activate : chr ""
 $ executable          : chr "/opt/venv/bin/python"
 $ base_executable     : chr "/usr/bin/python3.12"
 $ version_string      : chr "3.12.3 (main, Feb  4 2025, 14:48:35) [GCC 13.3.0]"
 $ version             :Classes 'package_version', 'numeric_version'  hidden list of 1
  ..$ : int [1:2] 3 12
 $ architecture        : chr "64bit"
 $ anaconda            : logi FALSE
 $ conda               : chr "False"
 $ numpy               : NULL
 $ required_module     : chr "QuantLib"
 $ required_module_path: NULL
 $ available           : logi TRUE
 $ python_versions     : chr "/opt/venv/bin/python"
 $ forced              : chr "VIRTUAL_ENV"
 - attr(*, "class")= chr "py_config"
Error: Python module QuantLib was not found.

Detected Python configuration:

>

Am I doing something wrong ? The docs say it is not recommended to use py_install anymore, so I am a bit stuck.

[eddelbuettel]

Is it there when you try to access it directly in Python? If so, is reticulate using the same Python setting / venv / ... ?

root@e15cc0a16015:/# /opt/ql/bin/python
Python 3.12.3 (main, Feb  4 2025, 14:48:35) [GCC 13.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import QuantLib;
>>> 

(This was in a fresh Docker container to test a pristine pip3 installation in a fresh venv.)

[eddelbuettel]

There is a whole other set of complications because of the recent PEP and the distro-managed / not-distro-managed packages. I maintain QuantLib for Debian/Ubuntu, and I have a (local, PPA-built, more recent than in the distro) quantlib-python here but that also only loads if I explicitly manage the PYTHONPATH:

edd@rob:~$ PYTHONPATH=/usr/lib/python3/dist-packages/QuantLib-1.38-py3.12-linux-x86_64.egg python3
Python 3.12.7 (main, Feb  4 2025, 14:46:03) [GCC 14.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import QuantLib;
>>> 

[t-kalinowski]

The py_require() call above is being ignored; forced indicates that Python was selected because VIRTUAL_ENV is set.

If the py_require() call were being used, you would see an ephemeral Python in the cache being used.

py_require() is only applicable if:

1. the user has opted in to using a reticulate-managed environment, or
2. the user has not specified some other self-managed Python (i.e., it’s a "green" fresh-out-of-the-box setup).

If you want to opt in and force usage of a py_require() Python, you can set this before reticulate has initialized Python:

Sys.setenv("RETICULATE_PYTHON" = "managed")

Or, make sure that nothing higher in the Python discovery order is present: https://rstudio.github.io/reticulate/articles/versions.html#order-of-discovery

[avhz]

Awesome, thank you ! This solved it:

.onLoad <- function(libname, pkgname) {
    Sys.setenv("RETICULATE_PYTHON" = "managed") ## <- added here

    reticulate::py_require("QuantLib")
    ql <<- reticulate::import("QuantLib", delay_load = TRUE)
}