You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When trying to package the new biogeme on conda forge we see many new dependencies in the pyproject.toml file and notice that some may not be hard dependencies:
IPython is only used in the examples
No Jinja2 imports are found in the source code at all
python-levenshtein does not seem to be imported in the source code anywhere
tomlkit and fuzzywuzzy seem to be imported once, could these be made into optional dependencies?
Also, does the package really not work with NumPy < 2? NumPy 2 is very new and putting such a restriction makes it hard to install in many environments because the whole ecosystem hasn't fully adopted NumPy 2 yet. For example, we hope to include this in the software installations at TU Delft in the computer labs but we will be running NumPy 1 for this coming year, as 2 can't be installed with many packages in our distribution.
Lastly, are the upper bound pins really necessary? This also creates unnecessary restrictions and makes it difficult to install the package in environments with other software. (TDLR about this topic: https://iscinumpy.dev/post/bound-version-constraints/)
Thanks,
Jason
The text was updated successfully, but these errors were encountered:
Hi,
When trying to package the new biogeme on conda forge we see many new dependencies in the
pyproject.toml
file and notice that some may not be hard dependencies:Also, does the package really not work with NumPy < 2? NumPy 2 is very new and putting such a restriction makes it hard to install in many environments because the whole ecosystem hasn't fully adopted NumPy 2 yet. For example, we hope to include this in the software installations at TU Delft in the computer labs but we will be running NumPy 1 for this coming year, as 2 can't be installed with many packages in our distribution.
Lastly, are the upper bound pins really necessary? This also creates unnecessary restrictions and makes it difficult to install the package in environments with other software. (TDLR about this topic: https://iscinumpy.dev/post/bound-version-constraints/)
Thanks,
Jason
The text was updated successfully, but these errors were encountered: