requirements.txt not being generated correctly
jamesramsden-bh opened this issue · 2 comments
In the interim-fixes branch, the requirements.txt does not get set up correctly. To fix,
- remove the two erroneous lines half way down
- remove black
- remove versions for all packages other than LB/HB related packages.
This is a priority fix as it prohibits the installation of the venv (without manually modifying the file anyway)
For reference, the following contents of requirements.txt result in a valid installation of the LB venv.
aiofiles
aiohttp
aiosignal
aiosqlite
anyio
argon2-cffi
argon2-cffi-bindings
arrow
astroid
async-timeout
asynctest
attrs
Babel
backcall
beautifulsoup4
bleach
cached-property
case-converter
certifi
cffi
charset-normalizer
click
click-plugins
colorama
coverage
cramjam
cycler
debugpy
decorator
defusedxml
dill
distlib
dnspython
docutils
dragonfly-core==1.41.16
dragonfly-doe2==0.11.3
dragonfly-energy==1.23.23
dragonfly-radiance==0.2.54
dragonfly-schema==1.9.2
dragonfly-uwg==0.5.352
entrypoints
et-xmlfile
exceptiongroup
fastjsonschema
fastparquet
filelock
fonttools
fortranformat
fqdn
frozenlist
fsspec
greenlet
honeybee-core==1.55.15
honeybee-display==0.2.19
honeybee-energy==1.100.2
honeybee-energy-standards==2.2.6
honeybee-ies==0.7.1
honeybee-radiance==1.64.183
honeybee-radiance-command==1.22.2
honeybee-radiance-folder==2.11.11
honeybee-radiance-postprocess==0.4.219
honeybee-schema==1.53.2
honeybee-standards==2.0.6
honeybee-vtk==0.39.1
idna
importlib-metadata
importlib-resources
iniconfig
ipykernel
ipython
ipython-genutils
isoduration
isort
jedi
Jinja2
joblib
json5
jsonpointer
jsonschema
jupyter-events
jupyter-server
jupyter_client
jupyter_core
jupyter_server_fileid
jupyterlab
jupyterlab-pygments
jupyterlab_server
kiwisolver
ladybug-comfort==0.16.35
ladybug-core==0.41.25
ladybug-display==0.9.1
ladybug-geometry==1.27.1
ladybug-geometry-polyskel==1.4.18
ladybug-radiance==0.2.0
ladybug-rhino==1.39.14
lazy-object-proxy
lbt-dragonfly==0.10.146
lbt-honeybee==0.7.238
lbt-ladybug==0.27.24
lbt-recipes==0.25.8
lockfile
luigi
MarkupSafe
matplotlib
matplotlib-inline
mccabe
meteostat
mistune
multidict
mypy-extensions
nbclassic
nbclient
nbconvert
nbformat
nest-asyncio
notebook
notebook_shim
numexpr
numpy
opencv-python
openpyxl
packaging
pandas
pandas-downcast
pandocfilters
parso
pathspec
pickleshare
Pillow
pip-chill
pkgutil_resolve_name
platformdirs
pluggy
pollination-handlers==0.10.0
prometheus-client
prompt-toolkit
psutil
pyarrow
pycparser
pydantic
pydantic-openapi-helper
Pygments
pylint
pymongo
pyodbc
pyparsing
pyperclip
pyrsistent
pytest
pytest-cov
pytest-order
python-daemon
python-dateutil
python-json-logger
pytz
pywin32
pywinpty
PyYAML
pyzmq
queenbee==1.26.7
queenbee-local==0.5.4
requests
rfc3339-validator
rfc3986-validator
scikit-learn
scipy
Send2Trash
shapely
six
sniffio
sockets
soupsieve
SQLAlchemy
tables
tenacity
terminado
threadpoolctl
tinycss2
tomli
tomlkit
tornado
tqdm
traitlets
typed-ast
typing_extensions
uri-template
urllib3
uwg==5.8.11
virtualenv
vtk
wcwidth
webcolors
webencodings
websocket-client
wrapt
wslink
xlrd
y-py
yarl
ypy-websocket
zipp
This issue is probably aligned with another issue encountered below. It're related to the @BHoM/ladybugtools_toolkit as a Python env being created off of the back of another existing full installation.
I think it's to do with change in file structure meaning the
./helpers
directory is now a./helpers.py
file which confuses robocopy. It would probably be best to delete thePythonCode/*_Toolkit
directory before copying files over for safety.
As a fun knock-on effect, after manually removing the files that stopped the installation first time round, BHoM installer encountered something like error 2830 (can't quite remember the exact code) due to a failed previous installation. I guess advice is to just delete the BHoM directory and reinstall when things like this happen, though I think changing the way that Python_Tk installs an environment might be necessary, being a bit more nuclear with the removal of old code and addition of new stuff.
It might also be worth completely removing any dependency on Python_Toolkit as an installed package into BHoM venvs to avoid this from happening in the future also. We can still use the root BHoM Python install (which is effectively a full Python install, albeit manually built up from the zipped versions from the official Python source) for jupyter server, formatting via Black and maybe linting, but it might be a bit of a set-up required for IDEs ... which can also be overcome with some helpful config files for those using them.