python binding is not installing dependencies

[xiazeyu]

Hello,

I tried the pip install -e . inside of the docker container from Quick Start guide. However, necessary dependencies like numpy and matplotlib are not installed.

Please take a look so it installs the dependcies that as well.

Ref: openjournals/joss-reviews#8680

[antonio-leblanc]

Hi @xiazeyu
Thanks for pointing it out!

Indeed python was not beeing installed on the Docker image..
Just pushed and updated that install python and the python bindings #112

I also set the default command to bash, which makes running the container a bit simpler (no need for bash at the end). The README and docs are updated accordingly.

Thanks again for the help!

[xiazeyu]

I tried to build and run the new docker image. Inside it, I can successfully execute python3, but numpy and matplotlib seems not installed with the python binding. Actually that also happens when I tried to install the binding using pip install -e .

Could you please check on that as well?

[antonio-leblanc]

okay sorry for that i will check again!

[antonio-leblanc]

@xiazeyu
Apologies for the confusion. Those dependencies where listed in pyproject.toml but not in setup.py
They were added.
Now when entering the container and checking pip list they are correctly installed

[antonio-leblanc]

Also tested the commands that are working... after entering the container

• cd bindings/python
• pip install -e .

[xiazeyu]

Have you tried

python3

import pyforefire as forefire

# Create an instance of the ForeFire class
ff = forefire.ForeFire()

# Example usage: define a domain command
sizeX = 300
sizeY = 200
myCmd = "FireDomain[sw=(0.,0.,0.);ne=(%f,%f,0.);t=0.]" % (sizeX, sizeY)

# Execute the command
ff.execute(myCmd)

[antonio-leblanc]

yes the code is running with some warnings though...

i will check with filippi if maybe there is a better example that can be used on the documentation

[antonio-leblanc]

For example running the tests inside tests/python and check that the code is running...

---

I've been contributing in other aspects of the project, but i will check with filippi the best test case that can be documented on the readme for the users that wanna use the pyforefire

Thanks again for your help! we will update soon

[filippi]

Hi @antonio-leblanc @xiazeyu

The python is kind of supposed to do nothing to show you it is well installed. It is just there so it does not say "no packages named pyforefire".. and that it can create a domain.. then you know it is technically well installed

There is also no 3 liners to perform a sensible simulation of something, to do that, it is best to check in the /tests/python directory
It will still emit a warning (because no fuel tables can be found.. I keep it as default because it is a very common misuse not to use user-defined fuel tables instead of default), but if you want actual simulation, you may change it to do :

import pyforefire as forefire
ff = forefire.ForeFire()
sim_shape = (10000,10000)
ff.execute(f'FireDomain[sw=(0,0,0);ne=({sim_shape[0]},{sim_shape[1]},0);t=0]')
ff.addLayer("propagation","Iso","propagationModel")
ff.execute(f'startFire[loc=({sim_shape[0]/2},{sim_shape[1]/2},0.0)]')
ff.execute("step[dt=1000]")
print(ff.execute("print[]"))
if(False): # Change to True to export KML
ff["dumpMode"] = "kml"
ff.execute("print[circle.kml]")

that will make a nice circle and event create a kml of this circle in Corsica, the default projection area when none is provided.

[xiazeyu]

LGTM.