| author | date | title | |
|---|---|---|---|
|
Wednesday 9th December 2020 |
'Software Sustainability: If a tree falls in a forest\...' - King's workshop exercise |
The below material assumes that the software sustainability course has been completed (https://www.youtube.com/playlist?list=PLxyHJ_wep1_DPbvtFl_-EGyoz2pVt-n1_), and that Python 3, nodejs, Git, Docker and Docker Compose are all installed.
{height="5cm"}
We are going to imagine we have just invented a perfect model of the weather, that allows us to make predictions for a specific location on a specific day.
Our model takes as input a location (e.g. "London") and the number of days from now at which to make a prediction about the weather in that location (e.g. 3 (days from now)).
To make a prediction, the model first takes the Unicode value of the first letter of the location (e.g. 'L', which is 76 in Unicode) and computers the remainder of this value when divided by 3. This produces a number between 0 and 2. Next, the model takes the number of days from now at which to make a prediction, and also computes the remainder of this value when divided by 3. This also produces a number between 0 and 2. These two numbers are then added together to produce a number between 0 and 4.
To make a final prediction, each number produced by the model is matched to a specific type of weather:
0 = sun, 1 = wind, 2 = rain, 3 = snow and 4 = cloud.
-
A user enters
London{.bash} and3{.bash}. -
The value of 'L' in Unicode is 76. The remainder of 76 / 3 is 1.
-
The remainder of 3 / 3 is 0.
-
Our final prediction value is 1 (1 + 0), which corresponds to 'wind'.
Realise the weather model described in Python code, adhering to the principles of coding practice discussed in the course.
Hint: the ord(){.python} function is useful for converting a character
to its Unicode value.
Your code should accept a location and a number of days as user input, and use the model to make a prediction based on this input.
Make a version of your software, and push it to a remote repository on King's Github (github.kcl.ac.uk). Call this repository 'simulation'.
You will need to log in with your k-number and password, if you have not done so before.
Public Github (github.com), should also work for this task, but should only be used as a backup.
Additional versions should be made at appropriate points while completing the remaining tasks.
To receive a mark for your work, add the user MC as a collaborator to your repository.
To add someone as a collaborator, from your repository visit Settings
If using public Github as a backup, the path to add a collaborator will instead be Settings $>$ Manage access. Click Invite a collaborator, and in the box that pops up search for the user 'martinchapman' and click the top result. Verify that 'martinchapman' is a collaborator on your repository.
If you are asked to set an access level for the collaborator, choose Admin.
Via this collaborator link, your repository will be accessed and your mark will be based on the number of passing tests in your code.
Write three tests to ensure the following:
-
The weather in London in 5 days time is "snow".
-
The weather in London in 365 days time is "snow".
-
The weather in London in -1 days time is not reported.
Modify your program such that these tests pass, if needed.
You are welcome to add additional tests, but these must all pass.
The supporting material for this session is available at github.com/martinteaching/sustainability2020.
Wrap your Python weather prediction model in a server and run it.
Move the request for user input (location and days) to a Javascript client, which, once acquired, sends this information to the server, waits for it to compute the result using the model, and then prints the result for the user.
Dockerise your program, using a Dockerfile and Docker compose, allowing the Python server to run in a container, and the Javascript client to issue requests to it.
If you wish to expand on your solution, separate your Python weather model server code into two servers (services). Each service should contain code to compute part of the weather prediction (e.g. one service for the location calculation, another for the day calculation). Designate one of your services as the 'user-facing' service, to be called by the client (Service A). Service A, when called by a client, should then subsequently call the other service (Service B) for its result, before combining Service B's result with its own, and returning this in a response to the client.
For example, Service A could compute the location value, call Service B to compute the day value, combine the results, and return them to a requesting client.