• The Fixer services is not giving USD/MXN rates since free accounts block the base rate to EURO.
• API User rate limit: I have to dig more onto this since i've never done it.
• Heroku deployment Unfinished

At this point i don't count with much information about how the service work. I would like to start by proposing a generla development Pipeline with two main blocks:

• Continous Integration Pipeline
• Continous Development Pipeline

Is the main set of tools that will helps with the work flow and implementation constant and with enough capabilities to keep growing and adding new features and developers. Some commons tools to use will be:

• Git: github, gitlab
• Pytest: Create unitary tests for every new module integrated to the project
• Virtual environments

Asuming the tools mentioned in the last section where used/integrated a continous Deployment pipeline can be set up in order to control the quality of code being sent to production before and during deployment.

By using pyTest a separated set of tests can be written that are connected to production services. Different code repository services offer Continous integration pipelines where pytest tests can be checked for 100% completition in order to accept any kind of deployment and merge requests.

At the same time continous integration services usually offer container services such as docker for standarization of the testing services. This services allow to create multiple images of the desired service where an instance of that service can be used to replicate what production services look like.

Based on what was explained above the next test pipeline diagram demonstrates the general idea of the production deployment pipeline.

• server connection check
• database response check
• client data check
• client parameters check
• model response time
• model output validation

Docker:

• Go to project folder
• from terminal run:

    docker-compose up 
  

• Enter http://localhost:5000/ in a browser to see the application running.
• If this doesn’t resolve, you can also try http://127.0.0.1:5000
• It will ask for user and password to get a token:

  user: appsecret
  pass: appsecret
  

• This will return an app token that can be used on query urls (instructions below).

Python:

• App uses flask simply running exchange_service.py will run a local host on port 5000 cd to project folder python exchange_service.py On browser: go to 127.0.0.1:5000/

http://localhost:49155/rates

• notes: fixer (provider0) free account doesnt allow to choose base as USD, defaul base EUR

http://localhost:49155/fixer?token=user_token

• notes: fixer (provider0) free account doesnt allow to choose base as USD, defaul base EUR

http://localhost:49155/bmx?=user_token

• notes: fixer (provider0) free account doesnt allow to choose base as USD, defaul base EUR

http://localhost:49155/dof?token=user_token

• notes: fixer (provider0) free account doesnt allow to choose base as USD, defaul base EUR

The scenario clear shows a case of ONE-WAY dependency. The fact that problems come up in production on the other team changes means that my team code is not being tested by the other team before production deployment.

The first thing i would do is to analyze with my team the main features that break on the other team changes. With this in mind i will try to talk to the othe tech lead and try to reach an agreement on which of our features they can and need to add test cases into their code.

Most of the time this type of cases result from resistance from teams to adapt to external test case since this result in less liberty for their team. Nevertheless it is always necessary to reach an agreement on how much freedom of implementation a team can have without considering other teams.

The main response I anticipate to have from the other team is some rejection to implement test cases for our code. A solution to this could be to propose a list of development constraints for future implementations. They could test over this constrains and constrains will not be change unless an agreement is reached by both teams and team leads.

A more General/LongTerm solution: Remove dependencies Implement a gate keeping strategy, where new stories with dependencies get blocked. Until dependencies are removed if possible.