# Development Guidelines for the GeoAdmin team

-> **None of us is as good as all of us** <-

## Definitions

Production/prod: Anything that we offer as a service to the public or a 3rd party.

## We embrace the [agile manifesto](http://www.agilemanifesto.org/)

- We work according to its [twelve principles](http://www.agilemanifesto.org/principles.html)

## What we consider a good Pull Request (PR)

- A PR addresses a well described and well documented issue
- A PR has a proper name
- A PR contains atomic and well named commits
- A PR describes the changes that are introduced and mentions possible side-effects for testing
- A PR describes the necessary steps to undertake before merging (infrastructure, data, varnish, etc.)
- A PR provides a testlink - and if that's not possible describes how it can be tested
- A PR should be of **production** quality

## We review each others work, including our own

- When creating a PR, you are the first to review it.
- Before creating a PR, we review a PR by another developer.

## We treat *master* as production

- Everything that is in master has production quality

## We never commit/push to *master*

- We always develop on branches

## We develop in *dev* environments, using *dev* resources

- We consider resources that are auto-deployed (db tables, others?) to be production resources
- We never develop using production resources
- We never develop code/scripts that read/manipulate production resources
- We never create testlinks or do tests that use production resources

## We treat each script/test/codepiece/data the same

- All code is production code, including testprojects, POC code, etc
- We put everything up for review (including bash scripts, SQL, etc)

## We write tests for each piece of code

- We do an extra effort to write tests for our code
- Test code are an integral part of any development

## We never deploy on Friday, before holidays or alone

- This concerns all production systems
- This concerns infrastructure too (puppet runs)
- Established and tested automatic scripts/processes are exceptions to this guideline

## We actively test production after each deploy

- We develop, then deploy. These are 2 separate processes.
- After each deploy, we test if the deployment was successfull.
- Auto-triggering scripts are tested directly after the deploy.

## We treat global resource with care

Gobal resources are resources shared by every developer. These
are currently:
- mf0 machine
- apache on mf0 (take care of your config!)
- sphinxsearch
- db's (_dev)

-> **Never stop improving** <-