Cleanup and make HACKING.rst DRYer

Reference the OpenStack hacking guide in HACKING.rst and remove
duplicate entries.  Add placeholder section for heat specific
rules.  heat specific rules can be created using hacking's local check

Change-Id: Ib6967ae769bd73857abb7ef89368c407c8b22053
Joe Gordon 10 years ago
parent 78d38ab038
commit 3b6e8d63e2
  1. 268

@ -1,214 +1,14 @@
Heat Style Commandments
- Step 1: Read
- Step 2: Read again
- Step 3: Read on
- Step 1: Read the OpenStack Style Commandments
- Step 2: Read on
Heat Specific Commandments
- Put two newlines between top-level code (funcs, classes, etc)
- Use only UNIX style newlines ("\n"), not Windows style ("\r\n")
- Put one newline between methods in classes and anywhere else
- Long lines should be wrapped in parentheses
in preference to using a backslash for line continuation.
- Do not write "except:", use "except Exception:" at the very least
- Include your name with TODOs as in "#TODO(termie)"
- Do not shadow a built-in or reserved word. Example::
def list():
return [1, 2, 3]
mylist = list() # BAD, shadows `list` built-in
class Foo(object):
def list(self):
return [1, 2, 3]
mylist = Foo().list() # OKAY, does not shadow built-in
- Use the "is not" operator when testing for unequal identities. Example::
if not X is Y: # BAD, intended behavior is ambiguous
if X is not Y: # OKAY, intuitive
- Use the "not in" operator for evaluating membership in a collection. Example::
if not X in Y: # BAD, intended behavior is ambiguous
if X not in Y: # OKAY, intuitive
if not (X in Y or X in Z): # OKAY, still better than all those 'not's
- Do not import objects, only modules (*)
- Do not import more than one module per line (*)
- Do not use wildcard ``*`` import (*)
- Do not make relative imports
- Order your imports by the full module path
- Organize your imports according to the following template
(*) exceptions are:
- imports from ``migrate`` package
- imports from ``sqlalchemy`` package
- imports from ``heat.db.sqlalchemy.session`` module
- imports from ``heat.db.sqlalchemy.migration.versioning_api`` package
# vim: tabstop=4 shiftwidth=4 softtabstop=4
{{stdlib imports in human alphabetical order}}
{{third-party lib imports in human alphabetical order}}
{{heat imports in human alphabetical order}}
{{begin your code}}
Human Alphabetical Order Examples
import httplib
import logging
import random
import StringIO
import time
import eventlet
import testtools
import webob.exc
import heat.api.ec2
from heat.api import openstack
from heat.auth import users
from heat.endpoint import cloud
import heat.flags
from heat import test
"""A one line docstring looks like this and ends in a period."""
"""A multi line docstring has a one-line summary, less than 80 characters.
Then a new paragraph after a newline that explains in more detail any
general information about the function, class or method. Example usages
are also great to have here if it is a complex class for function.
When writing the docstring for a class, an extra line should be placed
after the closing quotations. For more in-depth explanations for these
decisions see
If you are going to describe parameters and return values, use Sphinx, the
appropriate syntax is as follows.
:param foo: the foo parameter
:param bar: the bar parameter
:returns: return_type -- description of the return value
:returns: description of the return value
:raises: AttributeError, KeyError
If a dictionary (dict) or list object is longer than 80 characters, its items
should be split with newlines. Embedded iterables should have their items
indented. Additionally, the last item in the dictionary should have a trailing
comma. This increases readability and simplifies future diffs.
my_dictionary = {
"image": {
"name": "Just a Snapshot",
"size": 2749573,
"properties": {
"user_id": 12,
"arch": "x86_64",
"things": [
"status": "ACTIVE",
Calling Methods
Calls to methods 80 characters or longer should format each argument with
newlines. This is not a requirement, but a guideline::
unnecessarily_long_function_name('string one',
'string two',
kwarg2=['a', 'b', 'c'])
Rather than constructing parameters inline, it is better to break things up::
list_of_strings = [
'not as long',
dict_of_numbers = {
'one': 1,
'two': 2,
'twenty four': 24,
object_one.call_a_method('string three',
'string four',
Internationalization (i18n) Strings
In order to support multiple languages, we have a mechanism to support
automatic translations of exception and log strings.
msg = _("An error occurred")
raise HTTPBadRequest(explanation=msg)
If you have a variable to place within the string, first internationalize the
template string then do the replacement.
msg = _("Missing parameter: %s") % ("flavor",)
If you have multiple variables to place in the string, use keyword parameters.
This helps our translators reorder parameters when needed.
msg = _("The server with id %(s_id)s has no key %(m_key)s")
LOG.error(msg % {"s_id": "1234", "m_key": "imageId"})
None so far
Creating Unit Tests
@ -226,7 +26,7 @@ Running Tests
The testing system is based on a combination of tox and testr. The canonical
approach to running tests is to simply run the command `tox`. This will
create virtual environments, populate them with depenedencies and run all of
create virtual environments, populate them with dependencies and run all of
the tests that OpenStack CI systems run. Behind the scenes, tox is running
`testr run --parallel`, but is set up such that you can supply any additional
testr arguments that are needed to tox. For example, you can run:
@ -241,57 +41,3 @@ run --parallel` will run it in parallel (this is the default incantation tox
uses.) More information about testr can be found at:
A number of modules from openstack-common are imported into the project.
These modules are "incubating" in openstack-common and are kept in sync
with the help of openstack-common's script. See:
The copy of the code should never be directly modified here. Please
always update openstack-common first and then run the script to copy
the changes across.
OpenStack Trademark
OpenStack is a registered trademark of the OpenStack Foundation, and uses the
following capitalization:
Commit Messages
Using a common format for commit messages will help keep our git history
readable. Follow these guidelines:
First, provide a brief summary 72 characters in length or less.
The first line of the commit message should provide an accurate
description of the change, not just a reference to a bug or
blueprint. It must be followed by a single blank line.
If the change relates to a specific driver (libvirt, xenapi, qpid, etc...),
begin the first line of the commit message with the driver name, lowercased,
followed by a colon.
Following your brief summary, provide a more detailed description of
the patch, manually wrapping the text at 72 characters. This
description should provide enough detail that one does not have to
refer to external resources to determine its high-level functionality.
Once you use 'git review', two lines will be appended to the commit
message: a blank line followed by a 'Change-Id'. This is important
to correlate this commit with a specific review in Gerrit, and it
should not be modified.
For further information on constructing high quality commit messages,
and how to split up commits into a series of changes, consult the
project wiki: