-
Notifications
You must be signed in to change notification settings - Fork 79
Branching
The main master and develop branches names will not change and these branches will not be deleted.
Supporting branches should have a prefix so that they can be easily identified or searched for.
Branch Type | Prefix | Example | Used for |
---|---|---|---|
Feature | feature- | feature-shiny-new | Working on a new feature. General use branch for working on a feature. |
Hotfix | hotfix- | hotfix-crash-on-import | Fixing a critical bug. |
Release | release- | release-v1.0.0 | Preparing a release. |
Examples | example- | example-slide-info | Adding example code / notebooks. |
Bugfix | bug- | bug-slide-info-format | Working on a bug fix. |
Dev-Ops | dev- | dev-code-coverage | Working on a developer tools, operations tools or utility e.g. Travis or code coverage configurations. |
Tests | test- | test-slide-info | Addition or modification of tests. If adding code and test prefer feature. |
Documentation | doc- | doc-advanced-usage | Adding to or editing documentation or docstrings. If adding code and docs prefer feature. |
Enhancement | enhance- | enhance-wsireader-inherit | Enhancement of existing code. |
git checkout -b feature-shiny-new
This will create a new branch off of the current branch.
This repo follows Vincent Driessen's wildly popular git flow model of branching. Please do read his blog post for the full rundown at https://nvie.com/posts/a-successful-git-branching-model/. Below is just a condensed version.
Two main branches are active indefinitely for the lifetime of the repo: master and develop. Master is the default branch created by git. This branch contains a tagged commit for each release of the package. The develop branch is considered to be the main branch where HEAD points to the current working state of the repo. This is also called the "integration branch". When the develop branch is ready to be released as a new version, it is merged into master and tagged. A tag on a commit triggers the Travis continuous integration system to automatically build and release the package on PyPi as well as it's documentation on ReadTheDocs.
develop main
│ │
╭╌╌╌╌╌╌╌╌╌╌╮
│ ╭─┴─╮ ╎ Initial ╎
╭───╮ ┌────│ │◀─╎production╎
│ │◀───┘ ╰─┬─╯ ╎ version ╎
╰───╯ │ ╰╌╌╌╌╌╌╌╌╌╌╯
│ │
▼ │
╭─┴─╮ │
│ │ │
╰─┬─╯ │
│ │
▼ │
╭───╮ ▼
│ │────┐ ╭─┴─╮
╰───╯ └───▶│ │
│ ╰─┬─╯
▼ │
╭─┴─╮ │
│ │ │
╰─┬─╯ │
│ │
▼ │
╭───╮ ▼ ╭╌╌╌╌╌╌╌╌╌╌╮
│ │────┐ ╭─┴─╮ ╎ Next ╎
╰───╯ └───▶│ │◀─╎production╎
│ ╰─┬─╯ ╎ release ╎
▼ │ ╰╌╌╌╌╌╌╌╌╌╌╯
╭─┴─╮ │
│ │ │
╰─┬─╯ │
│ │
▼ │
╭───╮ ▼ ╭╌╌╌╌╌╌╌╌╌╌╮
│ │────┐ ╭─┴─╮ ╎ Next ╎
╰───╯ └───▶│ │◀─╎production╎
╭╌╌╌╌╌╌╌╌╌╌╮ │ ╰─┬─╯ ╎ release ╎
╎ Work in ╎ ▼ ╰╌╌╌╌╌╌╌╌╌╌╯
╎ progress ╎ ╭─┴─╮ │
╎ on "next ╎─▶│ │
╎ release" ╎ ╰─┬─╯ │
╰╌╌╌╌╌╌╌╌╌╌╯
│ │
Some supporting branches may also be used to allow parallel development. These are typically: hotfixes, feature branches, release preparation. These branches are removed when no longer needed.
{width="1.8645833333333333in" height="5.0in"}
<https://nvie.com/posts/a-successful-git-branching-model/#feature-branches Feature branches (AKA topic branches) should, in general, branch off of develop and be merged back into develop.
$ git checkout -b feature-example develop
Switched to a new branch "feature-example"
$ git checkout develop
Switched to branch 'develop'
$ git merge --no-ff feature-example
Updating ea1b82a..05e9557
(Summary of changes)
$ git branch -d feature-example
Deleted branch feature-example (was 05e9557).
$ git push origin develop
The use of --no-ff (no fast-forward) is important here. Also note that GitHub pull requests use --no-ff by default when merging.
From Vincent's post:
The --no-ff flag causes the merge to always create a new commit object, even if the merge could be performed with a fast-forward. This avoids losing information about the historical existence of a feature branch and groups together all commits that together added the feature. Compare:
{width="5.0in" height="4.427083333333333in"}
In the latter case, it is impossible to see from the Git history which of the commit objects together have implemented a feature---you would have to manually read all the log messages. Reverting a whole feature (i.e. a group of commits), is a true headache in the latter situation, whereas it is easily done if the --no-ff flag was used.
Yes, it will create a few more (empty) commit objects, but the gain is much bigger than the cost.
In short, --no-ff preserves the fact that development happened in parallel rather than sequentially. Fast forwarding, the default merge behaviour, makes the history appear as if the commits have all been applied sequential on top of the current HEAD of the develop branch. This is technically what is actually happening when merging. However, the history is usually more clear if the parallel development history is preserved.
Note that you don't necessarily ALWAYS want to use --no-ff. For example, if you have made a single commit feature branch then a plain merge will create a cleaner history.
For more on this discussion see this stack overflow: <https://stackoverflow.com/questions/18126297/when-to-use-the-no-ff-merge-option-in-git
https://docs.microsoft.com/en-us/azure/devops/repos/git/git-branching-guidance?view=azure-devops