Official website

Codebase review

21 Dec 2011

min read

Codebase is a project hosting solution by aTech Media. It tries to be a complete solution, including repository hosting, issue and time tracking, wiki and discussion forum. We are currently considering it as a Gitorious replacement in the near future, and as a lightweight replacement for Jira and Confluence later.

Hierarchical structure

The main hierarchical unit in Codebase is a project. Every project can have an unlimited number of source code repositories, and will have a Tickets (issue tracker), a Notebook (wiki-like functionality), Files, Discussions and Time (a time tracker).

A project always belongs to a company, and optionally to a group. For example, we could have two companies in our Codebase account, one called ‘Lunatech’ and one called ‘[Minoto]’. Lunatech would have groups such as ‘Internal applications’ and ‘Customer X projects’, and the latter would contain projects such as ‘Project Y’, which in turn contains the repositories y-server and y-client.

Projects within Codebase can be archived. Archived projects cannot be written to, but they also do not count towards the project limit, which is what the Codebase pricing model is based on (that, and storage).


Repositories in Codebase are not necessarily git repositories – when you create a repository you can choose between Git, Subversion, Mercurial and Bazaar. But we’ll probably stick with git.

Unlike most other (git) hosting solutions, there is no notion of a ‘fork’ in Codebase. There’s the repository, which you can of course clone, but that’s it. You cannot fork it to your own account, every repository is in the context of the company account (see hierarchy). Of course you are free to create a second repository and push the exact same commits to it, but Codebase does not know about any relation between the two repositories. By extension, there are no pull/merge requests. This makes certain development workflows hard or impossible.

Access control

Users are assigned a Role. Roles are groups of permissions that are granted to users with that role. Every user can have one, and only one, role. This keeps things nice and simple. Roles are customisable, but a Codebase account comes with a sane set of default roles:

  • Account administrator – Full access to the account.

  • Administrator – Full access to the account

  • Staff members - Read/Write access to all projects. This is the role all employees should have.

  • Trusted users - Read access to all projects. We probably don’t need this.

  • Users – No permissions at all, which means users with this role will have read/write access only to projects to which they have been assigned.

Access to repositories is configured per-project (in other words, you get access to a project, which includes repositories). Users can gain access through a role which gives them automatic permission, or they can be assigned to a project, giving them full read/write access by default (but this can be customised per user).


Each project can be assigned a Notebook, which is Codebase’s way of saying ‘wiki’. Notebook pages are written in Markdown (with Codebase specific extensions), Textile or plain text, but Markdown is recommended. Pages can be edited by checking out the notebook as a git repository, or through an editor built into codebase. Note that there is no WYSISWIG editor, but there is a preview feature.

Some features are missing when compared to Confluence: There is no way to add images or attachments to a wiki page, although there is an area per project where you can upload files. Files can be assigned a category from a pre-defined list, customisable per-project. Also MIA are comments on pages, but there is a separate ‘discussion’ area per project. Discussion can also be assigned a category, and again the list of categories is customisable per-project.

Issue tracker

The issue tracker is not as comprehensive as, say, Jira, but that’s a good thing, since it is less complex. There is an issue tracker per project, and tickets have the following properties, the possible values of which can be customised, you guessed it, per-project:


Default values

Jira equivalent


Bug, Enhancement, Feature, Task

Issue type





New, Accepted, In Progress, Completed, Invalid



Critical, High, Normal, Low



API, Cosmetic, General, Refactoring, Security


While the extra level of hierarchy that Jira calls ‘components’ isn’t there, Codebase has the concept of a ticket ‘category’, which could be used for the same feature, or some sort of project organisation.

Time Tracking

Time tracking in Codebase is simple and straightforward: You log chunks of time spent (expressed in hours and minutes) with a date and summary, and optionally a group (which just groups hours). Hours can be marked billed or unbilled, which is useful for, well, billing. The time tracking system is integrated with the issue tracker, so it’s easy to enter hours spent on a certain issue. There are reporting features, other than just basic browsing and filtering of hours. However, all hours can be exported as CSV and also read through the API for integration with other applications.

Unfortunately, this functionality does not appear to be enough for customer project use. For example, it does not appear to be possible to show all hours for one person (on all projects), even the in the CSV export (which does not include a project column); this is needed to check whether hours are complete. For billing one customer, however, it does appear to be possible to show all hours (for all people) for one project between two dates.


Codebase has a RESTFul, JSON-based HTTP API that allows read/write access to all features listed above, with the exception of ACL management. Authentication is HTTP basic with an API access token per user. In addition to the REST API it’s possible to configure an URL per repository that will receive an HTTP POST whenever someone pushes to that repository.


Codebase is an obvious migration path from JIRA/Confluence to hosted DVCS with integrated lightweight issue tracker and wiki. Our specific conclusions are as follows.

  • There is no functionality to fork repos and submit pull requests in the web interface, so you have to manually create new repositories and send merge requests some other way.

  • The management functionality includes useful repository grouping (projects and project groups) and users (organisations).

  • It’s useful to be able to archive old projects.

  • There’s a simple git-based wiki, that supports Textile and Markdown.

  • There’s a simple issue tracker.

  • Codebase’s pricing model is a better fit for our situation (many repos) than GitHub.

  • GitHub is more mature, seems to generally do more and have more development effort behind it.

Our conclusion is therefore that we want to use GitHub, even though it’s going to cost more and we’ll be limited in the number of repositories.