Source code management on UNIX and Linux systems
Identifying and tracking the changes made by multiple developers and merging them into a single, up-to-date codebase makes collaborative, multi-developer projects possible. VCS software, also referred to as revision control systems (RCS) or source code management (SCM) systems, enable multiple users to submit changes to the same files or project without one developer's changes accidentally overwriting another's changes.
Linux® and UNIX® systems are knee-deep in VCS software, ranging from dinosaurs such as the RCS and the Concurrent Versions System (CVS) to more modern systems such as Arch, Bazaar, Git, Subversion, and Mercurial. Like Git, Mercurial began life as an open source replacement for a commercial source code management system called BitKeeper, which was used to maintain and manage the source code for the Linux kernel. Since its inception, Mercurial has evolved into a popular VCS system that is used by many open source and commercial projects. Projects using Mercurial include Mozilla, IcedTea, and the MoinMoin wiki. See Resources for links to these and many more examples.
VCS systems generally refer to each collection of source code in which changes can be made and tracked as a repository. How developers interact with a repository is the key difference between more traditional VCS systems such as CVS and Subversion, referred to as centralized VCS systems, and more flexible VCS systems such as Mercurial and Git, which are referred to as distributed VCS systems. Developers interact with centralized VCS systems using a client/server model, where changes to your local copy of the source code can only be pushed back to the central repository. Developers interact with distributed VCS systems using a peer-to-peer model, where any copy of the central repository is itself a repository to which changes can be committed and from which they can be shared with any other copy. Distributed VCS systems do not actually have the notion of a central, master repository, but one is almost always defined by policy so that a single repository exists for building, testing, and maintaining a master version of your software.
Mercurial is a small, powerful distributed VCS system that is easy to get started with, while still providing the advanced commands that VCS power users may need (or want) to use. Mercurial's distributed nature makes it easy to work on projects locally, tracking and managing your changes via local commits and pushing those changes to remote repositories whenever necessary.
Among modern, distributed VCS systems, the closest VCS system to Mercurial is Git. Some differences between Mercurial and Git are the following:
- Multiple, built-in undo operations: Mercurial's
rollbackcommands make it easy to return to previous versions of specific files or previous sets of committed changes. Git provides a single built-in
revertcommand with its typical rocket-scientist-only syntax.
- Built-in web server: Mercurial provides a simple, integrated web server that makes it easy to host a repository quickly for others to pull from. Pushing requires either ignoring security or a more complex setup that supports Secure Sockets Layer (SSL).
- History preservation during copy/move operations:
movecommands both preserve complete history information, while Git does not preserve history in either case.
- Branches: Mercurial automatically shares all branches, while Git requires that each repository set up its own branches (either creating them locally or by mapping them to specific branches in a remote repository).
- Global and local tags: Mercurial supports global tags that are shared between repositories, which make it easy to share information about specific points in code development without branching.
- Native support on Windows platforms: Mercurial is written in Python, which is supported on Microsoft® Windows® systems. Mercurial is therefore available as a Windows executable (see Resources). Git on Windows is more complex—your choices are msysGit, using standard git under Cygwin, or using a web-based hosting system and repository.
- Automatic repository packing: Git requires that you explicitly pack and garbage-collect its repositories, while Mercurial performs its equivalent operations automatically. However, Mercurial repositories tend to be larger than Git repositories for the same codebase.
Mercurial and Git fans are also happy to discuss the learning curve, merits, and usability of each VCS system's command set. Space prevents that discussion here, but a web search on that topic will provide lots of interesting reading material.
Creating and using Mercurial repositories
Mercurial provides two basic ways of creating a local repository for a project's source code: either by explicitly creating a repository or by cloning an existing, remote repository:
- To create a local repository, use the
hg init [REPO-NAME]command. Supplying the name of a repository when executing this command creates a directory for that repository in the specified location. Not supplying the name of a repository turns the current working directory into a repository. The latter is handy when creating a Mercurial repository for an existing codebase.
To clone an existing repository, use the
hg clone REPO-NAME[LOCALNAME]command. Mercurial supports the Hypertext Transfer Protocol (HTTP) and Secure Shell (SSH) protocols for accessing remote repositories. Listing 1 shows an example
hgcommand and the resulting output produced when cloning a repository via SSH.
Listing 1. Cloning a Mercurial repository via SSH
$ hg clone ssh://codeserver//home/wvh/src/pop3check wvh@codeserver's password: destination directory: pop3check requesting all changes adding changesets adding manifests adding file changes added 1 changesets with 12 changes to 12 files updating to branch default 12 files updated, 0 files merged, 0 files removed, 0 files unresolved remote: 1 changesets found
Note: To use the HTTP protocol to access Mercurial
repositories, you must either start Mercurial's internal web
server in that repository (
hg serve -d)
or use Mercurial's
hgweb.cgi script to
integrate Mercurial with an existing web server such as
Apache. When cloning via HTTP, you will usually want to specify a
name for your local repository.
After you create or clone a repository and make that repository your working directory, you're ready to start working with the code that it contains, add new files, and so on.
Getting help in Mercurial
Mercurial's primary command is
which supports a set of sub-commands that are similar to those in
other VCS systems. To see a list of the most common commands,
hg command with no
arguments, which displays output similar to that shown in Listing 2.
Listing 2. Basic commands provided by Mercurial
Mercurial Distributed SCM basic commands: add add the specified files on the next commit annotate show changeset information by line for each file clone make a copy of an existing repository commit commit the specified files or all outstanding changes diff diff repository (or selected files) export dump the header and diffs for one or more changesets forget forget the specified files on the next commit init create a new repository in the given directory log show revision history of entire repository or files merge merge working directory with another revision pull pull changes from the specified source push push changes to the specified destination remove remove the specified files on the next commit serve export the repository via HTTP status show changed files in the working directory summary summarize working directory state update update working directory use "hg help" for the full list of commands or "hg -v" for details
This short list displays only basic Mercurial commands. To
obtain a full list, execute the
Tip: You can obtain detailed help on any Mercurial command
by executing the
hg help COMMAND
command, replacing COMMAND with the name of any valid
Checking repository status
Checking in changes is the most common operation in any VCS
system. You can use the
command to see any pending changes to the files in your
repository. For example, after creating a new file or modifying an
existing one, you see output like that shown in Listing 3.
Listing 3. Status output from Mercurial
$ hg status M Makefile ? hgrc.example
In this case, the Makefile file is an existing file that has been
modified (indicated by the letter
the beginning of the line), while the hgrc.example file is a new
file that isn't being tracked (indicated by the question mark
?) at the beginning of the line.
Adding files to a repository
To add the hgrc.example file to the
list of files that are being tracked in this repository, use the
hg add command. Specifying one or more
file names as arguments explicitly adds those files to the list of
files that are being tracked by Mercurial. If you don't specify any
files, all new files are added to the repository, as shown in Listing 4.
Listing 4. Adding a file to your repository
$ hg add adding hgrc.example
Tip: To add automatically all new files and mark any files
that have been removed for permanent removal, you can use
hg addremove command.
Checking the status of the repository shows that the new file has
been added (indicated by the letter
at the beginning of the line), as shown in Listing 5.
Listing 5. Repository status after modifications
$ hg status M Makefile A hgrc.example
Checking in changes is the most common operation in any VCS system. After making and testing your changes, you're ready to commit those changes to the local repository.
Before committing changes for the first time
If this is your first Mercurial project, you must provide some
basic information so that Mercurial can identify the user who is
committing those changes. If you do not do so, you'll see a
message along the lines of
abort: no username
supplied... when you try to commit changes, and your
changes will not be committed.
To add your user information, create a file called .hgrc in your home directory. This file is your personal Mercurial configuration file. You need to add at least the basic user information shown in Listing 6 to this file.
Listing 6. Mandatory information in a user's .hgrc file
[ui] username = Firstname Lastname <email@example.com>
Replace Firstname and Lastname with your first and last names; replace firstname.lastname@example.org with your email address; save the modified file.
You can set default Mercurial configuration values that apply to all users (which should not include user-specific information) in the /etc/mercurial/hgrc file on Linux and UNIX systems and in the Mercurial.ini file on Microsoft Windows systems, where this file is located in the directory of the Mercurial installation.
The standard commit process
After creating or verifying your ~/.hgrc file, you can commit your changes
hg commit command,
identifying the specific files that you want to commit or
committing all pending changes by not supplying an argument, as in
the following example:
$ hg commit Makefile hgrc.example committed changeset 1:3d7faeb12722
As shown in this example output, Mercurial refers to all changes that are associated with a single commit as a changeset.
When you commit changes, Mercurial starts your default editor
to enable you to add a commit message. To avoid this, you can
specify a commit message on the command line using the
-m "Message.." option. To use a different
editor, you can add an
editor entry in
[ui] section of your ~/.hgrc file, following the
editor keyword with the name of the editor
that you want to use and any associated command-line options. For
example, after adding an entry for using
emacs in no-window mode as my default editor,
my ~/.hgrc file looks like that
shown in Listing 7.
Listing 7. Additional customization in a user's .hgrc file
[ui] username = William von Hagen <email@example.com> editor = emacs -nw
Tip: To maximize the amount of information that Mercurial
provides about its activities, you can add the
verbose = True entry to the
[ui] section of your Mercurial configuration
Pushing changes to a remote repository
If you are using a clone of a remote repository, you want to
push those changes back to that repository after committing
changes to your local repository. To do so, use Mercurial's
hg push command, as shown in Listing 8.
Listing 8. Pushing changes via SSH
$ hg push wvh@codeserver's password: pushing to ssh://codeserver//home/wvh/src/pop3check searching for changes 1 changesets found remote: adding changesets remote: adding manifests remote: adding file changes remote: added 1 changesets with 2 changes to 2 files
Pulling changes from a remote repository
If you are using a clone of a remote repository and other users
are also using that same repository, you want to retrieve the
changes that they have made and pushed to that repository. To do
so, use Mercurial's
hg pull command, as
shown in Listing 9.
Listing 9. Pulling changes via SSH
$ hg pull wvh@codeserver's password: pulling from ssh://codeserver//home/wvh/src/pop3check searching for changes adding changesets adding manifests adding file changes added 1 changesets with 0 changes to 0 files (run 'hg update' to get a working copy) remote: 1 changesets found
As shown in the output from this command, this command only
retrieves information about remote changes—you must run the
hg update command to show the
associated changes in your local repository. This command
identifies the ways the repository has been updated, as
shown in Listing 10.
Listing 10. Updating your repository to show changes
$ hg update 0 files updated, 0 files merged, 1 files removed, 0 files unresolved
Undoing changes in Mercurial
Mercurial provides the following built-in commands that make it easy to undo committed changes:
hg backout CHANGESET: Undoes a specific changeset and creates a changeset that undoes that changeset. Unless you specify the
--mergeoption when executing this command, you have to merge that changeset into your current revision to push it back to a remote repository.
hg revert: Returns to previous versions of one or more files by specifying their names or returning to the previous version of all files by specifying the
hg rollback: Undoes the last Mercurial transaction, which is commonly a
pullfrom a remote repository, or a
pushto this repository. You can only undo a single transaction.
See the online help for all of these commands before attempting to use them!
Mercurial and other distributed source code management systems are the wave of the future. Mercurial is open source software, and pre-compiled versions of Mercurial are available for Linux, UNIX, Microsoft Windows, and Mac OS® X systems. This article highlighted how to use Mercurial to perform a number of common VCS tasks, showing how easy it is to get started using Mercurial. For more advanced purposes, Mercurial provides many more advanced commands and configuration options to help you manage your source code and customize your interaction with a Mercurial installation.
- The Mercurial home page is a great starting point for getting information about Mercurial, and it also provides links to many other sources of related information.
- Mercurial: The Definitive Guide by Bryan O'Sullivan is the definitive work on Mercurial. The complete text of the book is available online in HTML and epub formats, which should hold you until your paper copy comes in the mail.
- Joel Spolsky's hginit.com site provides a great introductory tutorial for using and working with Mercurial.
- Projects using Mercurial include Mozilla, IcedTea, and the MoinMoin wiki. See the list at the Mercurial site for many more examples.
- The Linux developerWorks zone and AIX and UNIX developerWorks zone provide a wealth of information relating to all aspects of Linux, AIX, and general UNIX systems administration and expanding your AIX, Linux, and UNIX skills.
- New to Linux or New to AIX and UNIX? Visit the appropriate New to page to learn more.
- Browse the technology bookstore for books on this and other technical topics.
Get products and technologies
- The Mercurial wiki's Download page provides links to compiled versions of Mercurial for all supported platforms.
- TortiseHG provides a shell extension and command-line Mercurial applications for Microsoft Windows systems.
- The MercurialEclipse plug-in provides support for Mercurial within the Eclipse Integrated Development Environment.
- FogCreek Software's Kiln provides free trials and student/start-up versions of its online, Mercurial-based hosting service that is similar to online Git hosting services such as GitHub, Repo.Org.Cz, and so on.
- BitBucket provides free hosting for Open Source projects in its online, Mercurial-based hosting service, as well as paid hosting plans for larger groups of developers.
- Explore msysGit, which is Git for Windows.
- Check out developerWorks blogs and get involved in the developerWorks community.
- Participate in the AIX, Linux, and UNIX forums: