bos@159: \chapter{Collaborating with other people}
bos@159: \label{cha:collab}
bos@159: 
bos@159: As a completely decentralised tool, Mercurial doesn't impose any
bos@159: policy on how people ought to work with each other.  However, if
bos@159: you're new to distributed revision control, it helps to have some
bos@159: tools and examples in mind when you're thinking about possible
bos@159: workflow models.
bos@159: 
bos@159: \section{Collaboration models}
bos@159: 
bos@159: With a suitably flexible tool, making decisions about workflow is much
bos@159: more of a social engineering challenge than a technical one.
bos@159: Mercurial imposes few limitations on how you can structure the flow of
bos@159: work in a project, so it's up to you and your group to set up and live
bos@159: with a model that matches your own particular needs.
bos@159: 
bos@159: \subsection{Factors to keep in mind}
bos@159: 
bos@159: The most important aspect of any model that you must keep in mind is
bos@159: how well it matches the needs and capabilities of the people who will
bos@159: be using it.  This might seem self-evident; even so, you still can't
bos@159: afford to forget it for a moment.
bos@159: 
bos@159: I once put together a workflow model that seemed to make perfect sense
bos@159: to me, but that caused a considerable amount of consternation and
bos@159: strife within my development team.  In spite of my attempts to explain
bos@159: why we needed a complex set of branches, and how changes ought to flow
bos@159: between them, a few team members revolted.  Even though they were
bos@159: smart people, they didn't want to pay attention to the constraints we
bos@159: were operating under, or face the consequences of those constraints in
bos@159: the details of the model that I was advocating.
bos@159: 
bos@159: Don't sweep foreseeable social or technical problems under the rug.
bos@159: Whatever scheme you put into effect, you should plan for mistakes and
bos@159: problem scenarios.  Consider adding automated machinery to prevent, or
bos@159: quickly recover from, trouble that you can anticipate.  As an example,
bos@159: if you intend to have a branch with not-for-release changes in it,
bos@159: you'd do well to think early about the possibility that someone might
bos@159: accidentally merge those changes into a release branch.  You could
bos@159: avoid this particular problem by writing a hook that prevents changes
bos@159: from being merged from an inappropriate branch.
bos@159: 
bos@159: \subsection{Informal anarchy}
bos@159: 
bos@159: I wouldn't suggest an ``anything goes'' approach as something
bos@159: sustainable, but it's a model that's easy to grasp, and it works
bos@159: perfectly well in a few unusual situations.
bos@159: 
bos@159: As one example, many projects have a loose-knit group of collaborators
bos@159: who rarely physically meet each other.  Some groups like to overcome
bos@159: the isolation of working at a distance by organising occasional
bos@159: ``sprints''.  In a sprint, a number of people get together in a single
bos@159: location (a company's conference room, a hotel meeting room, that kind
bos@159: of place) and spend several days more or less locked in there, hacking
bos@159: intensely on a handful of projects.
bos@159: 
bos@159: A sprint is the perfect place to use the \hgcmd{serve} command, since
bos@159: \hgcmd{serve} does not requires any fancy server infrastructure.  You
bos@159: can get started with \hgcmd{serve} in moments, by reading
bos@159: section~\ref{sec:collab:serve} below.  Then simply tell the person
bos@159: next to you that you're running a server, send the URL to them in an
bos@159: instant message, and you immediately have a quick-turnaround way to
bos@159: work together.  They can type your URL into their web browser and
bos@159: quickly review your changes; or they can pull a bugfix from you and
bos@159: verify it; or they can clone a branch containing a new feature and try
bos@159: it out.
bos@159: 
bos@159: The charm, and the problem, with doing things in an ad hoc fashion
bos@159: like this is that only people who know about your changes, and where
bos@159: they are, can see them.  Such an informal approach simply doesn't
bos@159: scale beyond a handful people, because each individual needs to know
bos@159: about $n$ different repositories to pull from.
bos@159: 
bos@159: \subsection{A single central repository}
bos@159: 
bos@159: For smaller projects, migrating from a centralised revision control
bos@159: tool, perhaps the easiest way to get started is to have changes flow
bos@159: through a single shared central repository.  This is also the
bos@159: most common ``building block'' for more ambitious workflow schemes.
bos@159: 
bos@159: Contributors start by cloning a copy of this repository.  They can
bos@159: pull changes from it whenever they need to, and some (perhaps all)
bos@159: developers have permission to push a change back when they're ready
bos@159: for other people to see it.
bos@159: 
bos@159: Under this model, it can still sometimes make sense for people to pull
bos@159: changes directly from each other, without going through the central
bos@159: repository.  Consider a case in which I have a tentative bug fix, but
bos@159: I am worried that if I were to publish it to the central repository,
bos@159: it might subsequently break everyone else's trees as they pull it.  To
bos@159: reduce the potential for damage, I can ask you to clone my repository
bos@159: into a temporary repository of your own and test it.  This lets us put
bos@159: off publishing the potentially unsafe change until it has had a little
bos@159: testing.
bos@159: 
bos@159: In this kind of scenario, people usually use the \command{ssh}
bos@159: protocol to securely push changes to the central repository, as
bos@159: documented in section~\ref{sec:collab:ssh}.  It's also usual to
bos@159: publish a read-only copy of the repository over HTTP using CGI, as in
bos@159: section~\ref{sec:collab:cgi}.  Publishing over HTTP satisfies the
bos@159: needs of people who don't have push access, and those who want to use
bos@159: web browsers to browse the repository's history.
bos@159: 
bos@159: \subsection{The Linux kernel model}
bos@159: 
bos@159: The development of the Linux kernel has a shallow hierarchical
bos@159: structure, surrounded by a cloud of apparent chaos.  Because most
bos@159: Linux developers use \command{git}, a distributed revision control
bos@159: tool with capabilities similar to Mercurial, it's useful to describe
bos@159: the way work flows in that environment; if you like the ideas, the
bos@159: approach translates well across tools.
bos@159: 
bos@159: At the center of the community sits Linus Torvalds, the creator of
bos@159: Linux.  He publishes a single source repository that is considered the
bos@159: ``authoritative'' current tree by the entire developer community.
bos@159: Anyone can clone Linus's tree, but he is very choosy about whose trees
bos@159: he pulls from.
bos@159: 
bos@159: Linus has a number of ``trusted lieutenants''.  As a general rule, he
bos@159: pulls whatever changes they publish, in most cases without even
bos@159: reviewing those changes.  Some of those lieutenants are generally
bos@159: agreed to be ``maintainers'', responsible for specific subsystems
bos@159: within the kernel.  If a random kernel hacker wants to make a change
bos@159: to a subsystem that they want to end up in Linus's tree, they must
bos@159: find out who the subsystem's maintainer is, and ask that maintainer to
bos@159: take their change.  If the maintainer reviews their changes and agrees
bos@159: to take them, they'll pass them along to Linus in due course.
bos@159: 
bos@159: Individual lieutenants have their own approaches to reviewing,
bos@159: accepting, and publishing changes; and for deciding when to feed them
bos@159: to Linus.  In addition, there are several well known branches that
bos@159: people use for different purposes.  For example, a few people maintain
bos@159: ``stable'' repositories of older versions of the kernel, to which they
bos@159: apply critical fixes as needed.
bos@159: 
bos@159: This model has two notable features.  The first is that it's ``pull
bos@159: only''.  You have to ask, convince, or beg another developer to take a
bos@159: change from you, because there are no shared trees, and there's no way
bos@159: to push changes into a tree that someone else controls.
bos@159: 
bos@159: The second is that it's based on reputation and acclaim.  If you're an
bos@159: unknown, Linus will probably ignore changes from you without even
bos@159: responding.  But a subsystem maintainer will probably review them, and
bos@159: will likely take them if they pass their criteria for suitability.
bos@159: The more ``good'' changes you contribute to a maintainer, the more
bos@159: likely they are to trust your judgment and accept your changes.  If
bos@159: you're well-known and maintain a long-lived branch for something Linus
bos@159: hasn't yet accepted, people with similar interests may pull your
bos@159: changes regularly to keep up with your work.
bos@159: 
bos@159: Reputation and acclaim don't necessarily cross subsystem or ``people''
bos@159: boundaries.  If you're a respected but specialised storage hacker, and
bos@159: you try to fix a networking bug, that change will receive a level of
bos@159: scrutiny from a network maintainer comparable to a change from a
bos@159: complete stranger.
bos@159: 
bos@159: To people who come from more orderly project backgrounds, the
bos@159: comparatively chaotic Linux kernel development process often seems
bos@159: completely insane.  It's subject to the whims of individuals; people
bos@159: make sweeping changes whenever they deem it appropriate; and the pace
bos@159: of development is astounding.  And yet Linux is a highly successful,
bos@159: well-regarded piece of software.
bos@159: 
bos@159: \section{The technical side of sharing}
bos@159: 
bos@159: \subsection{Informal sharing with \hgcmd{serve}}
bos@159: \label{sec:collab:serve}
bos@159: 
bos@159: Mercurial's \hgcmd{serve} command is wonderfully suited to small,
bos@159: tight-knit, and fast-paced group environments.  It also provides a
bos@159: great way to get a feel for using Mercurial commands over a network.
bos@159: 
bos@159: Run \hgcmd{serve} inside a repository, and in under a second it will
bos@159: bring up a specialised HTTP server; this will accept connections from
bos@159: any client, and serve up data for that repository until you terminate
bos@159: it.  Anyone who knows the URL of the server you just started, and can
bos@159: talk to your computer over the network, can then use a web browser or
bos@159: Mercurial to read data from that repository.  A URL for a
bos@159: \hgcmd{serve} instance running on a laptop is likely to look something
bos@159: like \Verb|http://my-laptop.local:8000/|.
bos@159: 
bos@159: The \hgcmd{serve} command is \emph{not} a general-purpose web server.
bos@159: It can do only two things:
bos@159: \begin{itemize}
bos@159: \item Allow people to browse the history of the repository it's
bos@159:   serving, from their normal web browsers.
bos@159: \item Speak Mercurial's wire protocol, so that people can
bos@159:   \hgcmd{clone} or \hgcmd{pull} changes from that repository.
bos@159: \end{itemize}
bos@159: In particular, \hgcmd{serve} won't allow remote users to \emph{modify}
bos@159: your repository.  It's intended for read-only use.
bos@159: 
bos@159: If you're getting started with Mercurial, there's nothing to prevent
bos@159: you from using \hgcmd{serve} to serve up a repository on your own
bos@159: computer, then use commands like \hgcmd{clone}, \hgcmd{incoming}, and
bos@159: so on to talk to that server as if the repository was hosted remotely.
bos@159: This can help you to quickly get acquainted with using commands on
bos@159: network-hosted repositories.
bos@159: 
bos@159: \subsubsection{A few things to keep in mind}
bos@159: 
bos@159: Because it provides unauthenticated read access to all clients, you
bos@159: should only use \hgcmd{serve} in an environment where you either don't
bos@159: care, or have complete control over, who can access your network and
bos@159: pull data from your repository.
bos@159: 
bos@159: The \hgcmd{serve} command knows nothing about any firewall software
bos@159: you might have installed on your system or network.  It cannot detect
bos@159: or control your firewall software.  If other people are unable to talk
bos@159: to a running \hgcmd{serve} instance, the second thing you should do
bos@159: (\emph{after} you make sure that they're using the correct URL) is
bos@159: check your firewall configuration.
bos@159: 
bos@159: By default, \hgcmd{serve} listens for incoming connections on
bos@159: port~8000.  If another process is already listening on the port you
bos@159: want to use, you can specify a different port to listen on using the
bos@159: \hgopt{serve}{-p} option.
bos@159: 
bos@159: Normally, when \hgcmd{serve} starts, it prints no output, which can be
bos@159: a bit unnerving.  If you'd like to confirm that it is indeed running
bos@159: correctly, and find out what URL you should send to your
bos@159: collaborators, start it with the \hggopt{-v} option.
bos@159: 
bos@159: \subsection{Using \command{ssh} as a tunnel}
bos@159: \label{sec:collab:ssh}
bos@159: 
bos@159: \subsection{Serving HTTP with a CGI script}
bos@159: \label{sec:collab:cgi}
bos@159: 
bos@159: 
bos@159: 
bos@159: %%% Local Variables: 
bos@159: %%% mode: latex
bos@159: %%% TeX-master: "00book"
bos@159: %%% End: