1.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     1.2 +++ b/en/collab.tex	Mon Apr 16 14:22:25 2007 -0700
     1.3 @@ -0,0 +1,236 @@
     1.4 +\chapter{Collaborating with other people}
     1.5 +\label{cha:collab}
     1.6 +
     1.7 +As a completely decentralised tool, Mercurial doesn't impose any
     1.8 +policy on how people ought to work with each other.  However, if
     1.9 +you're new to distributed revision control, it helps to have some
    1.10 +tools and examples in mind when you're thinking about possible
    1.11 +workflow models.
    1.12 +
    1.13 +\section{Collaboration models}
    1.14 +
    1.15 +With a suitably flexible tool, making decisions about workflow is much
    1.16 +more of a social engineering challenge than a technical one.
    1.17 +Mercurial imposes few limitations on how you can structure the flow of
    1.18 +work in a project, so it's up to you and your group to set up and live
    1.19 +with a model that matches your own particular needs.
    1.20 +
    1.21 +\subsection{Factors to keep in mind}
    1.22 +
    1.23 +The most important aspect of any model that you must keep in mind is
    1.24 +how well it matches the needs and capabilities of the people who will
    1.25 +be using it.  This might seem self-evident; even so, you still can't
    1.26 +afford to forget it for a moment.
    1.27 +
    1.28 +I once put together a workflow model that seemed to make perfect sense
    1.29 +to me, but that caused a considerable amount of consternation and
    1.30 +strife within my development team.  In spite of my attempts to explain
    1.31 +why we needed a complex set of branches, and how changes ought to flow
    1.32 +between them, a few team members revolted.  Even though they were
    1.33 +smart people, they didn't want to pay attention to the constraints we
    1.34 +were operating under, or face the consequences of those constraints in
    1.35 +the details of the model that I was advocating.
    1.36 +
    1.37 +Don't sweep foreseeable social or technical problems under the rug.
    1.38 +Whatever scheme you put into effect, you should plan for mistakes and
    1.39 +problem scenarios.  Consider adding automated machinery to prevent, or
    1.40 +quickly recover from, trouble that you can anticipate.  As an example,
    1.41 +if you intend to have a branch with not-for-release changes in it,
    1.42 +you'd do well to think early about the possibility that someone might
    1.43 +accidentally merge those changes into a release branch.  You could
    1.44 +avoid this particular problem by writing a hook that prevents changes
    1.45 +from being merged from an inappropriate branch.
    1.46 +
    1.47 +\subsection{Informal anarchy}
    1.48 +
    1.49 +I wouldn't suggest an ``anything goes'' approach as something
    1.50 +sustainable, but it's a model that's easy to grasp, and it works
    1.51 +perfectly well in a few unusual situations.
    1.52 +
    1.53 +As one example, many projects have a loose-knit group of collaborators
    1.54 +who rarely physically meet each other.  Some groups like to overcome
    1.55 +the isolation of working at a distance by organising occasional
    1.56 +``sprints''.  In a sprint, a number of people get together in a single
    1.57 +location (a company's conference room, a hotel meeting room, that kind
    1.58 +of place) and spend several days more or less locked in there, hacking
    1.59 +intensely on a handful of projects.
    1.60 +
    1.61 +A sprint is the perfect place to use the \hgcmd{serve} command, since
    1.62 +\hgcmd{serve} does not requires any fancy server infrastructure.  You
    1.63 +can get started with \hgcmd{serve} in moments, by reading
    1.64 +section~\ref{sec:collab:serve} below.  Then simply tell the person
    1.65 +next to you that you're running a server, send the URL to them in an
    1.66 +instant message, and you immediately have a quick-turnaround way to
    1.67 +work together.  They can type your URL into their web browser and
    1.68 +quickly review your changes; or they can pull a bugfix from you and
    1.69 +verify it; or they can clone a branch containing a new feature and try
    1.70 +it out.
    1.71 +
    1.72 +The charm, and the problem, with doing things in an ad hoc fashion
    1.73 +like this is that only people who know about your changes, and where
    1.74 +they are, can see them.  Such an informal approach simply doesn't
    1.75 +scale beyond a handful people, because each individual needs to know
    1.76 +about $n$ different repositories to pull from.
    1.77 +
    1.78 +\subsection{A single central repository}
    1.79 +
    1.80 +For smaller projects, migrating from a centralised revision control
    1.81 +tool, perhaps the easiest way to get started is to have changes flow
    1.82 +through a single shared central repository.  This is also the
    1.83 +most common ``building block'' for more ambitious workflow schemes.
    1.84 +
    1.85 +Contributors start by cloning a copy of this repository.  They can
    1.86 +pull changes from it whenever they need to, and some (perhaps all)
    1.87 +developers have permission to push a change back when they're ready
    1.88 +for other people to see it.
    1.89 +
    1.90 +Under this model, it can still sometimes make sense for people to pull
    1.91 +changes directly from each other, without going through the central
    1.92 +repository.  Consider a case in which I have a tentative bug fix, but
    1.93 +I am worried that if I were to publish it to the central repository,
    1.94 +it might subsequently break everyone else's trees as they pull it.  To
    1.95 +reduce the potential for damage, I can ask you to clone my repository
    1.96 +into a temporary repository of your own and test it.  This lets us put
    1.97 +off publishing the potentially unsafe change until it has had a little
    1.98 +testing.
    1.99 +
   1.100 +In this kind of scenario, people usually use the \command{ssh}
   1.101 +protocol to securely push changes to the central repository, as
   1.102 +documented in section~\ref{sec:collab:ssh}.  It's also usual to
   1.103 +publish a read-only copy of the repository over HTTP using CGI, as in
   1.104 +section~\ref{sec:collab:cgi}.  Publishing over HTTP satisfies the
   1.105 +needs of people who don't have push access, and those who want to use
   1.106 +web browsers to browse the repository's history.
   1.107 +
   1.108 +\subsection{The Linux kernel model}
   1.109 +
   1.110 +The development of the Linux kernel has a shallow hierarchical
   1.111 +structure, surrounded by a cloud of apparent chaos.  Because most
   1.112 +Linux developers use \command{git}, a distributed revision control
   1.113 +tool with capabilities similar to Mercurial, it's useful to describe
   1.114 +the way work flows in that environment; if you like the ideas, the
   1.115 +approach translates well across tools.
   1.116 +
   1.117 +At the center of the community sits Linus Torvalds, the creator of
   1.118 +Linux.  He publishes a single source repository that is considered the
   1.119 +``authoritative'' current tree by the entire developer community.
   1.120 +Anyone can clone Linus's tree, but he is very choosy about whose trees
   1.121 +he pulls from.
   1.122 +
   1.123 +Linus has a number of ``trusted lieutenants''.  As a general rule, he
   1.124 +pulls whatever changes they publish, in most cases without even
   1.125 +reviewing those changes.  Some of those lieutenants are generally
   1.126 +agreed to be ``maintainers'', responsible for specific subsystems
   1.127 +within the kernel.  If a random kernel hacker wants to make a change
   1.128 +to a subsystem that they want to end up in Linus's tree, they must
   1.129 +find out who the subsystem's maintainer is, and ask that maintainer to
   1.130 +take their change.  If the maintainer reviews their changes and agrees
   1.131 +to take them, they'll pass them along to Linus in due course.
   1.132 +
   1.133 +Individual lieutenants have their own approaches to reviewing,
   1.134 +accepting, and publishing changes; and for deciding when to feed them
   1.135 +to Linus.  In addition, there are several well known branches that
   1.136 +people use for different purposes.  For example, a few people maintain
   1.137 +``stable'' repositories of older versions of the kernel, to which they
   1.138 +apply critical fixes as needed.
   1.139 +
   1.140 +This model has two notable features.  The first is that it's ``pull
   1.141 +only''.  You have to ask, convince, or beg another developer to take a
   1.142 +change from you, because there are no shared trees, and there's no way
   1.143 +to push changes into a tree that someone else controls.
   1.144 +
   1.145 +The second is that it's based on reputation and acclaim.  If you're an
   1.146 +unknown, Linus will probably ignore changes from you without even
   1.147 +responding.  But a subsystem maintainer will probably review them, and
   1.148 +will likely take them if they pass their criteria for suitability.
   1.149 +The more ``good'' changes you contribute to a maintainer, the more
   1.150 +likely they are to trust your judgment and accept your changes.  If
   1.151 +you're well-known and maintain a long-lived branch for something Linus
   1.152 +hasn't yet accepted, people with similar interests may pull your
   1.153 +changes regularly to keep up with your work.
   1.154 +
   1.155 +Reputation and acclaim don't necessarily cross subsystem or ``people''
   1.156 +boundaries.  If you're a respected but specialised storage hacker, and
   1.157 +you try to fix a networking bug, that change will receive a level of
   1.158 +scrutiny from a network maintainer comparable to a change from a
   1.159 +complete stranger.
   1.160 +
   1.161 +To people who come from more orderly project backgrounds, the
   1.162 +comparatively chaotic Linux kernel development process often seems
   1.163 +completely insane.  It's subject to the whims of individuals; people
   1.164 +make sweeping changes whenever they deem it appropriate; and the pace
   1.165 +of development is astounding.  And yet Linux is a highly successful,
   1.166 +well-regarded piece of software.
   1.167 +
   1.168 +\section{The technical side of sharing}
   1.169 +
   1.170 +\subsection{Informal sharing with \hgcmd{serve}}
   1.171 +\label{sec:collab:serve}
   1.172 +
   1.173 +Mercurial's \hgcmd{serve} command is wonderfully suited to small,
   1.174 +tight-knit, and fast-paced group environments.  It also provides a
   1.175 +great way to get a feel for using Mercurial commands over a network.
   1.176 +
   1.177 +Run \hgcmd{serve} inside a repository, and in under a second it will
   1.178 +bring up a specialised HTTP server; this will accept connections from
   1.179 +any client, and serve up data for that repository until you terminate
   1.180 +it.  Anyone who knows the URL of the server you just started, and can
   1.181 +talk to your computer over the network, can then use a web browser or
   1.182 +Mercurial to read data from that repository.  A URL for a
   1.183 +\hgcmd{serve} instance running on a laptop is likely to look something
   1.184 +like \Verb|http://my-laptop.local:8000/|.
   1.185 +
   1.186 +The \hgcmd{serve} command is \emph{not} a general-purpose web server.
   1.187 +It can do only two things:
   1.188 +\begin{itemize}
   1.189 +\item Allow people to browse the history of the repository it's
   1.190 +  serving, from their normal web browsers.
   1.191 +\item Speak Mercurial's wire protocol, so that people can
   1.192 +  \hgcmd{clone} or \hgcmd{pull} changes from that repository.
   1.193 +\end{itemize}
   1.194 +In particular, \hgcmd{serve} won't allow remote users to \emph{modify}
   1.195 +your repository.  It's intended for read-only use.
   1.196 +
   1.197 +If you're getting started with Mercurial, there's nothing to prevent
   1.198 +you from using \hgcmd{serve} to serve up a repository on your own
   1.199 +computer, then use commands like \hgcmd{clone}, \hgcmd{incoming}, and
   1.200 +so on to talk to that server as if the repository was hosted remotely.
   1.201 +This can help you to quickly get acquainted with using commands on
   1.202 +network-hosted repositories.
   1.203 +
   1.204 +\subsubsection{A few things to keep in mind}
   1.205 +
   1.206 +Because it provides unauthenticated read access to all clients, you
   1.207 +should only use \hgcmd{serve} in an environment where you either don't
   1.208 +care, or have complete control over, who can access your network and
   1.209 +pull data from your repository.
   1.210 +
   1.211 +The \hgcmd{serve} command knows nothing about any firewall software
   1.212 +you might have installed on your system or network.  It cannot detect
   1.213 +or control your firewall software.  If other people are unable to talk
   1.214 +to a running \hgcmd{serve} instance, the second thing you should do
   1.215 +(\emph{after} you make sure that they're using the correct URL) is
   1.216 +check your firewall configuration.
   1.217 +
   1.218 +By default, \hgcmd{serve} listens for incoming connections on
   1.219 +port~8000.  If another process is already listening on the port you
   1.220 +want to use, you can specify a different port to listen on using the
   1.221 +\hgopt{serve}{-p} option.
   1.222 +
   1.223 +Normally, when \hgcmd{serve} starts, it prints no output, which can be
   1.224 +a bit unnerving.  If you'd like to confirm that it is indeed running
   1.225 +correctly, and find out what URL you should send to your
   1.226 +collaborators, start it with the \hggopt{-v} option.
   1.227 +
   1.228 +\subsection{Using \command{ssh} as a tunnel}
   1.229 +\label{sec:collab:ssh}
   1.230 +
   1.231 +\subsection{Serving HTTP with a CGI script}
   1.232 +\label{sec:collab:cgi}
   1.233 +
   1.234 +
   1.235 +
   1.236 +%%% Local Variables: 
   1.237 +%%% mode: latex
   1.238 +%%% TeX-master: "00book"
   1.239 +%%% End: