hgbook
view en/collab.tex @ 180:6413f88338df
Point to chapter on undoing mistakes.
| author | Bryan O'Sullivan <bos@serpentine.com> |
|---|---|
| date | Fri Mar 30 23:20:27 2007 -0700 (2007-03-30) |
| parents | 7355af913937 |
| children | 7b812c428074 |
line source
1 \chapter{Collaborating with other people}
2 \label{cha:collab}
4 As a completely decentralised tool, Mercurial doesn't impose any
5 policy on how people ought to work with each other. However, if
6 you're new to distributed revision control, it helps to have some
7 tools and examples in mind when you're thinking about possible
8 workflow models.
10 \section{Collaboration models}
12 With a suitably flexible tool, making decisions about workflow is much
13 more of a social engineering challenge than a technical one.
14 Mercurial imposes few limitations on how you can structure the flow of
15 work in a project, so it's up to you and your group to set up and live
16 with a model that matches your own particular needs.
18 \subsection{Factors to keep in mind}
20 The most important aspect of any model that you must keep in mind is
21 how well it matches the needs and capabilities of the people who will
22 be using it. This might seem self-evident; even so, you still can't
23 afford to forget it for a moment.
25 I once put together a workflow model that seemed to make perfect sense
26 to me, but that caused a considerable amount of consternation and
27 strife within my development team. In spite of my attempts to explain
28 why we needed a complex set of branches, and how changes ought to flow
29 between them, a few team members revolted. Even though they were
30 smart people, they didn't want to pay attention to the constraints we
31 were operating under, or face the consequences of those constraints in
32 the details of the model that I was advocating.
34 Don't sweep foreseeable social or technical problems under the rug.
35 Whatever scheme you put into effect, you should plan for mistakes and
36 problem scenarios. Consider adding automated machinery to prevent, or
37 quickly recover from, trouble that you can anticipate. As an example,
38 if you intend to have a branch with not-for-release changes in it,
39 you'd do well to think early about the possibility that someone might
40 accidentally merge those changes into a release branch. You could
41 avoid this particular problem by writing a hook that prevents changes
42 from being merged from an inappropriate branch.
44 \subsection{Informal anarchy}
46 I wouldn't suggest an ``anything goes'' approach as something
47 sustainable, but it's a model that's easy to grasp, and it works
48 perfectly well in a few unusual situations.
50 As one example, many projects have a loose-knit group of collaborators
51 who rarely physically meet each other. Some groups like to overcome
52 the isolation of working at a distance by organising occasional
53 ``sprints''. In a sprint, a number of people get together in a single
54 location (a company's conference room, a hotel meeting room, that kind
55 of place) and spend several days more or less locked in there, hacking
56 intensely on a handful of projects.
58 A sprint is the perfect place to use the \hgcmd{serve} command, since
59 \hgcmd{serve} does not requires any fancy server infrastructure. You
60 can get started with \hgcmd{serve} in moments, by reading
61 section~\ref{sec:collab:serve} below. Then simply tell the person
62 next to you that you're running a server, send the URL to them in an
63 instant message, and you immediately have a quick-turnaround way to
64 work together. They can type your URL into their web browser and
65 quickly review your changes; or they can pull a bugfix from you and
66 verify it; or they can clone a branch containing a new feature and try
67 it out.
69 The charm, and the problem, with doing things in an ad hoc fashion
70 like this is that only people who know about your changes, and where
71 they are, can see them. Such an informal approach simply doesn't
72 scale beyond a handful people, because each individual needs to know
73 about $n$ different repositories to pull from.
75 \subsection{A single central repository}
77 For smaller projects migrating from a centralised revision control
78 tool, perhaps the easiest way to get started is to have changes flow
79 through a single shared central repository. This is also the
80 most common ``building block'' for more ambitious workflow schemes.
82 Contributors start by cloning a copy of this repository. They can
83 pull changes from it whenever they need to, and some (perhaps all)
84 developers have permission to push a change back when they're ready
85 for other people to see it.
87 Under this model, it can still often make sense for people to pull
88 changes directly from each other, without going through the central
89 repository. Consider a case in which I have a tentative bug fix, but
90 I am worried that if I were to publish it to the central repository,
91 it might subsequently break everyone else's trees as they pull it. To
92 reduce the potential for damage, I can ask you to clone my repository
93 into a temporary repository of your own and test it. This lets us put
94 off publishing the potentially unsafe change until it has had a little
95 testing.
97 In this kind of scenario, people usually use the \command{ssh}
98 protocol to securely push changes to the central repository, as
99 documented in section~\ref{sec:collab:ssh}. It's also usual to
100 publish a read-only copy of the repository over HTTP using CGI, as in
101 section~\ref{sec:collab:cgi}. Publishing over HTTP satisfies the
102 needs of people who don't have push access, and those who want to use
103 web browsers to browse the repository's history.
105 \subsection{Working with multiple branches}
107 Projects of any significant size naturally tend to make progress on
108 several fronts simultaneously. In the case of software, it's common
109 for a project to go through periodic official releases. A release
110 might then go into ``maintenance mode'' for a while after its first
111 publication; maintenance releases tend to contain only bug fixes, not
112 new features. In parallel with these maintenance releases, one or
113 more future releases may be under development. People normally use
114 the word ``branch'' to refer to one of these many slightly different
115 directions in which development is proceeding.
117 Mercurial is particularly well suited to managing a number of
118 simultaneous, but not identical, branches. Each ``development
119 direction'' can live in its own central repository, and you can merge
120 changes from one to another as the need arises. Because repositories
121 are independent of each other, unstable changes in a development
122 branch will never affect a stable branch unless someone explicitly
123 merges those changes in.
125 Here's an example of how this can work in practice. Let's say you
126 have one ``main branch'' on a central server.
127 \interaction{branching.init}
128 People clone it, make changes locally, test them, and push them back.
130 Once the main branch reaches a release milestone, you can use the
131 \hgcmd{tag} command to give a permanent name to the milestone
132 revision.
133 \interaction{branching.tag}
134 Let's say some ongoing development occurs on the main branch.
135 \interaction{branching.main}
136 Using the tag that was recorded at the milestone, people who clone
137 that repository at any time in the future can use \hgcmd{update} to
138 get a copy of the working directory exactly as it was when that tagged
139 revision was committed.
140 \interaction{branching.update}
142 In addition, immediately after the main branch is tagged, someone can
143 then clone the main branch on the server to a new ``stable'' branch,
144 also on the server.
145 \interaction{branching.clone}
147 Someone who needs to make a change to the stable branch can then clone
148 \emph{that} repository, make their changes, commit, and push their
149 changes back there.
150 \interaction{branching.stable}
151 Because Mercurial repositories are independent, and Mercurial doesn't
152 move changes around automatically, the stable and main branches are
153 \emph{isolated} from each other. The changes that you made on the
154 main branch don't ``leak'' to the stable branch, and vice versa.
156 You'll often want all of your bugfixes on the stable branch to show up
157 on the main branch, too. Rather than rewrite a bugfix on the main
158 branch, you can simply pull and merge changes from the stable to the
159 main branch, and Mercurial will bring those bugfixes in for you.
160 \interaction{branching.merge}
161 The main branch will still contain changes that are not on the stable
162 branch, but it will also contain all of the bugfixes from the stable
163 branch. The stable branch remains unaffected by these changes.
165 \subsection{Feature branches}
167 For larger projects, an effective way to manage change is to break up
168 a team into smaller groups. Each group has a shared branch of its
169 own, cloned from a single ``master'' branch used by the entire
170 project. People working on an individual branch are typically quite
171 isolated from developments on other branches.
173 \begin{figure}[ht]
174 \centering
175 \grafix{feature-branches}
176 \caption{Feature branches}
177 \label{fig:collab:feature-branches}
178 \end{figure}
180 When a particular feature is deemed to be in suitable shape, someone
181 on that feature team pulls and merges from the master branch into the
182 feature branch, then pushes back up to the master branch.
184 \subsection{The release train}
186 Some projects are organised on a ``train'' basis: a release is
187 scheduled to happen every few months, and whatever features are ready
188 when the ``train'' is ready to leave are allowed in.
190 This model resembles working with feature branches. The difference is
191 that when a feature branch misses a train, someone on the feature team
192 pulls and merges the changes that went out on that train release, and
193 the team continues its work on top of that release so that their
194 feature can make the next release.
196 \subsection{The Linux kernel model}
198 The development of the Linux kernel has a shallow hierarchical
199 structure, surrounded by a cloud of apparent chaos. Because most
200 Linux developers use \command{git}, a distributed revision control
201 tool with capabilities similar to Mercurial, it's useful to describe
202 the way work flows in that environment; if you like the ideas, the
203 approach translates well across tools.
205 At the center of the community sits Linus Torvalds, the creator of
206 Linux. He publishes a single source repository that is considered the
207 ``authoritative'' current tree by the entire developer community.
208 Anyone can clone Linus's tree, but he is very choosy about whose trees
209 he pulls from.
211 Linus has a number of ``trusted lieutenants''. As a general rule, he
212 pulls whatever changes they publish, in most cases without even
213 reviewing those changes. Some of those lieutenants are generally
214 agreed to be ``maintainers'', responsible for specific subsystems
215 within the kernel. If a random kernel hacker wants to make a change
216 to a subsystem that they want to end up in Linus's tree, they must
217 find out who the subsystem's maintainer is, and ask that maintainer to
218 take their change. If the maintainer reviews their changes and agrees
219 to take them, they'll pass them along to Linus in due course.
221 Individual lieutenants have their own approaches to reviewing,
222 accepting, and publishing changes; and for deciding when to feed them
223 to Linus. In addition, there are several well known branches that
224 people use for different purposes. For example, a few people maintain
225 ``stable'' repositories of older versions of the kernel, to which they
226 apply critical fixes as needed.
228 This model has two notable features. The first is that it's ``pull
229 only''. You have to ask, convince, or beg another developer to take a
230 change from you, because there are no shared trees, and there's no way
231 to push changes into a tree that someone else controls.
233 The second is that it's based on reputation and acclaim. If you're an
234 unknown, Linus will probably ignore changes from you without even
235 responding. But a subsystem maintainer will probably review them, and
236 will likely take them if they pass their criteria for suitability.
237 The more ``good'' changes you contribute to a maintainer, the more
238 likely they are to trust your judgment and accept your changes. If
239 you're well-known and maintain a long-lived branch for something Linus
240 hasn't yet accepted, people with similar interests may pull your
241 changes regularly to keep up with your work.
243 Reputation and acclaim don't necessarily cross subsystem or ``people''
244 boundaries. If you're a respected but specialised storage hacker, and
245 you try to fix a networking bug, that change will receive a level of
246 scrutiny from a network maintainer comparable to a change from a
247 complete stranger.
249 To people who come from more orderly project backgrounds, the
250 comparatively chaotic Linux kernel development process often seems
251 completely insane. It's subject to the whims of individuals; people
252 make sweeping changes whenever they deem it appropriate; and the pace
253 of development is astounding. And yet Linux is a highly successful,
254 well-regarded piece of software.
256 \section{The technical side of sharing}
258 \subsection{Informal sharing with \hgcmd{serve}}
259 \label{sec:collab:serve}
261 Mercurial's \hgcmd{serve} command is wonderfully suited to small,
262 tight-knit, and fast-paced group environments. It also provides a
263 great way to get a feel for using Mercurial commands over a network.
265 Run \hgcmd{serve} inside a repository, and in under a second it will
266 bring up a specialised HTTP server; this will accept connections from
267 any client, and serve up data for that repository until you terminate
268 it. Anyone who knows the URL of the server you just started, and can
269 talk to your computer over the network, can then use a web browser or
270 Mercurial to read data from that repository. A URL for a
271 \hgcmd{serve} instance running on a laptop is likely to look something
272 like \Verb|http://my-laptop.local:8000/|.
274 The \hgcmd{serve} command is \emph{not} a general-purpose web server.
275 It can do only two things:
276 \begin{itemize}
277 \item Allow people to browse the history of the repository it's
278 serving, from their normal web browsers.
279 \item Speak Mercurial's wire protocol, so that people can
280 \hgcmd{clone} or \hgcmd{pull} changes from that repository.
281 \end{itemize}
282 In particular, \hgcmd{serve} won't allow remote users to \emph{modify}
283 your repository. It's intended for read-only use.
285 If you're getting started with Mercurial, there's nothing to prevent
286 you from using \hgcmd{serve} to serve up a repository on your own
287 computer, then use commands like \hgcmd{clone}, \hgcmd{incoming}, and
288 so on to talk to that server as if the repository was hosted remotely.
289 This can help you to quickly get acquainted with using commands on
290 network-hosted repositories.
292 \subsubsection{A few things to keep in mind}
294 Because it provides unauthenticated read access to all clients, you
295 should only use \hgcmd{serve} in an environment where you either don't
296 care, or have complete control over, who can access your network and
297 pull data from your repository.
299 The \hgcmd{serve} command knows nothing about any firewall software
300 you might have installed on your system or network. It cannot detect
301 or control your firewall software. If other people are unable to talk
302 to a running \hgcmd{serve} instance, the second thing you should do
303 (\emph{after} you make sure that they're using the correct URL) is
304 check your firewall configuration.
306 By default, \hgcmd{serve} listens for incoming connections on
307 port~8000. If another process is already listening on the port you
308 want to use, you can specify a different port to listen on using the
309 \hgopt{serve}{-p} option.
311 Normally, when \hgcmd{serve} starts, it prints no output, which can be
312 a bit unnerving. If you'd like to confirm that it is indeed running
313 correctly, and find out what URL you should send to your
314 collaborators, start it with the \hggopt{-v} option.
316 \subsection{Using \command{ssh} as a tunnel}
317 \label{sec:collab:ssh}
319 \subsection{Serving HTTP with a CGI script}
320 \label{sec:collab:cgi}
324 %%% Local Variables:
325 %%% mode: latex
326 %%% TeX-master: "00book"
327 %%% End: