rev |
line source |
jerojasro@393
|
1 \chapter{Tras bambalinas}
|
jerojasro@343
|
2 \label{chap:concepts}
|
jerojasro@343
|
3
|
jerojasro@393
|
4 A diferencia de varios sistemas de control de revisiones, los
|
jerojasro@393
|
5 conceptos en los que se fundamenta Mercurial son lo suficientemente
|
jerojasro@393
|
6 simples como para entender fácilmente cómo funciona el software.
|
jerojasro@393
|
7 Saber esto no es necesario, pero considero útil tener un ``modelo
|
jerojasro@393
|
8 mental'' de qué es lo que sucede.
|
jerojasro@393
|
9
|
jerojasro@393
|
10 Comprender esto me da la confianza de que Mercurial ha sido
|
jerojasro@393
|
11 cuidadosamente diseñado para ser tanto \emph{seguro} como
|
jerojasro@393
|
12 \emph{eficiente}. Y tal vez con la misma importancia, si es fácil
|
jerojasro@393
|
13 para mí hacerme a una idea adecuada de qué está haciendo el software
|
jerojasro@393
|
14 cuando llevo a cabo una tarea relacionada con control de revisiones,
|
jerojasro@393
|
15 es menos probable que me sosprenda su comportamiento.
|
jerojasro@393
|
16
|
jerojasro@393
|
17 En este capítulo, cubriremos inicialmente los conceptos centrales
|
jerojasro@393
|
18 del diseño de Mercurial, y luego discutiremos algunos detalles
|
jerojasro@393
|
19 interesantes de su implementación.
|
jerojasro@393
|
20
|
jerojasro@393
|
21 \section{Registro del historial de Mercurial}
|
jerojasro@393
|
22
|
jerojasro@393
|
23 \subsection{Seguir el historial de un único fichero}
|
jerojasro@393
|
24
|
jerojasro@393
|
25 Cuando Mercurial sigue las modificaciones a un fichero, guarda el
|
jerojasro@393
|
26 historial de dicho fichero en un objeto de metadatos llamado
|
jerojasro@395
|
27 \emph{filelog}\ndt{Fichero de registro}. Cada entrada en el fichero
|
jerojasro@395
|
28 de registro contiene suficiente información para reconstruir una
|
jerojasro@395
|
29 revisión del fichero que se está siguiendo. Los ficheros de registro
|
jerojasro@395
|
30 son almacenados como ficheros el el directorio
|
jerojasro@395
|
31 \sdirname{.hg/store/data}. Un fichero de registro contiene dos tipos
|
jerojasro@395
|
32 de información: datos de revisiones, y un índice para ayudar a
|
jerojasro@395
|
33 Mercurial a buscar revisiones eficientemente.
|
jerojasro@395
|
34
|
jerojasro@395
|
35 El fichero de registro de un fichero grande, o con un historial muy
|
jerojasro@395
|
36 largo, es guardado como ficheros separados para datos (sufijo
|
jerojasro@395
|
37 ``\texttt{.d}'') y para el índice (sufijo ``\texttt{.i}''). Para
|
jerojasro@395
|
38 ficheros pequeños con un historial pequeño, los datos de revisiones y
|
jerojasro@395
|
39 el índice son combinados en un único fichero ``\texttt{.i}''. La
|
jerojasro@395
|
40 correspondencia entre un fichero en el directorio de trabajo y el
|
jerojasro@395
|
41 fichero de registro que hace seguimiento a su historial en el
|
jerojasro@395
|
42 repositorio se ilustra en la figura~\ref{fig:concepts:filelog}.
|
jerojasro@343
|
43
|
jerojasro@343
|
44 \begin{figure}[ht]
|
jerojasro@343
|
45 \centering
|
jerojasro@343
|
46 \grafix{filelog}
|
jerojasro@396
|
47 \caption{Relación entre ficheros en el directorio de trabajo y
|
jerojasro@396
|
48 ficheros de registro en el repositorio}
|
jerojasro@343
|
49 \label{fig:concepts:filelog}
|
jerojasro@343
|
50 \end{figure}
|
jerojasro@343
|
51
|
jerojasro@396
|
52 \subsection{Administración de ficheros monitoreados}
|
jerojasro@396
|
53
|
jerojasro@396
|
54 Mercurial usa una estructura llamada \emph{manifiesto} para
|
jerojasro@396
|
55 % TODO collect together => centralizar
|
jerojasro@396
|
56 centralizar la información que maneja acerca de los ficheros que
|
jerojasro@396
|
57 monitorea. Cada entrada en el manifiesto contiene información acerca
|
jerojasro@396
|
58 de los ficheros involucrados en un único conjunto de cambios. Una
|
jerojasro@396
|
59 entrada registra qué ficheros están presentes en el conjunto de
|
jerojasro@396
|
60 cambios, la revisión de cada fichero, y otros cuantos metadatos del
|
jerojasro@396
|
61 mismo.
|
jerojasro@396
|
62
|
jerojasro@396
|
63 \subsection{Registro de información del conjunto de cambios}
|
jerojasro@396
|
64
|
jerojasro@396
|
65 La \emph{bitácora de cambios} contiene información acerca de cada
|
jerojasro@396
|
66 conjunto de cambios. Cada revisión indica quién consignó un cambio, el
|
jerojasro@396
|
67 comentario para el conjunto de cambios, otros datos relacionados con
|
jerojasro@396
|
68 el conjunto de cambios, y la revisión del manifiesto a usar.
|
jerojasro@396
|
69
|
jerojasro@396
|
70 \subsection{Relaciones entre revisiones}
|
jerojasro@396
|
71
|
jerojasro@396
|
72 Dentro de una bitácora de cambios, un manifiesto, o un fichero de
|
jerojasro@396
|
73 registro, cada revisión conserva un apuntador a su padre inmediato
|
jerojasro@396
|
74 (o sus dos padres, si es la revisión de una fusión). Como menciońe
|
jerojasro@396
|
75 anteriormente, también hay relaciones entre revisiones \emph{a través}
|
jerojasro@396
|
76 de estas estructuras, y tienen naturaleza jerárquica.
|
jerojasro@396
|
77
|
jerojasro@396
|
78 Por cada conjunto de cambios en un repositorio, hay exactamente una
|
jerojasro@396
|
79 revisión almacenada en la bitácora de cambios. Cada revisión de la
|
jerojasro@396
|
80 bitácora de cambios contiene un apuntador a una única revisión del
|
jerojasro@396
|
81 manifiesto. Una revisión del manifiesto almacena un apuntador a una
|
jerojasro@396
|
82 única revisión de cada fichero de registro al que se le hacía
|
jerojasro@396
|
83 seguimiento cuando fue creado el conjunto de cambios. Estas relaciones
|
jerojasro@396
|
84 se ilustran en la figura~\ref{fig:concepts:metadata}.
|
jerojasro@343
|
85
|
jerojasro@343
|
86 \begin{figure}[ht]
|
jerojasro@343
|
87 \centering
|
jerojasro@343
|
88 \grafix{metadata}
|
jerojasro@396
|
89 \caption{Relaciones entre metadatos}
|
jerojasro@343
|
90 \label{fig:concepts:metadata}
|
jerojasro@343
|
91 \end{figure}
|
jerojasro@343
|
92
|
jerojasro@406
|
93 Como lo muestra la figura, \emph{no} hay una relación ``uno a uno''
|
jerojasro@406
|
94 entre las revisiones en el conjunto de cambios, el manifiesto, o el
|
jerojasro@406
|
95 fichero de registro. Si el manifiesto no ha sido modificado de un
|
jerojasro@406
|
96 conjunto de cambios a otro, las entradas en la bitácora de cambios
|
jerojasro@406
|
97 para esos conjuntos de cambios apuntarán a la misma revisión del
|
jerojasro@406
|
98 manifiesto. Si un fichero monitoreado por Mercurial no sufre ningún
|
jerojasro@406
|
99 cambio de un conjunto de cambios a otro, la entrada para dicho fichero
|
jerojasro@406
|
100 en las dos revisiones del manifiesto apuntará a la misma revisión de
|
jerojasro@406
|
101 su fichero de registro.
|
jerojasro@406
|
102
|
jerojasro@406
|
103 \section{Almacenamiento seguro y eficiente}
|
jerojasro@406
|
104
|
jerojasro@406
|
105 La base común de las bitácoras de cambios, los manifiestos, y los
|
jerojasro@406
|
106 ficheros de registros es provista por una única estructura llamada el
|
jerojasro@406
|
107 \emph{revlog}\ndt{Contracción de \emph{revision log}, registro de
|
jerojasro@406
|
108 revisión.}.
|
jerojasro@406
|
109
|
jerojasro@406
|
110 \subsection{Almacenamiento eficiente}
|
jerojasro@406
|
111
|
jerojasro@406
|
112 El revlog provee almacenamiento eficiente de revisiones por medio del
|
jerojasro@406
|
113 mecanismo de \emph{deltas}\ndt{Diferencias.}. En vez de almacenar una
|
jerojasro@406
|
114 copia completa del fichero por cada revisión, almacena los cambios
|
jerojasro@406
|
115 necesarios para transformar una revisión anterior en la nueva
|
jerojasro@406
|
116 revisión. Para muchos tipos de fichero, estos deltas son típicamente
|
jerojasro@406
|
117 de una fracción porcentual del tamaño de una copia completa del
|
jerojasro@406
|
118 fichero.
|
jerojasro@406
|
119
|
jerojasro@406
|
120 Algunos sistemas de control de revisiones obsoletos sólo pueden
|
jerojasro@406
|
121 manipular deltas de ficheros de texto plano. Ellos o bien almacenan
|
jerojasro@406
|
122 los ficheros binarios como instantáneas completas, o codificados en
|
jerojasro@406
|
123 alguna representación de texto plano adecuada, y ambas alternativas
|
jerojasro@406
|
124 son enfoques que desperdician bastantes recursos. Mercurial puede
|
jerojasro@406
|
125 manejar deltas de ficheros con contenido binario arbitrario; no
|
jerojasro@406
|
126 necesita tratar el texto plano como un caso especial.
|
jerojasro@343
|
127
|
jerojasro@343
|
128 \subsection{Safe operation}
|
jerojasro@343
|
129 \label{sec:concepts:txn}
|
jerojasro@343
|
130
|
jerojasro@343
|
131 Mercurial only ever \emph{appends} data to the end of a revlog file.
|
jerojasro@343
|
132 It never modifies a section of a file after it has written it. This
|
jerojasro@343
|
133 is both more robust and efficient than schemes that need to modify or
|
jerojasro@343
|
134 rewrite data.
|
jerojasro@343
|
135
|
jerojasro@343
|
136 In addition, Mercurial treats every write as part of a
|
jerojasro@343
|
137 \emph{transaction} that can span a number of files. A transaction is
|
jerojasro@343
|
138 \emph{atomic}: either the entire transaction succeeds and its effects
|
jerojasro@343
|
139 are all visible to readers in one go, or the whole thing is undone.
|
jerojasro@343
|
140 This guarantee of atomicity means that if you're running two copies of
|
jerojasro@343
|
141 Mercurial, where one is reading data and one is writing it, the reader
|
jerojasro@343
|
142 will never see a partially written result that might confuse it.
|
jerojasro@343
|
143
|
jerojasro@343
|
144 The fact that Mercurial only appends to files makes it easier to
|
jerojasro@343
|
145 provide this transactional guarantee. The easier it is to do stuff
|
jerojasro@343
|
146 like this, the more confident you should be that it's done correctly.
|
jerojasro@343
|
147
|
jerojasro@343
|
148 \subsection{Fast retrieval}
|
jerojasro@343
|
149
|
jerojasro@343
|
150 Mercurial cleverly avoids a pitfall common to all earlier
|
jerojasro@343
|
151 revision control systems: the problem of \emph{inefficient retrieval}.
|
jerojasro@343
|
152 Most revision control systems store the contents of a revision as an
|
jerojasro@343
|
153 incremental series of modifications against a ``snapshot''. To
|
jerojasro@343
|
154 reconstruct a specific revision, you must first read the snapshot, and
|
jerojasro@343
|
155 then every one of the revisions between the snapshot and your target
|
jerojasro@343
|
156 revision. The more history that a file accumulates, the more
|
jerojasro@343
|
157 revisions you must read, hence the longer it takes to reconstruct a
|
jerojasro@343
|
158 particular revision.
|
jerojasro@343
|
159
|
jerojasro@343
|
160 \begin{figure}[ht]
|
jerojasro@343
|
161 \centering
|
jerojasro@343
|
162 \grafix{snapshot}
|
jerojasro@343
|
163 \caption{Snapshot of a revlog, with incremental deltas}
|
jerojasro@343
|
164 \label{fig:concepts:snapshot}
|
jerojasro@343
|
165 \end{figure}
|
jerojasro@343
|
166
|
jerojasro@343
|
167 The innovation that Mercurial applies to this problem is simple but
|
jerojasro@343
|
168 effective. Once the cumulative amount of delta information stored
|
jerojasro@343
|
169 since the last snapshot exceeds a fixed threshold, it stores a new
|
jerojasro@343
|
170 snapshot (compressed, of course), instead of another delta. This
|
jerojasro@343
|
171 makes it possible to reconstruct \emph{any} revision of a file
|
jerojasro@343
|
172 quickly. This approach works so well that it has since been copied by
|
jerojasro@343
|
173 several other revision control systems.
|
jerojasro@343
|
174
|
jerojasro@343
|
175 Figure~\ref{fig:concepts:snapshot} illustrates the idea. In an entry
|
jerojasro@343
|
176 in a revlog's index file, Mercurial stores the range of entries from
|
jerojasro@343
|
177 the data file that it must read to reconstruct a particular revision.
|
jerojasro@343
|
178
|
jerojasro@343
|
179 \subsubsection{Aside: the influence of video compression}
|
jerojasro@343
|
180
|
jerojasro@343
|
181 If you're familiar with video compression or have ever watched a TV
|
jerojasro@343
|
182 feed through a digital cable or satellite service, you may know that
|
jerojasro@343
|
183 most video compression schemes store each frame of video as a delta
|
jerojasro@343
|
184 against its predecessor frame. In addition, these schemes use
|
jerojasro@343
|
185 ``lossy'' compression techniques to increase the compression ratio, so
|
jerojasro@343
|
186 visual errors accumulate over the course of a number of inter-frame
|
jerojasro@343
|
187 deltas.
|
jerojasro@343
|
188
|
jerojasro@343
|
189 Because it's possible for a video stream to ``drop out'' occasionally
|
jerojasro@343
|
190 due to signal glitches, and to limit the accumulation of artefacts
|
jerojasro@343
|
191 introduced by the lossy compression process, video encoders
|
jerojasro@343
|
192 periodically insert a complete frame (called a ``key frame'') into the
|
jerojasro@343
|
193 video stream; the next delta is generated against that frame. This
|
jerojasro@343
|
194 means that if the video signal gets interrupted, it will resume once
|
jerojasro@343
|
195 the next key frame is received. Also, the accumulation of encoding
|
jerojasro@343
|
196 errors restarts anew with each key frame.
|
jerojasro@343
|
197
|
jerojasro@343
|
198 \subsection{Identification and strong integrity}
|
jerojasro@343
|
199
|
jerojasro@343
|
200 Along with delta or snapshot information, a revlog entry contains a
|
jerojasro@343
|
201 cryptographic hash of the data that it represents. This makes it
|
jerojasro@343
|
202 difficult to forge the contents of a revision, and easy to detect
|
jerojasro@343
|
203 accidental corruption.
|
jerojasro@343
|
204
|
jerojasro@343
|
205 Hashes provide more than a mere check against corruption; they are
|
jerojasro@343
|
206 used as the identifiers for revisions. The changeset identification
|
jerojasro@343
|
207 hashes that you see as an end user are from revisions of the
|
jerojasro@343
|
208 changelog. Although filelogs and the manifest also use hashes,
|
jerojasro@343
|
209 Mercurial only uses these behind the scenes.
|
jerojasro@343
|
210
|
jerojasro@343
|
211 Mercurial verifies that hashes are correct when it retrieves file
|
jerojasro@343
|
212 revisions and when it pulls changes from another repository. If it
|
jerojasro@343
|
213 encounters an integrity problem, it will complain and stop whatever
|
jerojasro@343
|
214 it's doing.
|
jerojasro@343
|
215
|
jerojasro@343
|
216 In addition to the effect it has on retrieval efficiency, Mercurial's
|
jerojasro@343
|
217 use of periodic snapshots makes it more robust against partial data
|
jerojasro@343
|
218 corruption. If a revlog becomes partly corrupted due to a hardware
|
jerojasro@343
|
219 error or system bug, it's often possible to reconstruct some or most
|
jerojasro@343
|
220 revisions from the uncorrupted sections of the revlog, both before and
|
jerojasro@343
|
221 after the corrupted section. This would not be possible with a
|
jerojasro@343
|
222 delta-only storage model.
|
jerojasro@343
|
223
|
jerojasro@343
|
224 \section{Revision history, branching,
|
jerojasro@343
|
225 and merging}
|
jerojasro@343
|
226
|
jerojasro@343
|
227 Every entry in a Mercurial revlog knows the identity of its immediate
|
jerojasro@343
|
228 ancestor revision, usually referred to as its \emph{parent}. In fact,
|
jerojasro@343
|
229 a revision contains room for not one parent, but two. Mercurial uses
|
jerojasro@343
|
230 a special hash, called the ``null ID'', to represent the idea ``there
|
jerojasro@343
|
231 is no parent here''. This hash is simply a string of zeroes.
|
jerojasro@343
|
232
|
jerojasro@343
|
233 In figure~\ref{fig:concepts:revlog}, you can see an example of the
|
jerojasro@343
|
234 conceptual structure of a revlog. Filelogs, manifests, and changelogs
|
jerojasro@343
|
235 all have this same structure; they differ only in the kind of data
|
jerojasro@343
|
236 stored in each delta or snapshot.
|
jerojasro@343
|
237
|
jerojasro@343
|
238 The first revision in a revlog (at the bottom of the image) has the
|
jerojasro@343
|
239 null ID in both of its parent slots. For a ``normal'' revision, its
|
jerojasro@343
|
240 first parent slot contains the ID of its parent revision, and its
|
jerojasro@343
|
241 second contains the null ID, indicating that the revision has only one
|
jerojasro@343
|
242 real parent. Any two revisions that have the same parent ID are
|
jerojasro@343
|
243 branches. A revision that represents a merge between branches has two
|
jerojasro@343
|
244 normal revision IDs in its parent slots.
|
jerojasro@343
|
245
|
jerojasro@343
|
246 \begin{figure}[ht]
|
jerojasro@343
|
247 \centering
|
jerojasro@343
|
248 \grafix{revlog}
|
jerojasro@343
|
249 \caption{}
|
jerojasro@343
|
250 \label{fig:concepts:revlog}
|
jerojasro@343
|
251 \end{figure}
|
jerojasro@343
|
252
|
jerojasro@343
|
253 \section{The working directory}
|
jerojasro@343
|
254
|
jerojasro@343
|
255 In the working directory, Mercurial stores a snapshot of the files
|
jerojasro@343
|
256 from the repository as of a particular changeset.
|
jerojasro@343
|
257
|
jerojasro@343
|
258 The working directory ``knows'' which changeset it contains. When you
|
jerojasro@343
|
259 update the working directory to contain a particular changeset,
|
jerojasro@343
|
260 Mercurial looks up the appropriate revision of the manifest to find
|
jerojasro@343
|
261 out which files it was tracking at the time that changeset was
|
jerojasro@343
|
262 committed, and which revision of each file was then current. It then
|
jerojasro@343
|
263 recreates a copy of each of those files, with the same contents it had
|
jerojasro@343
|
264 when the changeset was committed.
|
jerojasro@343
|
265
|
jerojasro@343
|
266 The \emph{dirstate} contains Mercurial's knowledge of the working
|
jerojasro@343
|
267 directory. This details which changeset the working directory is
|
jerojasro@343
|
268 updated to, and all of the files that Mercurial is tracking in the
|
jerojasro@343
|
269 working directory.
|
jerojasro@343
|
270
|
jerojasro@343
|
271 Just as a revision of a revlog has room for two parents, so that it
|
jerojasro@343
|
272 can represent either a normal revision (with one parent) or a merge of
|
jerojasro@343
|
273 two earlier revisions, the dirstate has slots for two parents. When
|
jerojasro@343
|
274 you use the \hgcmd{update} command, the changeset that you update to
|
jerojasro@343
|
275 is stored in the ``first parent'' slot, and the null ID in the second.
|
jerojasro@343
|
276 When you \hgcmd{merge} with another changeset, the first parent
|
jerojasro@343
|
277 remains unchanged, and the second parent is filled in with the
|
jerojasro@343
|
278 changeset you're merging with. The \hgcmd{parents} command tells you
|
jerojasro@343
|
279 what the parents of the dirstate are.
|
jerojasro@343
|
280
|
jerojasro@343
|
281 \subsection{What happens when you commit}
|
jerojasro@343
|
282
|
jerojasro@343
|
283 The dirstate stores parent information for more than just book-keeping
|
jerojasro@343
|
284 purposes. Mercurial uses the parents of the dirstate as \emph{the
|
jerojasro@343
|
285 parents of a new changeset} when you perform a commit.
|
jerojasro@343
|
286
|
jerojasro@343
|
287 \begin{figure}[ht]
|
jerojasro@343
|
288 \centering
|
jerojasro@343
|
289 \grafix{wdir}
|
jerojasro@343
|
290 \caption{The working directory can have two parents}
|
jerojasro@343
|
291 \label{fig:concepts:wdir}
|
jerojasro@343
|
292 \end{figure}
|
jerojasro@343
|
293
|
jerojasro@343
|
294 Figure~\ref{fig:concepts:wdir} shows the normal state of the working
|
jerojasro@343
|
295 directory, where it has a single changeset as parent. That changeset
|
jerojasro@343
|
296 is the \emph{tip}, the newest changeset in the repository that has no
|
jerojasro@343
|
297 children.
|
jerojasro@343
|
298
|
jerojasro@343
|
299 \begin{figure}[ht]
|
jerojasro@343
|
300 \centering
|
jerojasro@343
|
301 \grafix{wdir-after-commit}
|
jerojasro@343
|
302 \caption{The working directory gains new parents after a commit}
|
jerojasro@343
|
303 \label{fig:concepts:wdir-after-commit}
|
jerojasro@343
|
304 \end{figure}
|
jerojasro@343
|
305
|
jerojasro@343
|
306 It's useful to think of the working directory as ``the changeset I'm
|
jerojasro@343
|
307 about to commit''. Any files that you tell Mercurial that you've
|
jerojasro@343
|
308 added, removed, renamed, or copied will be reflected in that
|
jerojasro@343
|
309 changeset, as will modifications to any files that Mercurial is
|
jerojasro@343
|
310 already tracking; the new changeset will have the parents of the
|
jerojasro@343
|
311 working directory as its parents.
|
jerojasro@343
|
312
|
jerojasro@343
|
313 After a commit, Mercurial will update the parents of the working
|
jerojasro@343
|
314 directory, so that the first parent is the ID of the new changeset,
|
jerojasro@343
|
315 and the second is the null ID. This is shown in
|
jerojasro@343
|
316 figure~\ref{fig:concepts:wdir-after-commit}. Mercurial doesn't touch
|
jerojasro@343
|
317 any of the files in the working directory when you commit; it just
|
jerojasro@343
|
318 modifies the dirstate to note its new parents.
|
jerojasro@343
|
319
|
jerojasro@343
|
320 \subsection{Creating a new head}
|
jerojasro@343
|
321
|
jerojasro@343
|
322 It's perfectly normal to update the working directory to a changeset
|
jerojasro@343
|
323 other than the current tip. For example, you might want to know what
|
jerojasro@343
|
324 your project looked like last Tuesday, or you could be looking through
|
jerojasro@343
|
325 changesets to see which one introduced a bug. In cases like this, the
|
jerojasro@343
|
326 natural thing to do is update the working directory to the changeset
|
jerojasro@343
|
327 you're interested in, and then examine the files in the working
|
jerojasro@343
|
328 directory directly to see their contents as they werea when you
|
jerojasro@343
|
329 committed that changeset. The effect of this is shown in
|
jerojasro@343
|
330 figure~\ref{fig:concepts:wdir-pre-branch}.
|
jerojasro@343
|
331
|
jerojasro@343
|
332 \begin{figure}[ht]
|
jerojasro@343
|
333 \centering
|
jerojasro@343
|
334 \grafix{wdir-pre-branch}
|
jerojasro@343
|
335 \caption{The working directory, updated to an older changeset}
|
jerojasro@343
|
336 \label{fig:concepts:wdir-pre-branch}
|
jerojasro@343
|
337 \end{figure}
|
jerojasro@343
|
338
|
jerojasro@343
|
339 Having updated the working directory to an older changeset, what
|
jerojasro@343
|
340 happens if you make some changes, and then commit? Mercurial behaves
|
jerojasro@343
|
341 in the same way as I outlined above. The parents of the working
|
jerojasro@343
|
342 directory become the parents of the new changeset. This new changeset
|
jerojasro@343
|
343 has no children, so it becomes the new tip. And the repository now
|
jerojasro@343
|
344 contains two changesets that have no children; we call these
|
jerojasro@343
|
345 \emph{heads}. You can see the structure that this creates in
|
jerojasro@343
|
346 figure~\ref{fig:concepts:wdir-branch}.
|
jerojasro@343
|
347
|
jerojasro@343
|
348 \begin{figure}[ht]
|
jerojasro@343
|
349 \centering
|
jerojasro@343
|
350 \grafix{wdir-branch}
|
jerojasro@343
|
351 \caption{After a commit made while synced to an older changeset}
|
jerojasro@343
|
352 \label{fig:concepts:wdir-branch}
|
jerojasro@343
|
353 \end{figure}
|
jerojasro@343
|
354
|
jerojasro@343
|
355 \begin{note}
|
jerojasro@343
|
356 If you're new to Mercurial, you should keep in mind a common
|
jerojasro@343
|
357 ``error'', which is to use the \hgcmd{pull} command without any
|
jerojasro@343
|
358 options. By default, the \hgcmd{pull} command \emph{does not}
|
jerojasro@343
|
359 update the working directory, so you'll bring new changesets into
|
jerojasro@343
|
360 your repository, but the working directory will stay synced at the
|
jerojasro@343
|
361 same changeset as before the pull. If you make some changes and
|
jerojasro@343
|
362 commit afterwards, you'll thus create a new head, because your
|
jerojasro@343
|
363 working directory isn't synced to whatever the current tip is.
|
jerojasro@343
|
364
|
jerojasro@343
|
365 I put the word ``error'' in quotes because all that you need to do
|
jerojasro@343
|
366 to rectify this situation is \hgcmd{merge}, then \hgcmd{commit}. In
|
jerojasro@343
|
367 other words, this almost never has negative consequences; it just
|
jerojasro@343
|
368 surprises people. I'll discuss other ways to avoid this behaviour,
|
jerojasro@343
|
369 and why Mercurial behaves in this initially surprising way, later
|
jerojasro@343
|
370 on.
|
jerojasro@343
|
371 \end{note}
|
jerojasro@343
|
372
|
jerojasro@343
|
373 \subsection{Merging heads}
|
jerojasro@343
|
374
|
jerojasro@343
|
375 When you run the \hgcmd{merge} command, Mercurial leaves the first
|
jerojasro@343
|
376 parent of the working directory unchanged, and sets the second parent
|
jerojasro@343
|
377 to the changeset you're merging with, as shown in
|
jerojasro@343
|
378 figure~\ref{fig:concepts:wdir-merge}.
|
jerojasro@343
|
379
|
jerojasro@343
|
380 \begin{figure}[ht]
|
jerojasro@343
|
381 \centering
|
jerojasro@343
|
382 \grafix{wdir-merge}
|
jerojasro@343
|
383 \caption{Merging two heads}
|
jerojasro@343
|
384 \label{fig:concepts:wdir-merge}
|
jerojasro@343
|
385 \end{figure}
|
jerojasro@343
|
386
|
jerojasro@343
|
387 Mercurial also has to modify the working directory, to merge the files
|
jerojasro@343
|
388 managed in the two changesets. Simplified a little, the merging
|
jerojasro@343
|
389 process goes like this, for every file in the manifests of both
|
jerojasro@343
|
390 changesets.
|
jerojasro@343
|
391 \begin{itemize}
|
jerojasro@343
|
392 \item If neither changeset has modified a file, do nothing with that
|
jerojasro@343
|
393 file.
|
jerojasro@343
|
394 \item If one changeset has modified a file, and the other hasn't,
|
jerojasro@343
|
395 create the modified copy of the file in the working directory.
|
jerojasro@343
|
396 \item If one changeset has removed a file, and the other hasn't (or
|
jerojasro@343
|
397 has also deleted it), delete the file from the working directory.
|
jerojasro@343
|
398 \item If one changeset has removed a file, but the other has modified
|
jerojasro@343
|
399 the file, ask the user what to do: keep the modified file, or remove
|
jerojasro@343
|
400 it?
|
jerojasro@343
|
401 \item If both changesets have modified a file, invoke an external
|
jerojasro@343
|
402 merge program to choose the new contents for the merged file. This
|
jerojasro@343
|
403 may require input from the user.
|
jerojasro@343
|
404 \item If one changeset has modified a file, and the other has renamed
|
jerojasro@343
|
405 or copied the file, make sure that the changes follow the new name
|
jerojasro@343
|
406 of the file.
|
jerojasro@343
|
407 \end{itemize}
|
jerojasro@343
|
408 There are more details---merging has plenty of corner cases---but
|
jerojasro@343
|
409 these are the most common choices that are involved in a merge. As
|
jerojasro@343
|
410 you can see, most cases are completely automatic, and indeed most
|
jerojasro@343
|
411 merges finish automatically, without requiring your input to resolve
|
jerojasro@343
|
412 any conflicts.
|
jerojasro@343
|
413
|
jerojasro@343
|
414 When you're thinking about what happens when you commit after a merge,
|
jerojasro@343
|
415 once again the working directory is ``the changeset I'm about to
|
jerojasro@343
|
416 commit''. After the \hgcmd{merge} command completes, the working
|
jerojasro@343
|
417 directory has two parents; these will become the parents of the new
|
jerojasro@343
|
418 changeset.
|
jerojasro@343
|
419
|
jerojasro@343
|
420 Mercurial lets you perform multiple merges, but you must commit the
|
jerojasro@343
|
421 results of each individual merge as you go. This is necessary because
|
jerojasro@343
|
422 Mercurial only tracks two parents for both revisions and the working
|
jerojasro@343
|
423 directory. While it would be technically possible to merge multiple
|
jerojasro@343
|
424 changesets at once, the prospect of user confusion and making a
|
jerojasro@343
|
425 terrible mess of a merge immediately becomes overwhelming.
|
jerojasro@343
|
426
|
jerojasro@343
|
427 \section{Other interesting design features}
|
jerojasro@343
|
428
|
jerojasro@343
|
429 In the sections above, I've tried to highlight some of the most
|
jerojasro@343
|
430 important aspects of Mercurial's design, to illustrate that it pays
|
jerojasro@343
|
431 careful attention to reliability and performance. However, the
|
jerojasro@343
|
432 attention to detail doesn't stop there. There are a number of other
|
jerojasro@343
|
433 aspects of Mercurial's construction that I personally find
|
jerojasro@343
|
434 interesting. I'll detail a few of them here, separate from the ``big
|
jerojasro@343
|
435 ticket'' items above, so that if you're interested, you can gain a
|
jerojasro@343
|
436 better idea of the amount of thinking that goes into a well-designed
|
jerojasro@343
|
437 system.
|
jerojasro@343
|
438
|
jerojasro@343
|
439 \subsection{Clever compression}
|
jerojasro@343
|
440
|
jerojasro@343
|
441 When appropriate, Mercurial will store both snapshots and deltas in
|
jerojasro@343
|
442 compressed form. It does this by always \emph{trying to} compress a
|
jerojasro@343
|
443 snapshot or delta, but only storing the compressed version if it's
|
jerojasro@343
|
444 smaller than the uncompressed version.
|
jerojasro@343
|
445
|
jerojasro@343
|
446 This means that Mercurial does ``the right thing'' when storing a file
|
jerojasro@343
|
447 whose native form is compressed, such as a \texttt{zip} archive or a
|
jerojasro@343
|
448 JPEG image. When these types of files are compressed a second time,
|
jerojasro@343
|
449 the resulting file is usually bigger than the once-compressed form,
|
jerojasro@343
|
450 and so Mercurial will store the plain \texttt{zip} or JPEG.
|
jerojasro@343
|
451
|
jerojasro@343
|
452 Deltas between revisions of a compressed file are usually larger than
|
jerojasro@343
|
453 snapshots of the file, and Mercurial again does ``the right thing'' in
|
jerojasro@343
|
454 these cases. It finds that such a delta exceeds the threshold at
|
jerojasro@343
|
455 which it should store a complete snapshot of the file, so it stores
|
jerojasro@343
|
456 the snapshot, again saving space compared to a naive delta-only
|
jerojasro@343
|
457 approach.
|
jerojasro@343
|
458
|
jerojasro@343
|
459 \subsubsection{Network recompression}
|
jerojasro@343
|
460
|
jerojasro@343
|
461 When storing revisions on disk, Mercurial uses the ``deflate''
|
jerojasro@343
|
462 compression algorithm (the same one used by the popular \texttt{zip}
|
jerojasro@343
|
463 archive format), which balances good speed with a respectable
|
jerojasro@343
|
464 compression ratio. However, when transmitting revision data over a
|
jerojasro@343
|
465 network connection, Mercurial uncompresses the compressed revision
|
jerojasro@343
|
466 data.
|
jerojasro@343
|
467
|
jerojasro@343
|
468 If the connection is over HTTP, Mercurial recompresses the entire
|
jerojasro@343
|
469 stream of data using a compression algorithm that gives a better
|
jerojasro@343
|
470 compression ratio (the Burrows-Wheeler algorithm from the widely used
|
jerojasro@343
|
471 \texttt{bzip2} compression package). This combination of algorithm
|
jerojasro@343
|
472 and compression of the entire stream (instead of a revision at a time)
|
jerojasro@343
|
473 substantially reduces the number of bytes to be transferred, yielding
|
jerojasro@343
|
474 better network performance over almost all kinds of network.
|
jerojasro@343
|
475
|
jerojasro@343
|
476 (If the connection is over \command{ssh}, Mercurial \emph{doesn't}
|
jerojasro@343
|
477 recompress the stream, because \command{ssh} can already do this
|
jerojasro@343
|
478 itself.)
|
jerojasro@343
|
479
|
jerojasro@343
|
480 \subsection{Read/write ordering and atomicity}
|
jerojasro@343
|
481
|
jerojasro@343
|
482 Appending to files isn't the whole story when it comes to guaranteeing
|
jerojasro@343
|
483 that a reader won't see a partial write. If you recall
|
jerojasro@343
|
484 figure~\ref{fig:concepts:metadata}, revisions in the changelog point to
|
jerojasro@343
|
485 revisions in the manifest, and revisions in the manifest point to
|
jerojasro@343
|
486 revisions in filelogs. This hierarchy is deliberate.
|
jerojasro@343
|
487
|
jerojasro@343
|
488 A writer starts a transaction by writing filelog and manifest data,
|
jerojasro@343
|
489 and doesn't write any changelog data until those are finished. A
|
jerojasro@343
|
490 reader starts by reading changelog data, then manifest data, followed
|
jerojasro@343
|
491 by filelog data.
|
jerojasro@343
|
492
|
jerojasro@343
|
493 Since the writer has always finished writing filelog and manifest data
|
jerojasro@343
|
494 before it writes to the changelog, a reader will never read a pointer
|
jerojasro@343
|
495 to a partially written manifest revision from the changelog, and it will
|
jerojasro@343
|
496 never read a pointer to a partially written filelog revision from the
|
jerojasro@343
|
497 manifest.
|
jerojasro@343
|
498
|
jerojasro@343
|
499 \subsection{Concurrent access}
|
jerojasro@343
|
500
|
jerojasro@343
|
501 The read/write ordering and atomicity guarantees mean that Mercurial
|
jerojasro@343
|
502 never needs to \emph{lock} a repository when it's reading data, even
|
jerojasro@343
|
503 if the repository is being written to while the read is occurring.
|
jerojasro@343
|
504 This has a big effect on scalability; you can have an arbitrary number
|
jerojasro@343
|
505 of Mercurial processes safely reading data from a repository safely
|
jerojasro@343
|
506 all at once, no matter whether it's being written to or not.
|
jerojasro@343
|
507
|
jerojasro@343
|
508 The lockless nature of reading means that if you're sharing a
|
jerojasro@343
|
509 repository on a multi-user system, you don't need to grant other local
|
jerojasro@343
|
510 users permission to \emph{write} to your repository in order for them
|
jerojasro@343
|
511 to be able to clone it or pull changes from it; they only need
|
jerojasro@343
|
512 \emph{read} permission. (This is \emph{not} a common feature among
|
jerojasro@343
|
513 revision control systems, so don't take it for granted! Most require
|
jerojasro@343
|
514 readers to be able to lock a repository to access it safely, and this
|
jerojasro@343
|
515 requires write permission on at least one directory, which of course
|
jerojasro@343
|
516 makes for all kinds of nasty and annoying security and administrative
|
jerojasro@343
|
517 problems.)
|
jerojasro@343
|
518
|
jerojasro@343
|
519 Mercurial uses locks to ensure that only one process can write to a
|
jerojasro@343
|
520 repository at a time (the locking mechanism is safe even over
|
jerojasro@343
|
521 filesystems that are notoriously hostile to locking, such as NFS). If
|
jerojasro@343
|
522 a repository is locked, a writer will wait for a while to retry if the
|
jerojasro@343
|
523 repository becomes unlocked, but if the repository remains locked for
|
jerojasro@343
|
524 too long, the process attempting to write will time out after a while.
|
jerojasro@343
|
525 This means that your daily automated scripts won't get stuck forever
|
jerojasro@343
|
526 and pile up if a system crashes unnoticed, for example. (Yes, the
|
jerojasro@343
|
527 timeout is configurable, from zero to infinity.)
|
jerojasro@343
|
528
|
jerojasro@343
|
529 \subsubsection{Safe dirstate access}
|
jerojasro@343
|
530
|
jerojasro@343
|
531 As with revision data, Mercurial doesn't take a lock to read the
|
jerojasro@343
|
532 dirstate file; it does acquire a lock to write it. To avoid the
|
jerojasro@343
|
533 possibility of reading a partially written copy of the dirstate file,
|
jerojasro@343
|
534 Mercurial writes to a file with a unique name in the same directory as
|
jerojasro@343
|
535 the dirstate file, then renames the temporary file atomically to
|
jerojasro@343
|
536 \filename{dirstate}. The file named \filename{dirstate} is thus
|
jerojasro@343
|
537 guaranteed to be complete, not partially written.
|
jerojasro@343
|
538
|
jerojasro@343
|
539 \subsection{Avoiding seeks}
|
jerojasro@343
|
540
|
jerojasro@343
|
541 Critical to Mercurial's performance is the avoidance of seeks of the
|
jerojasro@343
|
542 disk head, since any seek is far more expensive than even a
|
jerojasro@343
|
543 comparatively large read operation.
|
jerojasro@343
|
544
|
jerojasro@343
|
545 This is why, for example, the dirstate is stored in a single file. If
|
jerojasro@343
|
546 there were a dirstate file per directory that Mercurial tracked, the
|
jerojasro@343
|
547 disk would seek once per directory. Instead, Mercurial reads the
|
jerojasro@343
|
548 entire single dirstate file in one step.
|
jerojasro@343
|
549
|
jerojasro@343
|
550 Mercurial also uses a ``copy on write'' scheme when cloning a
|
jerojasro@343
|
551 repository on local storage. Instead of copying every revlog file
|
jerojasro@343
|
552 from the old repository into the new repository, it makes a ``hard
|
jerojasro@343
|
553 link'', which is a shorthand way to say ``these two names point to the
|
jerojasro@343
|
554 same file''. When Mercurial is about to write to one of a revlog's
|
jerojasro@343
|
555 files, it checks to see if the number of names pointing at the file is
|
jerojasro@343
|
556 greater than one. If it is, more than one repository is using the
|
jerojasro@343
|
557 file, so Mercurial makes a new copy of the file that is private to
|
jerojasro@343
|
558 this repository.
|
jerojasro@343
|
559
|
jerojasro@343
|
560 A few revision control developers have pointed out that this idea of
|
jerojasro@343
|
561 making a complete private copy of a file is not very efficient in its
|
jerojasro@343
|
562 use of storage. While this is true, storage is cheap, and this method
|
jerojasro@343
|
563 gives the highest performance while deferring most book-keeping to the
|
jerojasro@343
|
564 operating system. An alternative scheme would most likely reduce
|
jerojasro@343
|
565 performance and increase the complexity of the software, each of which
|
jerojasro@343
|
566 is much more important to the ``feel'' of day-to-day use.
|
jerojasro@343
|
567
|
jerojasro@343
|
568 \subsection{Other contents of the dirstate}
|
jerojasro@343
|
569
|
jerojasro@343
|
570 Because Mercurial doesn't force you to tell it when you're modifying a
|
jerojasro@343
|
571 file, it uses the dirstate to store some extra information so it can
|
jerojasro@343
|
572 determine efficiently whether you have modified a file. For each file
|
jerojasro@343
|
573 in the working directory, it stores the time that it last modified the
|
jerojasro@343
|
574 file itself, and the size of the file at that time.
|
jerojasro@343
|
575
|
jerojasro@343
|
576 When you explicitly \hgcmd{add}, \hgcmd{remove}, \hgcmd{rename} or
|
jerojasro@343
|
577 \hgcmd{copy} files, Mercurial updates the dirstate so that it knows
|
jerojasro@343
|
578 what to do with those files when you commit.
|
jerojasro@343
|
579
|
jerojasro@343
|
580 When Mercurial is checking the states of files in the working
|
jerojasro@343
|
581 directory, it first checks a file's modification time. If that has
|
jerojasro@343
|
582 not changed, the file must not have been modified. If the file's size
|
jerojasro@343
|
583 has changed, the file must have been modified. If the modification
|
jerojasro@343
|
584 time has changed, but the size has not, only then does Mercurial need
|
jerojasro@343
|
585 to read the actual contents of the file to see if they've changed.
|
jerojasro@343
|
586 Storing these few extra pieces of information dramatically reduces the
|
jerojasro@343
|
587 amount of data that Mercurial needs to read, which yields large
|
jerojasro@343
|
588 performance improvements compared to other revision control systems.
|
jerojasro@343
|
589
|
jerojasro@343
|
590 %%% Local Variables:
|
jerojasro@343
|
591 %%% mode: latex
|
jerojasro@343
|
592 %%% TeX-master: "00book"
|
jerojasro@343
|
593 %%% End:
|