Mercurial > hgbook
annotate es/concepts.tex @ 505:446d1b4b7a71
translated up to section 4.2.1, included
author | Javier Rojas <jerojasro@devnull.li> |
---|---|
date | Sat, 08 Nov 2008 23:29:59 -0500 |
parents | 0f2170fcb79a |
children | 7c84967093e1 |
rev | line source |
---|---|
492
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
1 \chapter{Tras bambalinas} |
442 | 2 \label{chap:concepts} |
3 | |
492
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
4 A diferencia de varios sistemas de control de revisiones, los |
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
5 conceptos en los que se fundamenta Mercurial son lo suficientemente |
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
6 simples como para entender fácilmente cómo funciona el software. |
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
7 Saber esto no es necesario, pero considero útil tener un ``modelo |
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
8 mental'' de qué es lo que sucede. |
442 | 9 |
492
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
10 Comprender esto me da la confianza de que Mercurial ha sido |
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
11 cuidadosamente diseñado para ser tanto \emph{seguro} como |
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
12 \emph{eficiente}. Y tal vez con la misma importancia, si es fácil |
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
13 para mí hacerme a una idea adecuada de qué está haciendo el software |
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
14 cuando llevo a cabo una tarea relacionada con control de revisiones, |
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
15 es menos probable que me sosprenda su comportamiento. |
442 | 16 |
492
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
17 En este capítulo, cubriremos inicialmente los conceptos centrales |
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
18 del diseño de Mercurial, y luego discutiremos algunos detalles |
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
19 interesantes de su implementación. |
442 | 20 |
492
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
21 \section{Registro del historial de Mercurial} |
442 | 22 |
492
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
23 \subsection{Seguir el historial de un único fichero} |
442 | 24 |
492
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
25 Cuando Mercurial sigue las modificaciones a un fichero, guarda el |
2c2c86824c61
began translation of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
442
diff
changeset
|
26 historial de dicho fichero en un objeto de metadatos llamado |
494
149ea8ae39c4
translated a couple of paragraphs of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
492
diff
changeset
|
27 \emph{filelog}\ndt{Fichero de registro}. Cada entrada en el fichero |
149ea8ae39c4
translated a couple of paragraphs of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
492
diff
changeset
|
28 de registro contiene suficiente información para reconstruir una |
149ea8ae39c4
translated a couple of paragraphs of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
492
diff
changeset
|
29 revisión del fichero que se está siguiendo. Los ficheros de registro |
149ea8ae39c4
translated a couple of paragraphs of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
492
diff
changeset
|
30 son almacenados como ficheros el el directorio |
149ea8ae39c4
translated a couple of paragraphs of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
492
diff
changeset
|
31 \sdirname{.hg/store/data}. Un fichero de registro contiene dos tipos |
149ea8ae39c4
translated a couple of paragraphs of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
492
diff
changeset
|
32 de información: datos de revisiones, y un índice para ayudar a |
149ea8ae39c4
translated a couple of paragraphs of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
492
diff
changeset
|
33 Mercurial a buscar revisiones eficientemente. |
442 | 34 |
494
149ea8ae39c4
translated a couple of paragraphs of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
492
diff
changeset
|
35 El fichero de registro de un fichero grande, o con un historial muy |
149ea8ae39c4
translated a couple of paragraphs of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
492
diff
changeset
|
36 largo, es guardado como ficheros separados para datos (sufijo |
149ea8ae39c4
translated a couple of paragraphs of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
492
diff
changeset
|
37 ``\texttt{.d}'') y para el índice (sufijo ``\texttt{.i}''). Para |
149ea8ae39c4
translated a couple of paragraphs of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
492
diff
changeset
|
38 ficheros pequeños con un historial pequeño, los datos de revisiones y |
149ea8ae39c4
translated a couple of paragraphs of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
492
diff
changeset
|
39 el índice son combinados en un único fichero ``\texttt{.i}''. La |
149ea8ae39c4
translated a couple of paragraphs of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
492
diff
changeset
|
40 correspondencia entre un fichero en el directorio de trabajo y el |
149ea8ae39c4
translated a couple of paragraphs of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
492
diff
changeset
|
41 fichero de registro que hace seguimiento a su historial en el |
149ea8ae39c4
translated a couple of paragraphs of the "behind the scenes" chapter
Javier Rojas <jerojasro@devnull.li>
parents:
492
diff
changeset
|
42 repositorio se ilustra en la figura~\ref{fig:concepts:filelog}. |
442 | 43 |
44 \begin{figure}[ht] | |
45 \centering | |
46 \grafix{filelog} | |
495
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
47 \caption{Relación entre ficheros en el directorio de trabajo y |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
48 ficheros de registro en el repositorio} |
442 | 49 \label{fig:concepts:filelog} |
50 \end{figure} | |
51 | |
495
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
52 \subsection{Administración de ficheros monitoreados} |
442 | 53 |
495
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
54 Mercurial usa una estructura llamada \emph{manifiesto} para |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
55 % TODO collect together => centralizar |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
56 centralizar la información que maneja acerca de los ficheros que |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
57 monitorea. Cada entrada en el manifiesto contiene información acerca |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
58 de los ficheros involucrados en un único conjunto de cambios. Una |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
59 entrada registra qué ficheros están presentes en el conjunto de |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
60 cambios, la revisión de cada fichero, y otros cuantos metadatos del |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
61 mismo. |
442 | 62 |
495
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
63 \subsection{Registro de información del conjunto de cambios} |
442 | 64 |
495
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
65 La \emph{bitácora de cambios} contiene información acerca de cada |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
66 conjunto de cambios. Cada revisión indica quién consignó un cambio, el |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
67 comentario para el conjunto de cambios, otros datos relacionados con |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
68 el conjunto de cambios, y la revisión del manifiesto a usar. |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
69 |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
70 \subsection{Relaciones entre revisiones} |
442 | 71 |
495
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
72 Dentro de una bitácora de cambios, un manifiesto, o un fichero de |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
73 registro, cada revisión conserva un apuntador a su padre inmediato |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
74 (o sus dos padres, si es la revisión de una fusión). Como menciońe |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
75 anteriormente, también hay relaciones entre revisiones \emph{a través} |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
76 de estas estructuras, y tienen naturaleza jerárquica. |
442 | 77 |
495
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
78 Por cada conjunto de cambios en un repositorio, hay exactamente una |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
79 revisión almacenada en la bitácora de cambios. Cada revisión de la |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
80 bitácora de cambios contiene un apuntador a una única revisión del |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
81 manifiesto. Una revisión del manifiesto almacena un apuntador a una |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
82 única revisión de cada fichero de registro al que se le hacía |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
83 seguimiento cuando fue creado el conjunto de cambios. Estas relaciones |
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
84 se ilustran en la figura~\ref{fig:concepts:metadata}. |
442 | 85 |
86 \begin{figure}[ht] | |
87 \centering | |
88 \grafix{metadata} | |
495
0f2170fcb79a
added the term "tracking", and translated a few paragraphs of concepts.tex
Javier Rojas <jerojasro@devnull.li>
parents:
494
diff
changeset
|
89 \caption{Relaciones entre metadatos} |
442 | 90 \label{fig:concepts:metadata} |
91 \end{figure} | |
92 | |
505
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
93 Como lo muestra la figura, \emph{no} hay una relación ``uno a uno'' |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
94 entre las revisiones en el conjunto de cambios, el manifiesto, o el |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
95 fichero de registro. Si el manifiesto no ha sido modificado de un |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
96 conjunto de cambios a otro, las entradas en la bitácora de cambios |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
97 para esos conjuntos de cambios apuntarán a la misma revisión del |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
98 manifiesto. Si un fichero monitoreado por Mercurial no sufre ningún |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
99 cambio de un conjunto de cambios a otro, la entrada para dicho fichero |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
100 en las dos revisiones del manifiesto apuntará a la misma revisión de |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
101 su fichero de registro. |
442 | 102 |
505
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
103 \section{Almacenamiento seguro y eficiente} |
442 | 104 |
505
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
105 La base común de las bitácoras de cambios, los manifiestos, y los |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
106 ficheros de registros es provista por una única estructura llamada el |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
107 \emph{revlog}\ndt{Contracción de \emph{revision log}, registro de |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
108 revisión.}. |
442 | 109 |
505
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
110 \subsection{Almacenamiento eficiente} |
442 | 111 |
505
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
112 El revlog provee almacenamiento eficiente de revisiones por medio del |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
113 mecanismo de \emph{deltas}\ndt{Diferencias.}. En vez de almacenar una |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
114 copia completa del fichero por cada revisión, almacena los cambios |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
115 necesarios para transformar una revisión anterior en la nueva |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
116 revisión. Para muchos tipos de fichero, estos deltas son típicamente |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
117 de una fracción porcentual del tamaño de una copia completa del |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
118 fichero. |
442 | 119 |
505
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
120 Algunos sistemas de control de revisiones obsoletos sólo pueden |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
121 manipular deltas de ficheros de texto plano. Ellos o bien almacenan |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
122 los ficheros binarios como instantáneas completas, o codificados en |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
123 alguna representación de texto plano adecuada, y ambas alternativas |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
124 son enfoques que desperdician bastantes recursos. Mercurial puede |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
125 manejar deltas de ficheros con contenido binario arbitrario; no |
446d1b4b7a71
translated up to section 4.2.1, included
Javier Rojas <jerojasro@devnull.li>
parents:
495
diff
changeset
|
126 necesita tratar el texto plano como un caso especial. |
442 | 127 |
128 \subsection{Safe operation} | |
129 \label{sec:concepts:txn} | |
130 | |
131 Mercurial only ever \emph{appends} data to the end of a revlog file. | |
132 It never modifies a section of a file after it has written it. This | |
133 is both more robust and efficient than schemes that need to modify or | |
134 rewrite data. | |
135 | |
136 In addition, Mercurial treats every write as part of a | |
137 \emph{transaction} that can span a number of files. A transaction is | |
138 \emph{atomic}: either the entire transaction succeeds and its effects | |
139 are all visible to readers in one go, or the whole thing is undone. | |
140 This guarantee of atomicity means that if you're running two copies of | |
141 Mercurial, where one is reading data and one is writing it, the reader | |
142 will never see a partially written result that might confuse it. | |
143 | |
144 The fact that Mercurial only appends to files makes it easier to | |
145 provide this transactional guarantee. The easier it is to do stuff | |
146 like this, the more confident you should be that it's done correctly. | |
147 | |
148 \subsection{Fast retrieval} | |
149 | |
150 Mercurial cleverly avoids a pitfall common to all earlier | |
151 revision control systems: the problem of \emph{inefficient retrieval}. | |
152 Most revision control systems store the contents of a revision as an | |
153 incremental series of modifications against a ``snapshot''. To | |
154 reconstruct a specific revision, you must first read the snapshot, and | |
155 then every one of the revisions between the snapshot and your target | |
156 revision. The more history that a file accumulates, the more | |
157 revisions you must read, hence the longer it takes to reconstruct a | |
158 particular revision. | |
159 | |
160 \begin{figure}[ht] | |
161 \centering | |
162 \grafix{snapshot} | |
163 \caption{Snapshot of a revlog, with incremental deltas} | |
164 \label{fig:concepts:snapshot} | |
165 \end{figure} | |
166 | |
167 The innovation that Mercurial applies to this problem is simple but | |
168 effective. Once the cumulative amount of delta information stored | |
169 since the last snapshot exceeds a fixed threshold, it stores a new | |
170 snapshot (compressed, of course), instead of another delta. This | |
171 makes it possible to reconstruct \emph{any} revision of a file | |
172 quickly. This approach works so well that it has since been copied by | |
173 several other revision control systems. | |
174 | |
175 Figure~\ref{fig:concepts:snapshot} illustrates the idea. In an entry | |
176 in a revlog's index file, Mercurial stores the range of entries from | |
177 the data file that it must read to reconstruct a particular revision. | |
178 | |
179 \subsubsection{Aside: the influence of video compression} | |
180 | |
181 If you're familiar with video compression or have ever watched a TV | |
182 feed through a digital cable or satellite service, you may know that | |
183 most video compression schemes store each frame of video as a delta | |
184 against its predecessor frame. In addition, these schemes use | |
185 ``lossy'' compression techniques to increase the compression ratio, so | |
186 visual errors accumulate over the course of a number of inter-frame | |
187 deltas. | |
188 | |
189 Because it's possible for a video stream to ``drop out'' occasionally | |
190 due to signal glitches, and to limit the accumulation of artefacts | |
191 introduced by the lossy compression process, video encoders | |
192 periodically insert a complete frame (called a ``key frame'') into the | |
193 video stream; the next delta is generated against that frame. This | |
194 means that if the video signal gets interrupted, it will resume once | |
195 the next key frame is received. Also, the accumulation of encoding | |
196 errors restarts anew with each key frame. | |
197 | |
198 \subsection{Identification and strong integrity} | |
199 | |
200 Along with delta or snapshot information, a revlog entry contains a | |
201 cryptographic hash of the data that it represents. This makes it | |
202 difficult to forge the contents of a revision, and easy to detect | |
203 accidental corruption. | |
204 | |
205 Hashes provide more than a mere check against corruption; they are | |
206 used as the identifiers for revisions. The changeset identification | |
207 hashes that you see as an end user are from revisions of the | |
208 changelog. Although filelogs and the manifest also use hashes, | |
209 Mercurial only uses these behind the scenes. | |
210 | |
211 Mercurial verifies that hashes are correct when it retrieves file | |
212 revisions and when it pulls changes from another repository. If it | |
213 encounters an integrity problem, it will complain and stop whatever | |
214 it's doing. | |
215 | |
216 In addition to the effect it has on retrieval efficiency, Mercurial's | |
217 use of periodic snapshots makes it more robust against partial data | |
218 corruption. If a revlog becomes partly corrupted due to a hardware | |
219 error or system bug, it's often possible to reconstruct some or most | |
220 revisions from the uncorrupted sections of the revlog, both before and | |
221 after the corrupted section. This would not be possible with a | |
222 delta-only storage model. | |
223 | |
224 \section{Revision history, branching, | |
225 and merging} | |
226 | |
227 Every entry in a Mercurial revlog knows the identity of its immediate | |
228 ancestor revision, usually referred to as its \emph{parent}. In fact, | |
229 a revision contains room for not one parent, but two. Mercurial uses | |
230 a special hash, called the ``null ID'', to represent the idea ``there | |
231 is no parent here''. This hash is simply a string of zeroes. | |
232 | |
233 In figure~\ref{fig:concepts:revlog}, you can see an example of the | |
234 conceptual structure of a revlog. Filelogs, manifests, and changelogs | |
235 all have this same structure; they differ only in the kind of data | |
236 stored in each delta or snapshot. | |
237 | |
238 The first revision in a revlog (at the bottom of the image) has the | |
239 null ID in both of its parent slots. For a ``normal'' revision, its | |
240 first parent slot contains the ID of its parent revision, and its | |
241 second contains the null ID, indicating that the revision has only one | |
242 real parent. Any two revisions that have the same parent ID are | |
243 branches. A revision that represents a merge between branches has two | |
244 normal revision IDs in its parent slots. | |
245 | |
246 \begin{figure}[ht] | |
247 \centering | |
248 \grafix{revlog} | |
249 \caption{} | |
250 \label{fig:concepts:revlog} | |
251 \end{figure} | |
252 | |
253 \section{The working directory} | |
254 | |
255 In the working directory, Mercurial stores a snapshot of the files | |
256 from the repository as of a particular changeset. | |
257 | |
258 The working directory ``knows'' which changeset it contains. When you | |
259 update the working directory to contain a particular changeset, | |
260 Mercurial looks up the appropriate revision of the manifest to find | |
261 out which files it was tracking at the time that changeset was | |
262 committed, and which revision of each file was then current. It then | |
263 recreates a copy of each of those files, with the same contents it had | |
264 when the changeset was committed. | |
265 | |
266 The \emph{dirstate} contains Mercurial's knowledge of the working | |
267 directory. This details which changeset the working directory is | |
268 updated to, and all of the files that Mercurial is tracking in the | |
269 working directory. | |
270 | |
271 Just as a revision of a revlog has room for two parents, so that it | |
272 can represent either a normal revision (with one parent) or a merge of | |
273 two earlier revisions, the dirstate has slots for two parents. When | |
274 you use the \hgcmd{update} command, the changeset that you update to | |
275 is stored in the ``first parent'' slot, and the null ID in the second. | |
276 When you \hgcmd{merge} with another changeset, the first parent | |
277 remains unchanged, and the second parent is filled in with the | |
278 changeset you're merging with. The \hgcmd{parents} command tells you | |
279 what the parents of the dirstate are. | |
280 | |
281 \subsection{What happens when you commit} | |
282 | |
283 The dirstate stores parent information for more than just book-keeping | |
284 purposes. Mercurial uses the parents of the dirstate as \emph{the | |
285 parents of a new changeset} when you perform a commit. | |
286 | |
287 \begin{figure}[ht] | |
288 \centering | |
289 \grafix{wdir} | |
290 \caption{The working directory can have two parents} | |
291 \label{fig:concepts:wdir} | |
292 \end{figure} | |
293 | |
294 Figure~\ref{fig:concepts:wdir} shows the normal state of the working | |
295 directory, where it has a single changeset as parent. That changeset | |
296 is the \emph{tip}, the newest changeset in the repository that has no | |
297 children. | |
298 | |
299 \begin{figure}[ht] | |
300 \centering | |
301 \grafix{wdir-after-commit} | |
302 \caption{The working directory gains new parents after a commit} | |
303 \label{fig:concepts:wdir-after-commit} | |
304 \end{figure} | |
305 | |
306 It's useful to think of the working directory as ``the changeset I'm | |
307 about to commit''. Any files that you tell Mercurial that you've | |
308 added, removed, renamed, or copied will be reflected in that | |
309 changeset, as will modifications to any files that Mercurial is | |
310 already tracking; the new changeset will have the parents of the | |
311 working directory as its parents. | |
312 | |
313 After a commit, Mercurial will update the parents of the working | |
314 directory, so that the first parent is the ID of the new changeset, | |
315 and the second is the null ID. This is shown in | |
316 figure~\ref{fig:concepts:wdir-after-commit}. Mercurial doesn't touch | |
317 any of the files in the working directory when you commit; it just | |
318 modifies the dirstate to note its new parents. | |
319 | |
320 \subsection{Creating a new head} | |
321 | |
322 It's perfectly normal to update the working directory to a changeset | |
323 other than the current tip. For example, you might want to know what | |
324 your project looked like last Tuesday, or you could be looking through | |
325 changesets to see which one introduced a bug. In cases like this, the | |
326 natural thing to do is update the working directory to the changeset | |
327 you're interested in, and then examine the files in the working | |
328 directory directly to see their contents as they werea when you | |
329 committed that changeset. The effect of this is shown in | |
330 figure~\ref{fig:concepts:wdir-pre-branch}. | |
331 | |
332 \begin{figure}[ht] | |
333 \centering | |
334 \grafix{wdir-pre-branch} | |
335 \caption{The working directory, updated to an older changeset} | |
336 \label{fig:concepts:wdir-pre-branch} | |
337 \end{figure} | |
338 | |
339 Having updated the working directory to an older changeset, what | |
340 happens if you make some changes, and then commit? Mercurial behaves | |
341 in the same way as I outlined above. The parents of the working | |
342 directory become the parents of the new changeset. This new changeset | |
343 has no children, so it becomes the new tip. And the repository now | |
344 contains two changesets that have no children; we call these | |
345 \emph{heads}. You can see the structure that this creates in | |
346 figure~\ref{fig:concepts:wdir-branch}. | |
347 | |
348 \begin{figure}[ht] | |
349 \centering | |
350 \grafix{wdir-branch} | |
351 \caption{After a commit made while synced to an older changeset} | |
352 \label{fig:concepts:wdir-branch} | |
353 \end{figure} | |
354 | |
355 \begin{note} | |
356 If you're new to Mercurial, you should keep in mind a common | |
357 ``error'', which is to use the \hgcmd{pull} command without any | |
358 options. By default, the \hgcmd{pull} command \emph{does not} | |
359 update the working directory, so you'll bring new changesets into | |
360 your repository, but the working directory will stay synced at the | |
361 same changeset as before the pull. If you make some changes and | |
362 commit afterwards, you'll thus create a new head, because your | |
363 working directory isn't synced to whatever the current tip is. | |
364 | |
365 I put the word ``error'' in quotes because all that you need to do | |
366 to rectify this situation is \hgcmd{merge}, then \hgcmd{commit}. In | |
367 other words, this almost never has negative consequences; it just | |
368 surprises people. I'll discuss other ways to avoid this behaviour, | |
369 and why Mercurial behaves in this initially surprising way, later | |
370 on. | |
371 \end{note} | |
372 | |
373 \subsection{Merging heads} | |
374 | |
375 When you run the \hgcmd{merge} command, Mercurial leaves the first | |
376 parent of the working directory unchanged, and sets the second parent | |
377 to the changeset you're merging with, as shown in | |
378 figure~\ref{fig:concepts:wdir-merge}. | |
379 | |
380 \begin{figure}[ht] | |
381 \centering | |
382 \grafix{wdir-merge} | |
383 \caption{Merging two heads} | |
384 \label{fig:concepts:wdir-merge} | |
385 \end{figure} | |
386 | |
387 Mercurial also has to modify the working directory, to merge the files | |
388 managed in the two changesets. Simplified a little, the merging | |
389 process goes like this, for every file in the manifests of both | |
390 changesets. | |
391 \begin{itemize} | |
392 \item If neither changeset has modified a file, do nothing with that | |
393 file. | |
394 \item If one changeset has modified a file, and the other hasn't, | |
395 create the modified copy of the file in the working directory. | |
396 \item If one changeset has removed a file, and the other hasn't (or | |
397 has also deleted it), delete the file from the working directory. | |
398 \item If one changeset has removed a file, but the other has modified | |
399 the file, ask the user what to do: keep the modified file, or remove | |
400 it? | |
401 \item If both changesets have modified a file, invoke an external | |
402 merge program to choose the new contents for the merged file. This | |
403 may require input from the user. | |
404 \item If one changeset has modified a file, and the other has renamed | |
405 or copied the file, make sure that the changes follow the new name | |
406 of the file. | |
407 \end{itemize} | |
408 There are more details---merging has plenty of corner cases---but | |
409 these are the most common choices that are involved in a merge. As | |
410 you can see, most cases are completely automatic, and indeed most | |
411 merges finish automatically, without requiring your input to resolve | |
412 any conflicts. | |
413 | |
414 When you're thinking about what happens when you commit after a merge, | |
415 once again the working directory is ``the changeset I'm about to | |
416 commit''. After the \hgcmd{merge} command completes, the working | |
417 directory has two parents; these will become the parents of the new | |
418 changeset. | |
419 | |
420 Mercurial lets you perform multiple merges, but you must commit the | |
421 results of each individual merge as you go. This is necessary because | |
422 Mercurial only tracks two parents for both revisions and the working | |
423 directory. While it would be technically possible to merge multiple | |
424 changesets at once, the prospect of user confusion and making a | |
425 terrible mess of a merge immediately becomes overwhelming. | |
426 | |
427 \section{Other interesting design features} | |
428 | |
429 In the sections above, I've tried to highlight some of the most | |
430 important aspects of Mercurial's design, to illustrate that it pays | |
431 careful attention to reliability and performance. However, the | |
432 attention to detail doesn't stop there. There are a number of other | |
433 aspects of Mercurial's construction that I personally find | |
434 interesting. I'll detail a few of them here, separate from the ``big | |
435 ticket'' items above, so that if you're interested, you can gain a | |
436 better idea of the amount of thinking that goes into a well-designed | |
437 system. | |
438 | |
439 \subsection{Clever compression} | |
440 | |
441 When appropriate, Mercurial will store both snapshots and deltas in | |
442 compressed form. It does this by always \emph{trying to} compress a | |
443 snapshot or delta, but only storing the compressed version if it's | |
444 smaller than the uncompressed version. | |
445 | |
446 This means that Mercurial does ``the right thing'' when storing a file | |
447 whose native form is compressed, such as a \texttt{zip} archive or a | |
448 JPEG image. When these types of files are compressed a second time, | |
449 the resulting file is usually bigger than the once-compressed form, | |
450 and so Mercurial will store the plain \texttt{zip} or JPEG. | |
451 | |
452 Deltas between revisions of a compressed file are usually larger than | |
453 snapshots of the file, and Mercurial again does ``the right thing'' in | |
454 these cases. It finds that such a delta exceeds the threshold at | |
455 which it should store a complete snapshot of the file, so it stores | |
456 the snapshot, again saving space compared to a naive delta-only | |
457 approach. | |
458 | |
459 \subsubsection{Network recompression} | |
460 | |
461 When storing revisions on disk, Mercurial uses the ``deflate'' | |
462 compression algorithm (the same one used by the popular \texttt{zip} | |
463 archive format), which balances good speed with a respectable | |
464 compression ratio. However, when transmitting revision data over a | |
465 network connection, Mercurial uncompresses the compressed revision | |
466 data. | |
467 | |
468 If the connection is over HTTP, Mercurial recompresses the entire | |
469 stream of data using a compression algorithm that gives a better | |
470 compression ratio (the Burrows-Wheeler algorithm from the widely used | |
471 \texttt{bzip2} compression package). This combination of algorithm | |
472 and compression of the entire stream (instead of a revision at a time) | |
473 substantially reduces the number of bytes to be transferred, yielding | |
474 better network performance over almost all kinds of network. | |
475 | |
476 (If the connection is over \command{ssh}, Mercurial \emph{doesn't} | |
477 recompress the stream, because \command{ssh} can already do this | |
478 itself.) | |
479 | |
480 \subsection{Read/write ordering and atomicity} | |
481 | |
482 Appending to files isn't the whole story when it comes to guaranteeing | |
483 that a reader won't see a partial write. If you recall | |
484 figure~\ref{fig:concepts:metadata}, revisions in the changelog point to | |
485 revisions in the manifest, and revisions in the manifest point to | |
486 revisions in filelogs. This hierarchy is deliberate. | |
487 | |
488 A writer starts a transaction by writing filelog and manifest data, | |
489 and doesn't write any changelog data until those are finished. A | |
490 reader starts by reading changelog data, then manifest data, followed | |
491 by filelog data. | |
492 | |
493 Since the writer has always finished writing filelog and manifest data | |
494 before it writes to the changelog, a reader will never read a pointer | |
495 to a partially written manifest revision from the changelog, and it will | |
496 never read a pointer to a partially written filelog revision from the | |
497 manifest. | |
498 | |
499 \subsection{Concurrent access} | |
500 | |
501 The read/write ordering and atomicity guarantees mean that Mercurial | |
502 never needs to \emph{lock} a repository when it's reading data, even | |
503 if the repository is being written to while the read is occurring. | |
504 This has a big effect on scalability; you can have an arbitrary number | |
505 of Mercurial processes safely reading data from a repository safely | |
506 all at once, no matter whether it's being written to or not. | |
507 | |
508 The lockless nature of reading means that if you're sharing a | |
509 repository on a multi-user system, you don't need to grant other local | |
510 users permission to \emph{write} to your repository in order for them | |
511 to be able to clone it or pull changes from it; they only need | |
512 \emph{read} permission. (This is \emph{not} a common feature among | |
513 revision control systems, so don't take it for granted! Most require | |
514 readers to be able to lock a repository to access it safely, and this | |
515 requires write permission on at least one directory, which of course | |
516 makes for all kinds of nasty and annoying security and administrative | |
517 problems.) | |
518 | |
519 Mercurial uses locks to ensure that only one process can write to a | |
520 repository at a time (the locking mechanism is safe even over | |
521 filesystems that are notoriously hostile to locking, such as NFS). If | |
522 a repository is locked, a writer will wait for a while to retry if the | |
523 repository becomes unlocked, but if the repository remains locked for | |
524 too long, the process attempting to write will time out after a while. | |
525 This means that your daily automated scripts won't get stuck forever | |
526 and pile up if a system crashes unnoticed, for example. (Yes, the | |
527 timeout is configurable, from zero to infinity.) | |
528 | |
529 \subsubsection{Safe dirstate access} | |
530 | |
531 As with revision data, Mercurial doesn't take a lock to read the | |
532 dirstate file; it does acquire a lock to write it. To avoid the | |
533 possibility of reading a partially written copy of the dirstate file, | |
534 Mercurial writes to a file with a unique name in the same directory as | |
535 the dirstate file, then renames the temporary file atomically to | |
536 \filename{dirstate}. The file named \filename{dirstate} is thus | |
537 guaranteed to be complete, not partially written. | |
538 | |
539 \subsection{Avoiding seeks} | |
540 | |
541 Critical to Mercurial's performance is the avoidance of seeks of the | |
542 disk head, since any seek is far more expensive than even a | |
543 comparatively large read operation. | |
544 | |
545 This is why, for example, the dirstate is stored in a single file. If | |
546 there were a dirstate file per directory that Mercurial tracked, the | |
547 disk would seek once per directory. Instead, Mercurial reads the | |
548 entire single dirstate file in one step. | |
549 | |
550 Mercurial also uses a ``copy on write'' scheme when cloning a | |
551 repository on local storage. Instead of copying every revlog file | |
552 from the old repository into the new repository, it makes a ``hard | |
553 link'', which is a shorthand way to say ``these two names point to the | |
554 same file''. When Mercurial is about to write to one of a revlog's | |
555 files, it checks to see if the number of names pointing at the file is | |
556 greater than one. If it is, more than one repository is using the | |
557 file, so Mercurial makes a new copy of the file that is private to | |
558 this repository. | |
559 | |
560 A few revision control developers have pointed out that this idea of | |
561 making a complete private copy of a file is not very efficient in its | |
562 use of storage. While this is true, storage is cheap, and this method | |
563 gives the highest performance while deferring most book-keeping to the | |
564 operating system. An alternative scheme would most likely reduce | |
565 performance and increase the complexity of the software, each of which | |
566 is much more important to the ``feel'' of day-to-day use. | |
567 | |
568 \subsection{Other contents of the dirstate} | |
569 | |
570 Because Mercurial doesn't force you to tell it when you're modifying a | |
571 file, it uses the dirstate to store some extra information so it can | |
572 determine efficiently whether you have modified a file. For each file | |
573 in the working directory, it stores the time that it last modified the | |
574 file itself, and the size of the file at that time. | |
575 | |
576 When you explicitly \hgcmd{add}, \hgcmd{remove}, \hgcmd{rename} or | |
577 \hgcmd{copy} files, Mercurial updates the dirstate so that it knows | |
578 what to do with those files when you commit. | |
579 | |
580 When Mercurial is checking the states of files in the working | |
581 directory, it first checks a file's modification time. If that has | |
582 not changed, the file must not have been modified. If the file's size | |
583 has changed, the file must have been modified. If the modification | |
584 time has changed, but the size has not, only then does Mercurial need | |
585 to read the actual contents of the file to see if they've changed. | |
586 Storing these few extra pieces of information dramatically reduces the | |
587 amount of data that Mercurial needs to read, which yields large | |
588 performance improvements compared to other revision control systems. | |
589 | |
590 %%% Local Variables: | |
591 %%% mode: latex | |
592 %%% TeX-master: "00book" | |
593 %%% End: |