hgbook

annotate es/concepts.tex @ 393:2c2c86824c61

find changesets by author, revision, files, or words in the commit message
began translation of the "behind the scenes" chapter
author Javier Rojas <jerojasro@devnull.li>
date Mon Nov 03 21:00:32 2008 -0500 (2008-11-03)
parents 2fb78d342e07
children 149ea8ae39c4
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@343 27 of that file in a metadata object called a \emph{filelog}. Each entry
jerojasro@343 28 in the filelog contains enough information to reconstruct one revision
jerojasro@343 29 of the file that is being tracked. Filelogs are stored as files in
jerojasro@343 30 the \sdirname{.hg/store/data} directory. A filelog contains two kinds
jerojasro@343 31 of information: revision data, and an index to help Mercurial to find
jerojasro@343 32 a revision efficiently.
jerojasro@343 33
jerojasro@343 34 A file that is large, or has a lot of history, has its filelog stored
jerojasro@343 35 in separate data (``\texttt{.d}'' suffix) and index (``\texttt{.i}''
jerojasro@343 36 suffix) files. For small files without much history, the revision
jerojasro@343 37 data and index are combined in a single ``\texttt{.i}'' file. The
jerojasro@343 38 correspondence between a file in the working directory and the filelog
jerojasro@343 39 that tracks its history in the repository is illustrated in
jerojasro@343 40 figure~\ref{fig:concepts:filelog}.
jerojasro@343 41
jerojasro@343 42 \begin{figure}[ht]
jerojasro@343 43 \centering
jerojasro@343 44 \grafix{filelog}
jerojasro@343 45 \caption{Relationships between files in working directory and
jerojasro@343 46 filelogs in repository}
jerojasro@343 47 \label{fig:concepts:filelog}
jerojasro@343 48 \end{figure}
jerojasro@343 49
jerojasro@343 50 \subsection{Managing tracked files}
jerojasro@343 51
jerojasro@343 52 Mercurial uses a structure called a \emph{manifest} to collect
jerojasro@343 53 together information about the files that it tracks. Each entry in
jerojasro@343 54 the manifest contains information about the files present in a single
jerojasro@343 55 changeset. An entry records which files are present in the changeset,
jerojasro@343 56 the revision of each file, and a few other pieces of file metadata.
jerojasro@343 57
jerojasro@343 58 \subsection{Recording changeset information}
jerojasro@343 59
jerojasro@343 60 The \emph{changelog} contains information about each changeset. Each
jerojasro@343 61 revision records who committed a change, the changeset comment, other
jerojasro@343 62 pieces of changeset-related information, and the revision of the
jerojasro@343 63 manifest to use.
jerojasro@343 64
jerojasro@343 65 \subsection{Relationships between revisions}
jerojasro@343 66
jerojasro@343 67 Within a changelog, a manifest, or a filelog, each revision stores a
jerojasro@343 68 pointer to its immediate parent (or to its two parents, if it's a
jerojasro@343 69 merge revision). As I mentioned above, there are also relationships
jerojasro@343 70 between revisions \emph{across} these structures, and they are
jerojasro@343 71 hierarchical in nature.
jerojasro@343 72
jerojasro@343 73 For every changeset in a repository, there is exactly one revision
jerojasro@343 74 stored in the changelog. Each revision of the changelog contains a
jerojasro@343 75 pointer to a single revision of the manifest. A revision of the
jerojasro@343 76 manifest stores a pointer to a single revision of each filelog tracked
jerojasro@343 77 when that changeset was created. These relationships are illustrated
jerojasro@343 78 in figure~\ref{fig:concepts:metadata}.
jerojasro@343 79
jerojasro@343 80 \begin{figure}[ht]
jerojasro@343 81 \centering
jerojasro@343 82 \grafix{metadata}
jerojasro@343 83 \caption{Metadata relationships}
jerojasro@343 84 \label{fig:concepts:metadata}
jerojasro@343 85 \end{figure}
jerojasro@343 86
jerojasro@343 87 As the illustration shows, there is \emph{not} a ``one to one''
jerojasro@343 88 relationship between revisions in the changelog, manifest, or filelog.
jerojasro@343 89 If the manifest hasn't changed between two changesets, the changelog
jerojasro@343 90 entries for those changesets will point to the same revision of the
jerojasro@343 91 manifest. If a file that Mercurial tracks hasn't changed between two
jerojasro@343 92 changesets, the entry for that file in the two revisions of the
jerojasro@343 93 manifest will point to the same revision of its filelog.
jerojasro@343 94
jerojasro@343 95 \section{Safe, efficient storage}
jerojasro@343 96
jerojasro@343 97 The underpinnings of changelogs, manifests, and filelogs are provided
jerojasro@343 98 by a single structure called the \emph{revlog}.
jerojasro@343 99
jerojasro@343 100 \subsection{Efficient storage}
jerojasro@343 101
jerojasro@343 102 The revlog provides efficient storage of revisions using a
jerojasro@343 103 \emph{delta} mechanism. Instead of storing a complete copy of a file
jerojasro@343 104 for each revision, it stores the changes needed to transform an older
jerojasro@343 105 revision into the new revision. For many kinds of file data, these
jerojasro@343 106 deltas are typically a fraction of a percent of the size of a full
jerojasro@343 107 copy of a file.
jerojasro@343 108
jerojasro@343 109 Some obsolete revision control systems can only work with deltas of
jerojasro@343 110 text files. They must either store binary files as complete snapshots
jerojasro@343 111 or encoded into a text representation, both of which are wasteful
jerojasro@343 112 approaches. Mercurial can efficiently handle deltas of files with
jerojasro@343 113 arbitrary binary contents; it doesn't need to treat text as special.
jerojasro@343 114
jerojasro@343 115 \subsection{Safe operation}
jerojasro@343 116 \label{sec:concepts:txn}
jerojasro@343 117
jerojasro@343 118 Mercurial only ever \emph{appends} data to the end of a revlog file.
jerojasro@343 119 It never modifies a section of a file after it has written it. This
jerojasro@343 120 is both more robust and efficient than schemes that need to modify or
jerojasro@343 121 rewrite data.
jerojasro@343 122
jerojasro@343 123 In addition, Mercurial treats every write as part of a
jerojasro@343 124 \emph{transaction} that can span a number of files. A transaction is
jerojasro@343 125 \emph{atomic}: either the entire transaction succeeds and its effects
jerojasro@343 126 are all visible to readers in one go, or the whole thing is undone.
jerojasro@343 127 This guarantee of atomicity means that if you're running two copies of
jerojasro@343 128 Mercurial, where one is reading data and one is writing it, the reader
jerojasro@343 129 will never see a partially written result that might confuse it.
jerojasro@343 130
jerojasro@343 131 The fact that Mercurial only appends to files makes it easier to
jerojasro@343 132 provide this transactional guarantee. The easier it is to do stuff
jerojasro@343 133 like this, the more confident you should be that it's done correctly.
jerojasro@343 134
jerojasro@343 135 \subsection{Fast retrieval}
jerojasro@343 136
jerojasro@343 137 Mercurial cleverly avoids a pitfall common to all earlier
jerojasro@343 138 revision control systems: the problem of \emph{inefficient retrieval}.
jerojasro@343 139 Most revision control systems store the contents of a revision as an
jerojasro@343 140 incremental series of modifications against a ``snapshot''. To
jerojasro@343 141 reconstruct a specific revision, you must first read the snapshot, and
jerojasro@343 142 then every one of the revisions between the snapshot and your target
jerojasro@343 143 revision. The more history that a file accumulates, the more
jerojasro@343 144 revisions you must read, hence the longer it takes to reconstruct a
jerojasro@343 145 particular revision.
jerojasro@343 146
jerojasro@343 147 \begin{figure}[ht]
jerojasro@343 148 \centering
jerojasro@343 149 \grafix{snapshot}
jerojasro@343 150 \caption{Snapshot of a revlog, with incremental deltas}
jerojasro@343 151 \label{fig:concepts:snapshot}
jerojasro@343 152 \end{figure}
jerojasro@343 153
jerojasro@343 154 The innovation that Mercurial applies to this problem is simple but
jerojasro@343 155 effective. Once the cumulative amount of delta information stored
jerojasro@343 156 since the last snapshot exceeds a fixed threshold, it stores a new
jerojasro@343 157 snapshot (compressed, of course), instead of another delta. This
jerojasro@343 158 makes it possible to reconstruct \emph{any} revision of a file
jerojasro@343 159 quickly. This approach works so well that it has since been copied by
jerojasro@343 160 several other revision control systems.
jerojasro@343 161
jerojasro@343 162 Figure~\ref{fig:concepts:snapshot} illustrates the idea. In an entry
jerojasro@343 163 in a revlog's index file, Mercurial stores the range of entries from
jerojasro@343 164 the data file that it must read to reconstruct a particular revision.
jerojasro@343 165
jerojasro@343 166 \subsubsection{Aside: the influence of video compression}
jerojasro@343 167
jerojasro@343 168 If you're familiar with video compression or have ever watched a TV
jerojasro@343 169 feed through a digital cable or satellite service, you may know that
jerojasro@343 170 most video compression schemes store each frame of video as a delta
jerojasro@343 171 against its predecessor frame. In addition, these schemes use
jerojasro@343 172 ``lossy'' compression techniques to increase the compression ratio, so
jerojasro@343 173 visual errors accumulate over the course of a number of inter-frame
jerojasro@343 174 deltas.
jerojasro@343 175
jerojasro@343 176 Because it's possible for a video stream to ``drop out'' occasionally
jerojasro@343 177 due to signal glitches, and to limit the accumulation of artefacts
jerojasro@343 178 introduced by the lossy compression process, video encoders
jerojasro@343 179 periodically insert a complete frame (called a ``key frame'') into the
jerojasro@343 180 video stream; the next delta is generated against that frame. This
jerojasro@343 181 means that if the video signal gets interrupted, it will resume once
jerojasro@343 182 the next key frame is received. Also, the accumulation of encoding
jerojasro@343 183 errors restarts anew with each key frame.
jerojasro@343 184
jerojasro@343 185 \subsection{Identification and strong integrity}
jerojasro@343 186
jerojasro@343 187 Along with delta or snapshot information, a revlog entry contains a
jerojasro@343 188 cryptographic hash of the data that it represents. This makes it
jerojasro@343 189 difficult to forge the contents of a revision, and easy to detect
jerojasro@343 190 accidental corruption.
jerojasro@343 191
jerojasro@343 192 Hashes provide more than a mere check against corruption; they are
jerojasro@343 193 used as the identifiers for revisions. The changeset identification
jerojasro@343 194 hashes that you see as an end user are from revisions of the
jerojasro@343 195 changelog. Although filelogs and the manifest also use hashes,
jerojasro@343 196 Mercurial only uses these behind the scenes.
jerojasro@343 197
jerojasro@343 198 Mercurial verifies that hashes are correct when it retrieves file
jerojasro@343 199 revisions and when it pulls changes from another repository. If it
jerojasro@343 200 encounters an integrity problem, it will complain and stop whatever
jerojasro@343 201 it's doing.
jerojasro@343 202
jerojasro@343 203 In addition to the effect it has on retrieval efficiency, Mercurial's
jerojasro@343 204 use of periodic snapshots makes it more robust against partial data
jerojasro@343 205 corruption. If a revlog becomes partly corrupted due to a hardware
jerojasro@343 206 error or system bug, it's often possible to reconstruct some or most
jerojasro@343 207 revisions from the uncorrupted sections of the revlog, both before and
jerojasro@343 208 after the corrupted section. This would not be possible with a
jerojasro@343 209 delta-only storage model.
jerojasro@343 210
jerojasro@343 211 \section{Revision history, branching,
jerojasro@343 212 and merging}
jerojasro@343 213
jerojasro@343 214 Every entry in a Mercurial revlog knows the identity of its immediate
jerojasro@343 215 ancestor revision, usually referred to as its \emph{parent}. In fact,
jerojasro@343 216 a revision contains room for not one parent, but two. Mercurial uses
jerojasro@343 217 a special hash, called the ``null ID'', to represent the idea ``there
jerojasro@343 218 is no parent here''. This hash is simply a string of zeroes.
jerojasro@343 219
jerojasro@343 220 In figure~\ref{fig:concepts:revlog}, you can see an example of the
jerojasro@343 221 conceptual structure of a revlog. Filelogs, manifests, and changelogs
jerojasro@343 222 all have this same structure; they differ only in the kind of data
jerojasro@343 223 stored in each delta or snapshot.
jerojasro@343 224
jerojasro@343 225 The first revision in a revlog (at the bottom of the image) has the
jerojasro@343 226 null ID in both of its parent slots. For a ``normal'' revision, its
jerojasro@343 227 first parent slot contains the ID of its parent revision, and its
jerojasro@343 228 second contains the null ID, indicating that the revision has only one
jerojasro@343 229 real parent. Any two revisions that have the same parent ID are
jerojasro@343 230 branches. A revision that represents a merge between branches has two
jerojasro@343 231 normal revision IDs in its parent slots.
jerojasro@343 232
jerojasro@343 233 \begin{figure}[ht]
jerojasro@343 234 \centering
jerojasro@343 235 \grafix{revlog}
jerojasro@343 236 \caption{}
jerojasro@343 237 \label{fig:concepts:revlog}
jerojasro@343 238 \end{figure}
jerojasro@343 239
jerojasro@343 240 \section{The working directory}
jerojasro@343 241
jerojasro@343 242 In the working directory, Mercurial stores a snapshot of the files
jerojasro@343 243 from the repository as of a particular changeset.
jerojasro@343 244
jerojasro@343 245 The working directory ``knows'' which changeset it contains. When you
jerojasro@343 246 update the working directory to contain a particular changeset,
jerojasro@343 247 Mercurial looks up the appropriate revision of the manifest to find
jerojasro@343 248 out which files it was tracking at the time that changeset was
jerojasro@343 249 committed, and which revision of each file was then current. It then
jerojasro@343 250 recreates a copy of each of those files, with the same contents it had
jerojasro@343 251 when the changeset was committed.
jerojasro@343 252
jerojasro@343 253 The \emph{dirstate} contains Mercurial's knowledge of the working
jerojasro@343 254 directory. This details which changeset the working directory is
jerojasro@343 255 updated to, and all of the files that Mercurial is tracking in the
jerojasro@343 256 working directory.
jerojasro@343 257
jerojasro@343 258 Just as a revision of a revlog has room for two parents, so that it
jerojasro@343 259 can represent either a normal revision (with one parent) or a merge of
jerojasro@343 260 two earlier revisions, the dirstate has slots for two parents. When
jerojasro@343 261 you use the \hgcmd{update} command, the changeset that you update to
jerojasro@343 262 is stored in the ``first parent'' slot, and the null ID in the second.
jerojasro@343 263 When you \hgcmd{merge} with another changeset, the first parent
jerojasro@343 264 remains unchanged, and the second parent is filled in with the
jerojasro@343 265 changeset you're merging with. The \hgcmd{parents} command tells you
jerojasro@343 266 what the parents of the dirstate are.
jerojasro@343 267
jerojasro@343 268 \subsection{What happens when you commit}
jerojasro@343 269
jerojasro@343 270 The dirstate stores parent information for more than just book-keeping
jerojasro@343 271 purposes. Mercurial uses the parents of the dirstate as \emph{the
jerojasro@343 272 parents of a new changeset} when you perform a commit.
jerojasro@343 273
jerojasro@343 274 \begin{figure}[ht]
jerojasro@343 275 \centering
jerojasro@343 276 \grafix{wdir}
jerojasro@343 277 \caption{The working directory can have two parents}
jerojasro@343 278 \label{fig:concepts:wdir}
jerojasro@343 279 \end{figure}
jerojasro@343 280
jerojasro@343 281 Figure~\ref{fig:concepts:wdir} shows the normal state of the working
jerojasro@343 282 directory, where it has a single changeset as parent. That changeset
jerojasro@343 283 is the \emph{tip}, the newest changeset in the repository that has no
jerojasro@343 284 children.
jerojasro@343 285
jerojasro@343 286 \begin{figure}[ht]
jerojasro@343 287 \centering
jerojasro@343 288 \grafix{wdir-after-commit}
jerojasro@343 289 \caption{The working directory gains new parents after a commit}
jerojasro@343 290 \label{fig:concepts:wdir-after-commit}
jerojasro@343 291 \end{figure}
jerojasro@343 292
jerojasro@343 293 It's useful to think of the working directory as ``the changeset I'm
jerojasro@343 294 about to commit''. Any files that you tell Mercurial that you've
jerojasro@343 295 added, removed, renamed, or copied will be reflected in that
jerojasro@343 296 changeset, as will modifications to any files that Mercurial is
jerojasro@343 297 already tracking; the new changeset will have the parents of the
jerojasro@343 298 working directory as its parents.
jerojasro@343 299
jerojasro@343 300 After a commit, Mercurial will update the parents of the working
jerojasro@343 301 directory, so that the first parent is the ID of the new changeset,
jerojasro@343 302 and the second is the null ID. This is shown in
jerojasro@343 303 figure~\ref{fig:concepts:wdir-after-commit}. Mercurial doesn't touch
jerojasro@343 304 any of the files in the working directory when you commit; it just
jerojasro@343 305 modifies the dirstate to note its new parents.
jerojasro@343 306
jerojasro@343 307 \subsection{Creating a new head}
jerojasro@343 308
jerojasro@343 309 It's perfectly normal to update the working directory to a changeset
jerojasro@343 310 other than the current tip. For example, you might want to know what
jerojasro@343 311 your project looked like last Tuesday, or you could be looking through
jerojasro@343 312 changesets to see which one introduced a bug. In cases like this, the
jerojasro@343 313 natural thing to do is update the working directory to the changeset
jerojasro@343 314 you're interested in, and then examine the files in the working
jerojasro@343 315 directory directly to see their contents as they werea when you
jerojasro@343 316 committed that changeset. The effect of this is shown in
jerojasro@343 317 figure~\ref{fig:concepts:wdir-pre-branch}.
jerojasro@343 318
jerojasro@343 319 \begin{figure}[ht]
jerojasro@343 320 \centering
jerojasro@343 321 \grafix{wdir-pre-branch}
jerojasro@343 322 \caption{The working directory, updated to an older changeset}
jerojasro@343 323 \label{fig:concepts:wdir-pre-branch}
jerojasro@343 324 \end{figure}
jerojasro@343 325
jerojasro@343 326 Having updated the working directory to an older changeset, what
jerojasro@343 327 happens if you make some changes, and then commit? Mercurial behaves
jerojasro@343 328 in the same way as I outlined above. The parents of the working
jerojasro@343 329 directory become the parents of the new changeset. This new changeset
jerojasro@343 330 has no children, so it becomes the new tip. And the repository now
jerojasro@343 331 contains two changesets that have no children; we call these
jerojasro@343 332 \emph{heads}. You can see the structure that this creates in
jerojasro@343 333 figure~\ref{fig:concepts:wdir-branch}.
jerojasro@343 334
jerojasro@343 335 \begin{figure}[ht]
jerojasro@343 336 \centering
jerojasro@343 337 \grafix{wdir-branch}
jerojasro@343 338 \caption{After a commit made while synced to an older changeset}
jerojasro@343 339 \label{fig:concepts:wdir-branch}
jerojasro@343 340 \end{figure}
jerojasro@343 341
jerojasro@343 342 \begin{note}
jerojasro@343 343 If you're new to Mercurial, you should keep in mind a common
jerojasro@343 344 ``error'', which is to use the \hgcmd{pull} command without any
jerojasro@343 345 options. By default, the \hgcmd{pull} command \emph{does not}
jerojasro@343 346 update the working directory, so you'll bring new changesets into
jerojasro@343 347 your repository, but the working directory will stay synced at the
jerojasro@343 348 same changeset as before the pull. If you make some changes and
jerojasro@343 349 commit afterwards, you'll thus create a new head, because your
jerojasro@343 350 working directory isn't synced to whatever the current tip is.
jerojasro@343 351
jerojasro@343 352 I put the word ``error'' in quotes because all that you need to do
jerojasro@343 353 to rectify this situation is \hgcmd{merge}, then \hgcmd{commit}. In
jerojasro@343 354 other words, this almost never has negative consequences; it just
jerojasro@343 355 surprises people. I'll discuss other ways to avoid this behaviour,
jerojasro@343 356 and why Mercurial behaves in this initially surprising way, later
jerojasro@343 357 on.
jerojasro@343 358 \end{note}
jerojasro@343 359
jerojasro@343 360 \subsection{Merging heads}
jerojasro@343 361
jerojasro@343 362 When you run the \hgcmd{merge} command, Mercurial leaves the first
jerojasro@343 363 parent of the working directory unchanged, and sets the second parent
jerojasro@343 364 to the changeset you're merging with, as shown in
jerojasro@343 365 figure~\ref{fig:concepts:wdir-merge}.
jerojasro@343 366
jerojasro@343 367 \begin{figure}[ht]
jerojasro@343 368 \centering
jerojasro@343 369 \grafix{wdir-merge}
jerojasro@343 370 \caption{Merging two heads}
jerojasro@343 371 \label{fig:concepts:wdir-merge}
jerojasro@343 372 \end{figure}
jerojasro@343 373
jerojasro@343 374 Mercurial also has to modify the working directory, to merge the files
jerojasro@343 375 managed in the two changesets. Simplified a little, the merging
jerojasro@343 376 process goes like this, for every file in the manifests of both
jerojasro@343 377 changesets.
jerojasro@343 378 \begin{itemize}
jerojasro@343 379 \item If neither changeset has modified a file, do nothing with that
jerojasro@343 380 file.
jerojasro@343 381 \item If one changeset has modified a file, and the other hasn't,
jerojasro@343 382 create the modified copy of the file in the working directory.
jerojasro@343 383 \item If one changeset has removed a file, and the other hasn't (or
jerojasro@343 384 has also deleted it), delete the file from the working directory.
jerojasro@343 385 \item If one changeset has removed a file, but the other has modified
jerojasro@343 386 the file, ask the user what to do: keep the modified file, or remove
jerojasro@343 387 it?
jerojasro@343 388 \item If both changesets have modified a file, invoke an external
jerojasro@343 389 merge program to choose the new contents for the merged file. This
jerojasro@343 390 may require input from the user.
jerojasro@343 391 \item If one changeset has modified a file, and the other has renamed
jerojasro@343 392 or copied the file, make sure that the changes follow the new name
jerojasro@343 393 of the file.
jerojasro@343 394 \end{itemize}
jerojasro@343 395 There are more details---merging has plenty of corner cases---but
jerojasro@343 396 these are the most common choices that are involved in a merge. As
jerojasro@343 397 you can see, most cases are completely automatic, and indeed most
jerojasro@343 398 merges finish automatically, without requiring your input to resolve
jerojasro@343 399 any conflicts.
jerojasro@343 400
jerojasro@343 401 When you're thinking about what happens when you commit after a merge,
jerojasro@343 402 once again the working directory is ``the changeset I'm about to
jerojasro@343 403 commit''. After the \hgcmd{merge} command completes, the working
jerojasro@343 404 directory has two parents; these will become the parents of the new
jerojasro@343 405 changeset.
jerojasro@343 406
jerojasro@343 407 Mercurial lets you perform multiple merges, but you must commit the
jerojasro@343 408 results of each individual merge as you go. This is necessary because
jerojasro@343 409 Mercurial only tracks two parents for both revisions and the working
jerojasro@343 410 directory. While it would be technically possible to merge multiple
jerojasro@343 411 changesets at once, the prospect of user confusion and making a
jerojasro@343 412 terrible mess of a merge immediately becomes overwhelming.
jerojasro@343 413
jerojasro@343 414 \section{Other interesting design features}
jerojasro@343 415
jerojasro@343 416 In the sections above, I've tried to highlight some of the most
jerojasro@343 417 important aspects of Mercurial's design, to illustrate that it pays
jerojasro@343 418 careful attention to reliability and performance. However, the
jerojasro@343 419 attention to detail doesn't stop there. There are a number of other
jerojasro@343 420 aspects of Mercurial's construction that I personally find
jerojasro@343 421 interesting. I'll detail a few of them here, separate from the ``big
jerojasro@343 422 ticket'' items above, so that if you're interested, you can gain a
jerojasro@343 423 better idea of the amount of thinking that goes into a well-designed
jerojasro@343 424 system.
jerojasro@343 425
jerojasro@343 426 \subsection{Clever compression}
jerojasro@343 427
jerojasro@343 428 When appropriate, Mercurial will store both snapshots and deltas in
jerojasro@343 429 compressed form. It does this by always \emph{trying to} compress a
jerojasro@343 430 snapshot or delta, but only storing the compressed version if it's
jerojasro@343 431 smaller than the uncompressed version.
jerojasro@343 432
jerojasro@343 433 This means that Mercurial does ``the right thing'' when storing a file
jerojasro@343 434 whose native form is compressed, such as a \texttt{zip} archive or a
jerojasro@343 435 JPEG image. When these types of files are compressed a second time,
jerojasro@343 436 the resulting file is usually bigger than the once-compressed form,
jerojasro@343 437 and so Mercurial will store the plain \texttt{zip} or JPEG.
jerojasro@343 438
jerojasro@343 439 Deltas between revisions of a compressed file are usually larger than
jerojasro@343 440 snapshots of the file, and Mercurial again does ``the right thing'' in
jerojasro@343 441 these cases. It finds that such a delta exceeds the threshold at
jerojasro@343 442 which it should store a complete snapshot of the file, so it stores
jerojasro@343 443 the snapshot, again saving space compared to a naive delta-only
jerojasro@343 444 approach.
jerojasro@343 445
jerojasro@343 446 \subsubsection{Network recompression}
jerojasro@343 447
jerojasro@343 448 When storing revisions on disk, Mercurial uses the ``deflate''
jerojasro@343 449 compression algorithm (the same one used by the popular \texttt{zip}
jerojasro@343 450 archive format), which balances good speed with a respectable
jerojasro@343 451 compression ratio. However, when transmitting revision data over a
jerojasro@343 452 network connection, Mercurial uncompresses the compressed revision
jerojasro@343 453 data.
jerojasro@343 454
jerojasro@343 455 If the connection is over HTTP, Mercurial recompresses the entire
jerojasro@343 456 stream of data using a compression algorithm that gives a better
jerojasro@343 457 compression ratio (the Burrows-Wheeler algorithm from the widely used
jerojasro@343 458 \texttt{bzip2} compression package). This combination of algorithm
jerojasro@343 459 and compression of the entire stream (instead of a revision at a time)
jerojasro@343 460 substantially reduces the number of bytes to be transferred, yielding
jerojasro@343 461 better network performance over almost all kinds of network.
jerojasro@343 462
jerojasro@343 463 (If the connection is over \command{ssh}, Mercurial \emph{doesn't}
jerojasro@343 464 recompress the stream, because \command{ssh} can already do this
jerojasro@343 465 itself.)
jerojasro@343 466
jerojasro@343 467 \subsection{Read/write ordering and atomicity}
jerojasro@343 468
jerojasro@343 469 Appending to files isn't the whole story when it comes to guaranteeing
jerojasro@343 470 that a reader won't see a partial write. If you recall
jerojasro@343 471 figure~\ref{fig:concepts:metadata}, revisions in the changelog point to
jerojasro@343 472 revisions in the manifest, and revisions in the manifest point to
jerojasro@343 473 revisions in filelogs. This hierarchy is deliberate.
jerojasro@343 474
jerojasro@343 475 A writer starts a transaction by writing filelog and manifest data,
jerojasro@343 476 and doesn't write any changelog data until those are finished. A
jerojasro@343 477 reader starts by reading changelog data, then manifest data, followed
jerojasro@343 478 by filelog data.
jerojasro@343 479
jerojasro@343 480 Since the writer has always finished writing filelog and manifest data
jerojasro@343 481 before it writes to the changelog, a reader will never read a pointer
jerojasro@343 482 to a partially written manifest revision from the changelog, and it will
jerojasro@343 483 never read a pointer to a partially written filelog revision from the
jerojasro@343 484 manifest.
jerojasro@343 485
jerojasro@343 486 \subsection{Concurrent access}
jerojasro@343 487
jerojasro@343 488 The read/write ordering and atomicity guarantees mean that Mercurial
jerojasro@343 489 never needs to \emph{lock} a repository when it's reading data, even
jerojasro@343 490 if the repository is being written to while the read is occurring.
jerojasro@343 491 This has a big effect on scalability; you can have an arbitrary number
jerojasro@343 492 of Mercurial processes safely reading data from a repository safely
jerojasro@343 493 all at once, no matter whether it's being written to or not.
jerojasro@343 494
jerojasro@343 495 The lockless nature of reading means that if you're sharing a
jerojasro@343 496 repository on a multi-user system, you don't need to grant other local
jerojasro@343 497 users permission to \emph{write} to your repository in order for them
jerojasro@343 498 to be able to clone it or pull changes from it; they only need
jerojasro@343 499 \emph{read} permission. (This is \emph{not} a common feature among
jerojasro@343 500 revision control systems, so don't take it for granted! Most require
jerojasro@343 501 readers to be able to lock a repository to access it safely, and this
jerojasro@343 502 requires write permission on at least one directory, which of course
jerojasro@343 503 makes for all kinds of nasty and annoying security and administrative
jerojasro@343 504 problems.)
jerojasro@343 505
jerojasro@343 506 Mercurial uses locks to ensure that only one process can write to a
jerojasro@343 507 repository at a time (the locking mechanism is safe even over
jerojasro@343 508 filesystems that are notoriously hostile to locking, such as NFS). If
jerojasro@343 509 a repository is locked, a writer will wait for a while to retry if the
jerojasro@343 510 repository becomes unlocked, but if the repository remains locked for
jerojasro@343 511 too long, the process attempting to write will time out after a while.
jerojasro@343 512 This means that your daily automated scripts won't get stuck forever
jerojasro@343 513 and pile up if a system crashes unnoticed, for example. (Yes, the
jerojasro@343 514 timeout is configurable, from zero to infinity.)
jerojasro@343 515
jerojasro@343 516 \subsubsection{Safe dirstate access}
jerojasro@343 517
jerojasro@343 518 As with revision data, Mercurial doesn't take a lock to read the
jerojasro@343 519 dirstate file; it does acquire a lock to write it. To avoid the
jerojasro@343 520 possibility of reading a partially written copy of the dirstate file,
jerojasro@343 521 Mercurial writes to a file with a unique name in the same directory as
jerojasro@343 522 the dirstate file, then renames the temporary file atomically to
jerojasro@343 523 \filename{dirstate}. The file named \filename{dirstate} is thus
jerojasro@343 524 guaranteed to be complete, not partially written.
jerojasro@343 525
jerojasro@343 526 \subsection{Avoiding seeks}
jerojasro@343 527
jerojasro@343 528 Critical to Mercurial's performance is the avoidance of seeks of the
jerojasro@343 529 disk head, since any seek is far more expensive than even a
jerojasro@343 530 comparatively large read operation.
jerojasro@343 531
jerojasro@343 532 This is why, for example, the dirstate is stored in a single file. If
jerojasro@343 533 there were a dirstate file per directory that Mercurial tracked, the
jerojasro@343 534 disk would seek once per directory. Instead, Mercurial reads the
jerojasro@343 535 entire single dirstate file in one step.
jerojasro@343 536
jerojasro@343 537 Mercurial also uses a ``copy on write'' scheme when cloning a
jerojasro@343 538 repository on local storage. Instead of copying every revlog file
jerojasro@343 539 from the old repository into the new repository, it makes a ``hard
jerojasro@343 540 link'', which is a shorthand way to say ``these two names point to the
jerojasro@343 541 same file''. When Mercurial is about to write to one of a revlog's
jerojasro@343 542 files, it checks to see if the number of names pointing at the file is
jerojasro@343 543 greater than one. If it is, more than one repository is using the
jerojasro@343 544 file, so Mercurial makes a new copy of the file that is private to
jerojasro@343 545 this repository.
jerojasro@343 546
jerojasro@343 547 A few revision control developers have pointed out that this idea of
jerojasro@343 548 making a complete private copy of a file is not very efficient in its
jerojasro@343 549 use of storage. While this is true, storage is cheap, and this method
jerojasro@343 550 gives the highest performance while deferring most book-keeping to the
jerojasro@343 551 operating system. An alternative scheme would most likely reduce
jerojasro@343 552 performance and increase the complexity of the software, each of which
jerojasro@343 553 is much more important to the ``feel'' of day-to-day use.
jerojasro@343 554
jerojasro@343 555 \subsection{Other contents of the dirstate}
jerojasro@343 556
jerojasro@343 557 Because Mercurial doesn't force you to tell it when you're modifying a
jerojasro@343 558 file, it uses the dirstate to store some extra information so it can
jerojasro@343 559 determine efficiently whether you have modified a file. For each file
jerojasro@343 560 in the working directory, it stores the time that it last modified the
jerojasro@343 561 file itself, and the size of the file at that time.
jerojasro@343 562
jerojasro@343 563 When you explicitly \hgcmd{add}, \hgcmd{remove}, \hgcmd{rename} or
jerojasro@343 564 \hgcmd{copy} files, Mercurial updates the dirstate so that it knows
jerojasro@343 565 what to do with those files when you commit.
jerojasro@343 566
jerojasro@343 567 When Mercurial is checking the states of files in the working
jerojasro@343 568 directory, it first checks a file's modification time. If that has
jerojasro@343 569 not changed, the file must not have been modified. If the file's size
jerojasro@343 570 has changed, the file must have been modified. If the modification
jerojasro@343 571 time has changed, but the size has not, only then does Mercurial need
jerojasro@343 572 to read the actual contents of the file to see if they've changed.
jerojasro@343 573 Storing these few extra pieces of information dramatically reduces the
jerojasro@343 574 amount of data that Mercurial needs to read, which yields large
jerojasro@343 575 performance improvements compared to other revision control systems.
jerojasro@343 576
jerojasro@343 577 %%% Local Variables:
jerojasro@343 578 %%% mode: latex
jerojasro@343 579 %%% TeX-master: "00book"
jerojasro@343 580 %%% End: