www/cvs2git.html

   1 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
   2 "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
   3 <html xmlns="http://www.w3.org/1999/xhtml">
   4 <head>
   5 <style type="text/css"> /* <![CDATA[ */
   6   @import "tigris-branding/css/tigris.css";
   7   @import "tigris-branding/css/inst.css";
   8   /* ]]> */</style>
   9 <link rel="stylesheet" type="text/css" media="print"
  10   href="tigris-branding/css/print.css"/>
  11 <script type="text/javascript" src="tigris-branding/scripts/tigris.js"></script>
  12 <title>cvs2git Documentation</title>
  13 </head>
  14
  15 <body id="bodycol">
  16 <div class="app">
  17
  18 <h1>cvs2git</h1>
  19
  20 <h2>Index</h2>
  21
  22 <ul>
  23
  24   <li><a href="#intro">Introduction</a></li>
  25
  26   <li><a href="#reqs">Requirements</a></li>
  27
  28   <li><a href="#status">Development status</a></li>
  29
  30   <li><a href="#docs">Documentation</a></li>
  31
  32   <li><a href="#usage">Usage</a></li>
  33
  34 </ul>
  35
  36 <hr />
  37
  38 <h2><a name="intro">Introduction</a></h2>
  39
  40 <p>cvs2svn/cvs2git is a tool that can be used to migrate CVS
  41 repositories to newer version control tools, including <a
  42 href="http://git.or.cz/">git</a>.  git is a distributed version
  43 control system most famous for being used for Linux kernel
  44 development.  The program used to convert to git, called cvs2git, is
  45 distributed as part of the cvs2svn project.</p>
  46
  47 <p><strong>If you are reading this documentation on the <a
  48 href="http://cvs2svn.tigris.org">cvs2svn website</a>, then please be
  49 aware that it describes the current trunk version of cvs2svn, which
  50 may be different than the most recent released version.  Please refer
  51 to the documentation that was included with your version of cvs2svn.
  52 </strong></p>
  53
  54 <p>Conversion to git was added in release 2.1 of cvs2svn and has
  55 improved significantly since then.  Please make sure you are using an
  56 up-to-date version of cvs2svn--perhaps even the development trunk
  57 version.</p>
  58
  59
  60 <h2><a name="reqs">Requirements</a></h2>
  61
  62 <p>cvs2git requires the following:</p>
  63
  64 <ul>
  65
  66   <li>Direct (filesystem) access to a copy of the CVS repository that
  67     you want to convert.  cvs2git parses the files in the CVS
  68     repository directly, so it is not enough to have remote CVS
  69     access.  See the <a href="faq.html#repoaccess">FAQ</a> for more
  70     information and a possible workaround.</li>
  71
  72   <li>Python 2, version 2.4 or later.  See <a
  73     href="http://www.python.org/">http://www.python.org/</a>.
  74     (cvs2git does <strong>not</strong> work with Python 3.x.)</li>
  75
  76   <li>If you use the <tt>--use-rcs</tt> option, then RCS's `co'
  77     program is required.  The RCS home page is
  78     <a href="http://www.cs.purdue.edu/homes/trinkle/RCS/"
  79             >http://www.cs.purdue.edu/homes/trinkle/RCS/</a>.
  80     See the <a href="cvs2svn.html#use-rcs"><tt>--use-rcs</tt> flag</a> for more
  81     details.</li>
  82
  83   <li>If you use the <tt>--use-cvs</tt> option, then the `cvs' command
  84     is required.  The CVS home page is
  85     <a href="http://ccvs.cvshome.org/">http://ccvs.cvshome.org/</a>.
  86     See the <a href="cvs2svn.html#use-cvs"><tt>--use-cvs</tt> flag</a> for more
  87     details.</li>
  88
  89   <li> Git version 1.5.4.4 or later (earlier versions have a bug in
  90     "git fast-import" that prevent them from loading the files
  91     generated by cvs2git).</li>
  92
  93 </ul>
  94
  95
  96 <h2><a name="status">Development status</a></h2>
  97
  98 <p>Most of the work of converting a repository from CVS to a more
  99 modern version control system is inferring the most likely history
 100 given the incomplete information that CVS records.  cvs2svn has a long
 101 history of making sense of even the most convoluted CVS repositories,
 102 and cvs2git uses this same machinery.  Therefore, cvs2git inherits the
 103 robustness and many of the <a href="features.html">features of
 104 cvs2svn</a>.  cvs2svn can convert just about every CVS repository we
 105 have ever seen, and includes a plethora of options for customizing
 106 your conversion.</p>
 107
 108 <p>The output of cvs2git is one or more dump files that can be
 109 imported into git using the excellent <a
 110 href="http://www.kernel.org/pub/software/scm/git/docs/git
 111 fast-import.html">git fast-import</a> tool.</p>
 112
 113 <p>Although cvs2git is considerably newer than cvs2svn, and much less
 114 well tested, it is believed that cvs2git can be used for production
 115 conversions.  If you use cvs2git, please let us know how it worked for
 116 you!</p>
 117
 118
 119 <h2><a name="limitations">cvs2git limitations</a></h2>
 120
 121 <p>cvs2git still has many limitations compared to
 122 cvs2svn.  <strong>Help would be much appreciated!</strong> Some of
 123 these missing features would be pretty easy to program, and I'd be
 124 happy to help you get started.</p>
 125
 126 <ul>
 127
 128   <li>The cvs2git documentation is still not as complete as that for
 129     cvs2svn.  See <a href="#docs">below</a> for more references.</li>
 130
 131   <li>Differences between CVS and git branch/tag models: CVS allows a
 132     branch or tag to be created from arbitrary combinations of source
 133     revisions from multiple source branches.  It even allows file
 134     revisions that were never contemporaneous to be added to a single
 135     branch/tag.  Git, on the other hand, only allows the full source
 136     tree, as it existed at some instant in the history, to be branched
 137     or tagged as a unit.  Moreover, the ancestry of a git revision
 138     makes <a
 139     href="http://softwareswirl.blogspot.com/2009/08/git-mercurial-and-bazaarsimplicity.html">implications
 140     about the contents of that revision</a>.  This difference means
 141     that it is fundamentally impossible to represent an arbitrary CVS
 142     history in a git repository 100% faithfully.  cvs2git uses the
 143     following workarounds:
 144
 145     <ul>
 146
 147       <li>cvs2git tries to create a branch from a single source, but
 148         if it can't figure out how to, it creates the branch using a
 149         "merge" from multiple source branches.  In pathological
 150         situations, the number of merge sources for a branch can be
 151         arbitrarily large.  The resulting history implies that
 152         whenever any file was added to a branch, the <em>entire</em>
 153         source branch was merged into the destination branch, which is
 154         clearly incorrect.  (The alternative, to omit the merge, would
 155         discard the information that <em>some</em> content was moved
 156         from one branch to the other.)</li>
 157
 158       <li>If cvs2git cannot determine that a CVS tag can be created
 159         from a single revision, then it creates a <a
 160         href="http://www.kernel.org/pub/software/scm/git/docs/git-fast-import.html">tag
 161         fixup branch</a> named <tt>TAG.FIXUP</tt>, then tags this
 162         branch.  (This is a necessary workaround for the fact that git
 163         only allows existing revisions to be tagged.)  The
 164         <tt>TAG.FIXUP</tt> branch is created as a merge between all of
 165         the branches that contain file revisions included in the tag,
 166         which involves the same tradeoff described above for branches.
 167         The <tt>TAG.FIXUP</tt> branch is cleared at the end of the
 168         conversion, but (due to a technical limitation of the git
 169         fast-import file format) not deleted.  There are some
 170         situations when a tag <em>could</em> be created from a single
 171         revision, but cvs2git does not realize it and creates a
 172         superfluous tag fixup branch.  It is possible to delete
 173         superfluous tag fixup branches after the conversion by running
 174         the <tt>contrib/git-move-refs.py</tt> script within the
 175         resulting git repository.</li>
 176
 177     </ul>
 178
 179   </li>
 180
 181   <li>There are no checks that CVS branch and tag names are legal git
 182     names.  There are probably other git constraints that should also
 183     be checked.</li>
 184
 185   <li>The data that should be fed to git fast-import are written to
 186     two files, which have to be loaded into git fast-import manually.
 187     These files might grow to very large size.  It would be nice to
 188     add an option to invoke git fast-import automatically and pipe the
 189     output directly into git fast-import; this should also speed up
 190     the conversion.</li>
 191
 192   <li>Only single projects can be converted at a time.  Given the way
 193     git is typically used, this is probably what you want anyway.</li>
 194
 195   <li>The cvs2svn test suite does not include meaningful tests of git
 196     output.</li>
 197
 198   <li>cvs2git makes no attempt to convert <tt>.cvsignore</tt> files
 199     into <tt>.gitignore</tt> files.</li>
 200
 201   <li>cvs2git, like cvs2svn, does not support incremental conversion
 202     (i.e., tracking a live CVS repository).  However, at least one
 203     person has documented a <a
 204     href="http://www.oak.homeunix.org/~marcel/blog/2009/06/03/tracking-cvs-with-git-using-cvs2git">possible
 205     workaround</a>.</li>
 206
 207 </ul>
 208
 209
 210 <h2><a name="docs">Documentation</a></h2>
 211
 212 <p>There is some documentation specific to cvs2git, and much of the
 213 cvs2svn documentation also applies fairly straightforwardly to
 214 cvs2git.  See the following sources:</p>
 215
 216 <ul>
 217
 218   <li>This document.</li>
 219
 220   <li>The output of <tt>cvs2git --help</tt>.</li>
 221
 222   <li>The cvs2git man page.  If the man page is not installed on your
 223     Unix-like system, you can view it by typing a command like
 224     <tt>cvs2git --man | groff -man -Tascii | less</tt>.</li>
 225
 226   <li><a href="cvs2svn.html#intro">The cvs2svn documentation</a> and
 227     <a href="faq.html">the cvs2svn FAQ</a>, which contain much general
 228     discussion and describe many features that can also be used for
 229     cvs2git.</li>
 230
 231   <li><tt>cvs2git-example.options</tt> in the cvs2svn source tree,
 232     which is an example of an options file that can be used to
 233     configure a cvs2git conversion.  The file is extensively
 234     documented.</li>
 235
 236   <li>The cvs2svn mailing lists, IRC channel, etc., as described in <a
 237     href="faq.html#gettinghelp">the cvs2svn FAQ</a>.</li>
 238
 239 </ul>
 240
 241
 242 <h2><a name="usage">Usage</a></h2>
 243
 244 <p>This section outlines the steps needed to convert a CVS repository
 245 to git using cvs2git.</p>
 246
 247 <ol>
 248
 249   <li>Be sure that you have the <a href="#reqs">requirements</a>,
 250     including either RCS or CVS (used to read revision contents from
 251     the CVS repository).</li>
 252
 253   <li>Obtain a copy of cvs2svn/cvs2git version 2.1 or newer.  It is
 254     recommended that you use the most recent version available, or
 255     even the development version.
 256
 257     <ul>
 258
 259       <li>To install cvs2svn from a <a
 260         href="http://cvs2svn.tigris.org/servlets/ProjectDocumentList">tarball</a>,
 261         simply unpack the tarball into a directory on your conversion
 262         computer; cvs2git can be run directly from this
 263         directory.</li>
 264
 265       <li>
 266
 267         <p>To check out the current trunk version of cvs2svn, make
 268           sure that you have Subversion installed and then run:</p>
 269
 270 <pre>
 271 svn co --username=guest --password="" http://cvs2svn.tigris.org/svn/cvs2svn/trunk cvs2svn-trunk
 272 cd cvs2svn-trunk
 273 make man # If you want to create manpages for the main programs
 274 make check # ...optional
 275 </pre>
 276
 277         <p>Please note that the test suite includes tests that are
 278           marked "XFAIL" (expected failure); these are known and are
 279           not considered serious problems.</p>
 280
 281       </li>
 282
 283     </ul>
 284
 285   </li>
 286
 287   <li>
 288
 289     Configure cvs2git and run the conversion.  This can be done via
 290     command-line options or via an options file:
 291
 292     <ul>
 293
 294       <li>
 295
 296         <p>The command-line options for running cvs2git are documented
 297           in the cvs2git man page and in the output of <tt>cvs2git
 298           --help</tt>.  For example:</p>
 299
 300 <pre>
 301 cvs2git \
 302     --blobfile=cvs2git-tmp/git-blob.dat \
 303     --dumpfile=cvs2git-tmp/git-dump.dat \
 304     --username=cvs2git \
 305     /path/to/cvs/repo
 306 </pre>
 307
 308       </li>
 309
 310       <li>
 311
 312         <p>The more flexible <a
 313           href="cvs2svn.html#cmd-vs-options">options-file method</a>
 314           requires you to create an options file, then start cvs2git
 315           with</p>
 316
 317 <pre>
 318 cvs2git --options=OPTIONS-FILE
 319 </pre>
 320
 321         <p>Use <tt>cvs2git-example.options</tt> in the cvs2svn source
 322           tree as your starting point; the file contains lots of
 323           documentation.</p>
 324
 325       </li>
 326
 327     </ul>
 328
 329     <p>This creates two output files
 330       in <a href="http://www.kernel.org/pub/software/scm/git/docs/git-fast-import.html">git
 331       fast-import</a> format.  The names of these files are specified
 332       by your options file or command-line arguments.  In the example,
 333       these files are named <tt>cvs2git-tmp/git-blob.dat</tt> and
 334       <tt>cvs2git-tmp/git-dump.dat</tt>.</p>
 335
 336   </li>
 337
 338   <li>
 339
 340     <p>Initialize a git repository:</p>
 341
 342 <pre>
 343 git init --bare /path/to/myproject.git
 344 cd /path/to/myproject.git
 345 </pre>
 346
 347   </li>
 348
 349   <li>
 350
 351     <p>Load the dump files into the new git repository using <tt>git
 352       fast-import</tt>:</p>
 353
 354 <pre>
 355 git fast-import --export-marks=../cvs2git-tmp/git-marks.dat &lt;../cvs2git-tmp/git-blob.dat
 356 git fast-import --import-marks=../cvs2git-tmp/git-marks.dat &lt;../cvs2git-tmp/git-dump.dat
 357 </pre>
 358
 359     <p>On Linux/Unix this can be shortened to:</p>
 360
 361 <pre>
 362 cat ../cvs2git-tmp/git-blob.dat ../cvs2git-tmp/git-dump.dat | git fast-import
 363 </pre>
 364
 365   </li>
 366
 367   <li>
 368
 369     <p>Delete the <tt>TAG.FIXUP</tt> branch:</p>
 370
 371 <pre>
 372 git branch -D TAG.FIXUP
 373 </pre>
 374
 375   </li>
 376
 377   <li>
 378
 379     <p>(Optional) View the results of the conversion, for example:</p>
 380
 381 <pre>
 382 gitk --all
 383 </pre>
 384
 385   </li>
 386
 387   <li>(Recommended) To get rid of unnecessary tag fixup branches, run
 388     the <tt>contrib/git-move-refs.py</tt> script from within the git
 389     repository.</li>
 390
 391   <li>
 392
 393     <p>(Recommended) Re-compact the repository and discard any
 394       garbage:</p>
 395
 396 <pre>
 397 git gc --prune=now
 398 </pre>
 399
 400   </li>
 401
 402   <li>
 403
 404     <p>The result of the above procedure is a <em>bare</em> git
 405       repository (one that does not have a checked-out version of the
 406       source tree).  This is the type of repository that you would put
 407       on your server.  To work on your project, make a non-bare clone
 408       (one that includes a checked-out source tree):</p>
 409
 410 <pre>
 411 cd $HOME
 412 git clone /path/to/myproject.git
 413 cd myproject
 414 </pre>
 415
 416     <p>Now you are ready to start editing files and committing to
 417       git!</p>
 418
 419   </li>
 420
 421 </ol>
 422
 423 <h2><a name="non-bare">Converting to a non-bare repository</a></h2>
 424
 425 <p>If you want to convert into a non-bare git repository (one
 426   that includes a working tree), then you need to make two changes to
 427   the above procedure:</p>
 428
 429 <ul>
 430
 431   <li><p>Omit the <tt>--bare</tt> option in step 4; i.e., type</p>
 432
 433 <pre>
 434 git init /path/to/myproject.git
 435 cd /path/to/myproject.git
 436 </pre>
 437
 438   </li>
 439
 440   <li><p>When the conversion is done, instead of cloning as described
 441     in step 10, you need to explicitly check out the "master" version
 442     of the files into your working tree:</p>
 443
 444 <pre>
 445 git checkout master
 446 </pre>
 447
 448   </li>
 449
 450 </ul>
 451
 452 <p>Feedback would be much appreciated, including reports of success
 453   using cvs2git.  Please send comments, bug reports, and patches to
 454   the <a
 455   href="http://cvs2svn.tigris.org/servlets/ProjectMailingListList">cvs2svn
 456   mailing lists</a>.</p>
 457
 458 </div>
 459 </body>
 460 </html>