Documentation / git-cvsserver.txton commit merge-recursive --renormalize (7610fa5)
   1git-cvsserver(1)
   2================
   3
   4NAME
   5----
   6git-cvsserver - A CVS server emulator for git
   7
   8SYNOPSIS
   9--------
  10
  11SSH:
  12
  13[verse]
  14export CVS_SERVER="git cvsserver"
  15'cvs' -d :ext:user@server/path/repo.git co <HEAD_name>
  16
  17pserver (/etc/inetd.conf):
  18
  19[verse]
  20cvspserver stream tcp nowait nobody /usr/bin/git-cvsserver git-cvsserver pserver
  21
  22Usage:
  23
  24[verse]
  25'git-cvsserver' [options] [pserver|server] [<directory> ...]
  26
  27OPTIONS
  28-------
  29
  30All these options obviously only make sense if enforced by the server side.
  31They have been implemented to resemble the linkgit:git-daemon[1] options as
  32closely as possible.
  33
  34--base-path <path>::
  35Prepend 'path' to requested CVSROOT
  36
  37--strict-paths::
  38Don't allow recursing into subdirectories
  39
  40--export-all::
  41Don't check for `gitcvs.enabled` in config. You also have to specify a list
  42of allowed directories (see below) if you want to use this option.
  43
  44-V::
  45--version::
  46Print version information and exit
  47
  48-h::
  49-H::
  50--help::
  51Print usage information and exit
  52
  53<directory>::
  54You can specify a list of allowed directories. If no directories
  55are given, all are allowed. This is an additional restriction, gitcvs
  56access still needs to be enabled by the `gitcvs.enabled` config option
  57unless '--export-all' was given, too.
  58
  59
  60DESCRIPTION
  61-----------
  62
  63This application is a CVS emulation layer for git.
  64
  65It is highly functional. However, not all methods are implemented,
  66and for those methods that are implemented,
  67not all switches are implemented.
  68
  69Testing has been done using both the CLI CVS client, and the Eclipse CVS
  70plugin. Most functionality works fine with both of these clients.
  71
  72LIMITATIONS
  73-----------
  74
  75CVS clients cannot tag, branch or perform GIT merges.
  76
  77'git-cvsserver' maps GIT branches to CVS modules. This is very different
  78from what most CVS users would expect since in CVS modules usually represent
  79one or more directories.
  80
  81INSTALLATION
  82------------
  83
  841. If you are going to offer CVS access via pserver, add a line in
  85   /etc/inetd.conf like
  86+
  87--
  88------
  89   cvspserver stream tcp nowait nobody git-cvsserver pserver
  90
  91------
  92Note: Some inetd servers let you specify the name of the executable
  93independently of the value of argv[0] (i.e. the name the program assumes
  94it was executed with). In this case the correct line in /etc/inetd.conf
  95looks like
  96
  97------
  98   cvspserver stream tcp nowait nobody /usr/bin/git-cvsserver git-cvsserver pserver
  99
 100------
 101
 102Only anonymous access is provided by pserve by default. To commit you
 103will have to create pserver accounts, simply add a gitcvs.authdb
 104setting in the config file of the repositories you want the cvsserver
 105to allow writes to, for example:
 106
 107------
 108
 109   [gitcvs]
 110        authdb = /etc/cvsserver/passwd
 111
 112------
 113The format of these files is username followed by the crypted password,
 114for example:
 115
 116------
 117   myuser:$1Oyx5r9mdGZ2
 118   myuser:$1$BA)@$vbnMJMDym7tA32AamXrm./
 119------
 120You can use the 'htpasswd' facility that comes with Apache to make these
 121files, but Apache's MD5 crypt method differs from the one used by most C
 122library's crypt() function, so don't use the -m option.
 123
 124Alternatively you can produce the password with perl's crypt() operator:
 125-----
 126   perl -e 'my ($user, $pass) = @ARGV; printf "%s:%s\n", $user, crypt($user, $pass)' $USER password
 127-----
 128
 129Then provide your password via the pserver method, for example:
 130------
 131   cvs -d:pserver:someuser:somepassword <at> server/path/repo.git co <HEAD_name>
 132------
 133No special setup is needed for SSH access, other than having GIT tools
 134in the PATH. If you have clients that do not accept the CVS_SERVER
 135environment variable, you can rename 'git-cvsserver' to `cvs`.
 136
 137Note: Newer CVS versions (>= 1.12.11) also support specifying
 138CVS_SERVER directly in CVSROOT like
 139
 140------
 141cvs -d ":ext;CVS_SERVER=git cvsserver:user@server/path/repo.git" co <HEAD_name>
 142------
 143This has the advantage that it will be saved in your 'CVS/Root' files and
 144you don't need to worry about always setting the correct environment
 145variable.  SSH users restricted to 'git-shell' don't need to override the default
 146with CVS_SERVER (and shouldn't) as 'git-shell' understands `cvs` to mean
 147'git-cvsserver' and pretends that the other end runs the real 'cvs' better.
 148--
 1492. For each repo that you want accessible from CVS you need to edit config in
 150   the repo and add the following section.
 151+
 152--
 153------
 154   [gitcvs]
 155        enabled=1
 156        # optional for debugging
 157        logfile=/path/to/logfile
 158
 159------
 160Note: you need to ensure each user that is going to invoke 'git-cvsserver' has
 161write access to the log file and to the database (see
 162<<dbbackend,Database Backend>>. If you want to offer write access over
 163SSH, the users of course also need write access to the git repository itself.
 164
 165You also need to ensure that each repository is "bare" (without a git index
 166file) for `cvs commit` to work. See linkgit:gitcvs-migration[7].
 167
 168[[configaccessmethod]]
 169All configuration variables can also be overridden for a specific method of
 170access. Valid method names are "ext" (for SSH access) and "pserver". The
 171following example configuration would disable pserver access while still
 172allowing access over SSH.
 173------
 174   [gitcvs]
 175        enabled=0
 176
 177   [gitcvs "ext"]
 178        enabled=1
 179------
 180--
 1813. If you didn't specify the CVSROOT/CVS_SERVER directly in the checkout command,
 182   automatically saving it in your 'CVS/Root' files, then you need to set them
 183   explicitly in your environment.  CVSROOT should be set as per normal, but the
 184   directory should point at the appropriate git repo.  As above, for SSH clients
 185   _not_ restricted to 'git-shell', CVS_SERVER should be set to 'git-cvsserver'.
 186+
 187--
 188------
 189     export CVSROOT=:ext:user@server:/var/git/project.git
 190     export CVS_SERVER="git cvsserver"
 191------
 192--
 1934. For SSH clients that will make commits, make sure their server-side
 194   .ssh/environment files (or .bashrc, etc., according to their specific shell)
 195   export appropriate values for GIT_AUTHOR_NAME, GIT_AUTHOR_EMAIL,
 196   GIT_COMMITTER_NAME, and GIT_COMMITTER_EMAIL.  For SSH clients whose login
 197   shell is bash, .bashrc may be a reasonable alternative.
 198
 1995. Clients should now be able to check out the project. Use the CVS 'module'
 200   name to indicate what GIT 'head' you want to check out.  This also sets the
 201   name of your newly checked-out directory, unless you tell it otherwise with
 202   `-d <dir_name>`.  For example, this checks out 'master' branch to the
 203   `project-master` directory:
 204+
 205------
 206     cvs co -d project-master master
 207------
 208
 209[[dbbackend]]
 210Database Backend
 211----------------
 212
 213'git-cvsserver' uses one database per git head (i.e. CVS module) to
 214store information about the repository to maintain consistent
 215CVS revision numbers. The database needs to be
 216updated (i.e. written to) after every commit.
 217
 218If the commit is done directly by using `git` (as opposed to
 219using 'git-cvsserver') the update will need to happen on the
 220next repository access by 'git-cvsserver', independent of
 221access method and requested operation.
 222
 223That means that even if you offer only read access (e.g. by using
 224the pserver method), 'git-cvsserver' should have write access to
 225the database to work reliably (otherwise you need to make sure
 226that the database is up-to-date any time 'git-cvsserver' is executed).
 227
 228By default it uses SQLite databases in the git directory, named
 229`gitcvs.<module_name>.sqlite`. Note that the SQLite backend creates
 230temporary files in the same directory as the database file on
 231write so it might not be enough to grant the users using
 232'git-cvsserver' write access to the database file without granting
 233them write access to the directory, too.
 234
 235The database can not be reliably regenerated in a
 236consistent form after the branch it is tracking has changed.
 237Example: For merged branches, 'git-cvsserver' only tracks
 238one branch of development, and after a 'git merge' an
 239incrementally updated database may track a different branch
 240than a database regenerated from scratch, causing inconsistent
 241CVS revision numbers. `git-cvsserver` has no way of knowing which
 242branch it would have picked if it had been run incrementally
 243pre-merge. So if you have to fully or partially (from old
 244backup) regenerate the database, you should be suspicious
 245of pre-existing CVS sandboxes.
 246
 247You can configure the database backend with the following
 248configuration variables:
 249
 250Configuring database backend
 251~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 252
 253'git-cvsserver' uses the Perl DBI module. Please also read
 254its documentation if changing these variables, especially
 255about `DBI->connect()`.
 256
 257gitcvs.dbname::
 258        Database name. The exact meaning depends on the
 259        selected database driver, for SQLite this is a filename.
 260        Supports variable substitution (see below). May
 261        not contain semicolons (`;`).
 262        Default: '%Ggitcvs.%m.sqlite'
 263
 264gitcvs.dbdriver::
 265        Used DBI driver. You can specify any available driver
 266        for this here, but it might not work. cvsserver is tested
 267        with 'DBD::SQLite', reported to work with
 268        'DBD::Pg', and reported *not* to work with 'DBD::mysql'.
 269        Please regard this as an experimental feature. May not
 270        contain colons (`:`).
 271        Default: 'SQLite'
 272
 273gitcvs.dbuser::
 274        Database user. Only useful if setting `dbdriver`, since
 275        SQLite has no concept of database users. Supports variable
 276        substitution (see below).
 277
 278gitcvs.dbpass::
 279        Database password.  Only useful if setting `dbdriver`, since
 280        SQLite has no concept of database passwords.
 281
 282gitcvs.dbTableNamePrefix::
 283        Database table name prefix.  Supports variable substitution
 284        (see below).  Any non-alphabetic characters will be replaced
 285        with underscores.
 286
 287All variables can also be set per access method, see <<configaccessmethod,above>>.
 288
 289Variable substitution
 290^^^^^^^^^^^^^^^^^^^^^
 291In `dbdriver` and `dbuser` you can use the following variables:
 292
 293%G::
 294        git directory name
 295%g::
 296        git directory name, where all characters except for
 297        alpha-numeric ones, `.`, and `-` are replaced with
 298        `_` (this should make it easier to use the directory
 299        name in a filename if wanted)
 300%m::
 301        CVS module/git head name
 302%a::
 303        access method (one of "ext" or "pserver")
 304%u::
 305        Name of the user running 'git-cvsserver'.
 306        If no name can be determined, the
 307        numeric uid is used.
 308
 309ENVIRONMENT
 310-----------
 311
 312These variables obviate the need for command-line options in some
 313circumstances, allowing easier restricted usage through git-shell.
 314
 315GIT_CVSSERVER_BASE_PATH takes the place of the argument to --base-path.
 316
 317GIT_CVSSERVER_ROOT specifies a single-directory whitelist. The
 318repository must still be configured to allow access through
 319git-cvsserver, as described above.
 320
 321When these environment variables are set, the corresponding
 322command-line arguments may not be used.
 323
 324Eclipse CVS Client Notes
 325------------------------
 326
 327To get a checkout with the Eclipse CVS client:
 328
 3291. Select "Create a new project -> From CVS checkout"
 3302. Create a new location. See the notes below for details on how to choose the
 331   right protocol.
 3323. Browse the 'modules' available. It will give you a list of the heads in
 333   the repository. You will not be able to browse the tree from there. Only
 334   the heads.
 3354. Pick 'HEAD' when it asks what branch/tag to check out. Untick the
 336   "launch commit wizard" to avoid committing the .project file.
 337
 338Protocol notes: If you are using anonymous access via pserver, just select that.
 339Those using SSH access should choose the 'ext' protocol, and configure 'ext'
 340access on the Preferences->Team->CVS->ExtConnection pane. Set CVS_SERVER to
 341"`git cvsserver`". Note that password support is not good when using 'ext',
 342you will definitely want to have SSH keys setup.
 343
 344Alternatively, you can just use the non-standard extssh protocol that Eclipse
 345offer. In that case CVS_SERVER is ignored, and you will have to replace
 346the cvs utility on the server with 'git-cvsserver' or manipulate your `.bashrc`
 347so that calling 'cvs' effectively calls 'git-cvsserver'.
 348
 349Clients known to work
 350---------------------
 351
 352- CVS 1.12.9 on Debian
 353- CVS 1.11.17 on MacOSX (from Fink package)
 354- Eclipse 3.0, 3.1.2 on MacOSX (see Eclipse CVS Client Notes)
 355- TortoiseCVS
 356
 357Operations supported
 358--------------------
 359
 360All the operations required for normal use are supported, including
 361checkout, diff, status, update, log, add, remove, commit.
 362Legacy monitoring operations are not supported (edit, watch and related).
 363Exports and tagging (tags and branches) are not supported at this stage.
 364
 365CRLF Line Ending Conversions
 366~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 367
 368By default the server leaves the '-k' mode blank for all files,
 369which causes the cvs client to treat them as a text files, subject
 370to crlf conversion on some platforms.
 371
 372You can make the server use the end-of-line conversion attributes to
 373set the '-k' modes for files by setting the `gitcvs.usecrlfattr`
 374config variable.  See linkgit:gitattributes[5] for more information
 375about end-of-line conversion.
 376
 377Alternatively, if `gitcvs.usecrlfattr` config is not enabled
 378or the attributes do not allow automatic detection for a filename, then
 379the server uses the `gitcvs.allbinary` config for the default setting.
 380If `gitcvs.allbinary` is set, then file not otherwise
 381specified will default to '-kb' mode. Otherwise the '-k' mode
 382is left blank. But if `gitcvs.allbinary` is set to "guess", then
 383the correct '-k' mode will be guessed based on the contents of
 384the file.
 385
 386For best consistency with 'cvs', it is probably best to override the
 387defaults by setting `gitcvs.usecrlfattr` to true,
 388and `gitcvs.allbinary` to "guess".
 389
 390Dependencies
 391------------
 392'git-cvsserver' depends on DBD::SQLite.
 393
 394Copyright and Authors
 395---------------------
 396
 397This program is copyright The Open University UK - 2006.
 398
 399Authors:
 400
 401- Martyn Smith    <martyn@catalyst.net.nz>
 402- Martin Langhoff <martin@catalyst.net.nz>
 403
 404with ideas and patches from participants of the git-list <git@vger.kernel.org>.
 405
 406Documentation
 407--------------
 408Documentation by Martyn Smith <martyn@catalyst.net.nz>, Martin Langhoff <martin@catalyst.net.nz>, and Matthias Urlichs <smurf@smurf.noris.de>.
 409
 410GIT
 411---
 412Part of the linkgit:git[1] suite