6 git-cvsserver - A CVS server emulator for Git
14 export CVS_SERVER="git cvsserver"
15 'cvs' -d :ext:user@server/path/repo.git co <HEAD_name>
17 pserver (/etc/inetd.conf):
20 cvspserver stream tcp nowait nobody /usr/bin/git-cvsserver git-cvsserver pserver
25 'git-cvsserver' [<options>] [pserver|server] [<directory> ...]
30 This application is a CVS emulation layer for Git.
32 It is highly functional. However, not all methods are implemented,
33 and for those methods that are implemented,
34 not all switches are implemented.
36 Testing has been done using both the CLI CVS client, and the Eclipse CVS
37 plugin. Most functionality works fine with both of these clients.
42 All these options obviously only make sense if enforced by the server side.
43 They have been implemented to resemble the linkgit:git-daemon[1] options as
47 Prepend 'path' to requested CVSROOT
50 Don't allow recursing into subdirectories
53 Don't check for `gitcvs.enabled` in config. You also have to specify a list
54 of allowed directories (see below) if you want to use this option.
58 Print version information and exit
63 Print usage information and exit
66 You can specify a list of allowed directories. If no directories
67 are given, all are allowed. This is an additional restriction, gitcvs
68 access still needs to be enabled by the `gitcvs.enabled` config option
69 unless `--export-all` was given, too.
75 CVS clients cannot tag, branch or perform Git merges.
77 'git-cvsserver' maps Git branches to CVS modules. This is very different
78 from what most CVS users would expect since in CVS modules usually represent
79 one or more directories.
84 1. If you are going to offer CVS access via pserver, add a line in
89 cvspserver stream tcp nowait nobody git-cvsserver pserver
92 Note: Some inetd servers let you specify the name of the executable
93 independently of the value of argv[0] (i.e. the name the program assumes
94 it was executed with). In this case the correct line in /etc/inetd.conf
98 cvspserver stream tcp nowait nobody /usr/bin/git-cvsserver git-cvsserver pserver
102 Only anonymous access is provided by pserver by default. To commit you
103 will have to create pserver accounts, simply add a gitcvs.authdb
104 setting in the config file of the repositories you want the cvsserver
105 to allow writes to, for example:
110 authdb = /etc/cvsserver/passwd
113 The format of these files is username followed by the encrypted password,
118 myuser:$1$9K7FzU28$VfF6EoPYCJEYcVQwATgOP/
119 myuser:$5$.NqmNH1vwfzGpV8B$znZIcumu1tNLATgV2l6e1/mY8RzhUDHMOaVOeL1cxV3
121 You can use the 'htpasswd' facility that comes with Apache to make these
122 files, but only with the -d option (or -B if your system suports it).
124 Preferably use the system specific utility that manages password hash
125 creation in your platform (e.g. mkpasswd in Linux, encrypt in OpenBSD or
126 pwhash in NetBSD) and paste it in the right location.
128 Then provide your password via the pserver method, for example:
130 cvs -d:pserver:someuser:somepassword@server:/path/repo.git co <HEAD_name>
132 No special setup is needed for SSH access, other than having Git tools
133 in the PATH. If you have clients that do not accept the CVS_SERVER
134 environment variable, you can rename 'git-cvsserver' to `cvs`.
136 Note: Newer CVS versions (>= 1.12.11) also support specifying
137 CVS_SERVER directly in CVSROOT like
140 cvs -d ":ext;CVS_SERVER=git cvsserver:user@server/path/repo.git" co <HEAD_name>
142 This has the advantage that it will be saved in your 'CVS/Root' files and
143 you don't need to worry about always setting the correct environment
144 variable. SSH users restricted to 'git-shell' don't need to override the default
145 with CVS_SERVER (and shouldn't) as 'git-shell' understands `cvs` to mean
146 'git-cvsserver' and pretends that the other end runs the real 'cvs' better.
148 2. For each repo that you want accessible from CVS you need to edit config in
149 the repo and add the following section.
155 # optional for debugging
156 logFile=/path/to/logfile
159 Note: you need to ensure each user that is going to invoke 'git-cvsserver' has
160 write access to the log file and to the database (see
161 <<dbbackend,Database Backend>>. If you want to offer write access over
162 SSH, the users of course also need write access to the Git repository itself.
164 You also need to ensure that each repository is "bare" (without a Git index
165 file) for `cvs commit` to work. See linkgit:gitcvs-migration[7].
167 [[configaccessmethod]]
168 All configuration variables can also be overridden for a specific method of
169 access. Valid method names are "ext" (for SSH access) and "pserver". The
170 following example configuration would disable pserver access while still
171 allowing access over SSH.
180 3. If you didn't specify the CVSROOT/CVS_SERVER directly in the checkout command,
181 automatically saving it in your 'CVS/Root' files, then you need to set them
182 explicitly in your environment. CVSROOT should be set as per normal, but the
183 directory should point at the appropriate Git repo. As above, for SSH clients
184 _not_ restricted to 'git-shell', CVS_SERVER should be set to 'git-cvsserver'.
188 export CVSROOT=:ext:user@server:/var/git/project.git
189 export CVS_SERVER="git cvsserver"
192 4. For SSH clients that will make commits, make sure their server-side
193 .ssh/environment files (or .bashrc, etc., according to their specific shell)
194 export appropriate values for GIT_AUTHOR_NAME, GIT_AUTHOR_EMAIL,
195 GIT_COMMITTER_NAME, and GIT_COMMITTER_EMAIL. For SSH clients whose login
196 shell is bash, .bashrc may be a reasonable alternative.
198 5. Clients should now be able to check out the project. Use the CVS 'module'
199 name to indicate what Git 'head' you want to check out. This also sets the
200 name of your newly checked-out directory, unless you tell it otherwise with
201 `-d <dir_name>`. For example, this checks out 'master' branch to the
202 `project-master` directory:
205 cvs co -d project-master master
212 'git-cvsserver' uses one database per Git head (i.e. CVS module) to
213 store information about the repository to maintain consistent
214 CVS revision numbers. The database needs to be
215 updated (i.e. written to) after every commit.
217 If the commit is done directly by using `git` (as opposed to
218 using 'git-cvsserver') the update will need to happen on the
219 next repository access by 'git-cvsserver', independent of
220 access method and requested operation.
222 That means that even if you offer only read access (e.g. by using
223 the pserver method), 'git-cvsserver' should have write access to
224 the database to work reliably (otherwise you need to make sure
225 that the database is up to date any time 'git-cvsserver' is executed).
227 By default it uses SQLite databases in the Git directory, named
228 `gitcvs.<module_name>.sqlite`. Note that the SQLite backend creates
229 temporary files in the same directory as the database file on
230 write so it might not be enough to grant the users using
231 'git-cvsserver' write access to the database file without granting
232 them write access to the directory, too.
234 The database cannot be reliably regenerated in a
235 consistent form after the branch it is tracking has changed.
236 Example: For merged branches, 'git-cvsserver' only tracks
237 one branch of development, and after a 'git merge' an
238 incrementally updated database may track a different branch
239 than a database regenerated from scratch, causing inconsistent
240 CVS revision numbers. `git-cvsserver` has no way of knowing which
241 branch it would have picked if it had been run incrementally
242 pre-merge. So if you have to fully or partially (from old
243 backup) regenerate the database, you should be suspicious
244 of pre-existing CVS sandboxes.
246 You can configure the database backend with the following
247 configuration variables:
249 Configuring database backend
250 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
252 'git-cvsserver' uses the Perl DBI module. Please also read
253 its documentation if changing these variables, especially
254 about `DBI->connect()`.
257 Database name. The exact meaning depends on the
258 selected database driver, for SQLite this is a filename.
259 Supports variable substitution (see below). May
260 not contain semicolons (`;`).
261 Default: '%Ggitcvs.%m.sqlite'
264 Used DBI driver. You can specify any available driver
265 for this here, but it might not work. cvsserver is tested
266 with 'DBD::SQLite', reported to work with
267 'DBD::Pg', and reported *not* to work with 'DBD::mysql'.
268 Please regard this as an experimental feature. May not
269 contain colons (`:`).
273 Database user. Only useful if setting `dbDriver`, since
274 SQLite has no concept of database users. Supports variable
275 substitution (see below).
278 Database password. Only useful if setting `dbDriver`, since
279 SQLite has no concept of database passwords.
281 gitcvs.dbTableNamePrefix::
282 Database table name prefix. Supports variable substitution
283 (see below). Any non-alphabetic characters will be replaced
286 All variables can also be set per access method, see <<configaccessmethod,above>>.
288 Variable substitution
289 ^^^^^^^^^^^^^^^^^^^^^
290 In `dbDriver` and `dbUser` you can use the following variables:
295 Git directory name, where all characters except for
296 alphanumeric ones, `.`, and `-` are replaced with
297 `_` (this should make it easier to use the directory
298 name in a filename if wanted)
300 CVS module/Git head name
302 access method (one of "ext" or "pserver")
304 Name of the user running 'git-cvsserver'.
305 If no name can be determined, the
311 These variables obviate the need for command-line options in some
312 circumstances, allowing easier restricted usage through git-shell.
314 GIT_CVSSERVER_BASE_PATH takes the place of the argument to --base-path.
316 GIT_CVSSERVER_ROOT specifies a single-directory whitelist. The
317 repository must still be configured to allow access through
318 git-cvsserver, as described above.
320 When these environment variables are set, the corresponding
321 command-line arguments may not be used.
323 ECLIPSE CVS CLIENT NOTES
324 ------------------------
326 To get a checkout with the Eclipse CVS client:
328 1. Select "Create a new project -> From CVS checkout"
329 2. Create a new location. See the notes below for details on how to choose the
331 3. Browse the 'modules' available. It will give you a list of the heads in
332 the repository. You will not be able to browse the tree from there. Only
334 4. Pick `HEAD` when it asks what branch/tag to check out. Untick the
335 "launch commit wizard" to avoid committing the .project file.
337 Protocol notes: If you are using anonymous access via pserver, just select that.
338 Those using SSH access should choose the 'ext' protocol, and configure 'ext'
339 access on the Preferences->Team->CVS->ExtConnection pane. Set CVS_SERVER to
340 "`git cvsserver`". Note that password support is not good when using 'ext',
341 you will definitely want to have SSH keys setup.
343 Alternatively, you can just use the non-standard extssh protocol that Eclipse
344 offer. In that case CVS_SERVER is ignored, and you will have to replace
345 the cvs utility on the server with 'git-cvsserver' or manipulate your `.bashrc`
346 so that calling 'cvs' effectively calls 'git-cvsserver'.
348 CLIENTS KNOWN TO WORK
349 ---------------------
351 - CVS 1.12.9 on Debian
352 - CVS 1.11.17 on MacOSX (from Fink package)
353 - Eclipse 3.0, 3.1.2 on MacOSX (see Eclipse CVS Client Notes)
359 All the operations required for normal use are supported, including
360 checkout, diff, status, update, log, add, remove, commit.
362 Most CVS command arguments that read CVS tags or revision numbers
363 (typically -r) work, and also support any git refspec
364 (tag, branch, commit ID, etc).
365 However, CVS revision numbers for non-default branches are not well
366 emulated, and cvs log does not show tags or branches at
367 all. (Non-main-branch CVS revision numbers superficially resemble CVS
368 revision numbers, but they actually encode a git commit ID directly,
369 rather than represent the number of revisions since the branch point.)
371 Note that there are two ways to checkout a particular branch.
372 As described elsewhere on this page, the "module" parameter
373 of cvs checkout is interpreted as a branch name, and it becomes
374 the main branch. It remains the main branch for a given sandbox
375 even if you temporarily make another branch sticky with
376 cvs update -r. Alternatively, the -r argument can indicate
377 some other branch to actually checkout, even though the module
378 is still the "main" branch. Tradeoffs (as currently
379 implemented): Each new "module" creates a new database on disk with
380 a history for the given module, and after the database is created,
381 operations against that main branch are fast. Or alternatively,
382 -r doesn't take any extra disk space, but may be significantly slower for
383 many operations, like cvs update.
385 If you want to refer to a git refspec that has characters that are
386 not allowed by CVS, you have two options. First, it may just work
387 to supply the git refspec directly to the appropriate CVS -r argument;
388 some CVS clients don't seem to do much sanity checking of the argument.
389 Second, if that fails, you can use a special character escape mechanism
390 that only uses characters that are valid in CVS tags. A sequence
391 of 4 or 5 characters of the form (underscore (`"_"`), dash (`"-"`),
392 one or two characters, and dash (`"-"`)) can encode various characters based
393 on the one or two letters: `"s"` for slash (`"/"`), `"p"` for
394 period (`"."`), `"u"` for underscore (`"_"`), or two hexadecimal digits
395 for any byte value at all (typically an ASCII number, or perhaps a part
396 of a UTF-8 encoded character).
398 Legacy monitoring operations are not supported (edit, watch and related).
399 Exports and tagging (tags and branches) are not supported at this stage.
401 CRLF Line Ending Conversions
402 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
404 By default the server leaves the `-k` mode blank for all files,
405 which causes the CVS client to treat them as a text files, subject
406 to end-of-line conversion on some platforms.
408 You can make the server use the end-of-line conversion attributes to
409 set the `-k` modes for files by setting the `gitcvs.usecrlfattr`
410 config variable. See linkgit:gitattributes[5] for more information
411 about end-of-line conversion.
413 Alternatively, if `gitcvs.usecrlfattr` config is not enabled
414 or the attributes do not allow automatic detection for a filename, then
415 the server uses the `gitcvs.allBinary` config for the default setting.
416 If `gitcvs.allBinary` is set, then file not otherwise
417 specified will default to '-kb' mode. Otherwise the `-k` mode
418 is left blank. But if `gitcvs.allBinary` is set to "guess", then
419 the correct `-k` mode will be guessed based on the contents of
422 For best consistency with 'cvs', it is probably best to override the
423 defaults by setting `gitcvs.usecrlfattr` to true,
424 and `gitcvs.allBinary` to "guess".
428 'git-cvsserver' depends on DBD::SQLite.
432 Part of the linkgit:git[1] suite