Support gitlinks in fast-import.
[git/git.git] / Documentation / git-fast-import.txt
1 git-fast-import(1)
2 ==================
5 ----
6 git-fast-import - Backend for fast Git data importers
10 --------
11 frontend | 'git fast-import' [options]
14 -----------
15 This program is usually not what the end user wants to run directly.
16 Most end users want to use one of the existing frontend programs,
17 which parses a specific type of foreign source and feeds the contents
18 stored there to 'git-fast-import'.
20 fast-import reads a mixed command/data stream from standard input and
21 writes one or more packfiles directly into the current repository.
22 When EOF is received on standard input, fast import writes out
23 updated branch and tag refs, fully updating the current repository
24 with the newly imported data.
26 The fast-import backend itself can import into an empty repository (one that
27 has already been initialized by 'git-init') or incrementally
28 update an existing populated repository. Whether or not incremental
29 imports are supported from a particular foreign source depends on
30 the frontend program in use.
34 -------
35 --date-format=<fmt>::
36 Specify the type of dates the frontend will supply to
37 fast-import within `author`, `committer` and `tagger` commands.
38 See ``Date Formats'' below for details about which formats
39 are supported, and their syntax.
41 --force::
42 Force updating modified existing branches, even if doing
43 so would cause commits to be lost (as the new commit does
44 not contain the old commit).
46 --max-pack-size=<n>::
47 Maximum size of each output packfile, expressed in MiB.
48 The default is 4096 (4 GiB) as that is the maximum allowed
49 packfile size (due to file format limitations). Some
50 importers may wish to lower this, such as to ensure the
51 resulting packfiles fit on CDs.
53 --depth=<n>::
54 Maximum delta depth, for blob and tree deltification.
55 Default is 10.
57 --active-branches=<n>::
58 Maximum number of branches to maintain active at once.
59 See ``Memory Utilization'' below for details. Default is 5.
61 --export-marks=<file>::
62 Dumps the internal marks table to <file> when complete.
63 Marks are written one per line as `:markid SHA-1`.
64 Frontends can use this file to validate imports after they
65 have been completed, or to save the marks table across
66 incremental runs. As <file> is only opened and truncated
67 at checkpoint (or completion) the same path can also be
68 safely given to \--import-marks.
70 --import-marks=<file>::
71 Before processing any input, load the marks specified in
72 <file>. The input file must exist, must be readable, and
73 must use the same format as produced by \--export-marks.
74 Multiple options may be supplied to import more than one
75 set of marks. If a mark is defined to different values,
76 the last file wins.
78 --export-pack-edges=<file>::
79 After creating a packfile, print a line of data to
80 <file> listing the filename of the packfile and the last
81 commit on each branch that was written to that packfile.
82 This information may be useful after importing projects
83 whose total object set exceeds the 4 GiB packfile limit,
84 as these commits can be used as edge points during calls
85 to 'git-pack-objects'.
87 --quiet::
88 Disable all non-fatal output, making fast-import silent when it
89 is successful. This option disables the output shown by
90 \--stats.
92 --stats::
93 Display some basic statistics about the objects fast-import has
94 created, the packfiles they were stored into, and the
95 memory used by fast-import during this run. Showing this output
96 is currently the default, but can be disabled with \--quiet.
99 Performance
100 -----------
101 The design of fast-import allows it to import large projects in a minimum
102 amount of memory usage and processing time. Assuming the frontend
103 is able to keep up with fast-import and feed it a constant stream of data,
104 import times for projects holding 10+ years of history and containing
105 100,000+ individual commits are generally completed in just 1-2
106 hours on quite modest (~$2,000 USD) hardware.
108 Most bottlenecks appear to be in foreign source data access (the
109 source just cannot extract revisions fast enough) or disk IO (fast-import
110 writes as fast as the disk will take the data). Imports will run
111 faster if the source data is stored on a different drive than the
112 destination Git repository (due to less IO contention).
115 Development Cost
116 ----------------
117 A typical frontend for fast-import tends to weigh in at approximately 200
118 lines of Perl/Python/Ruby code. Most developers have been able to
119 create working importers in just a couple of hours, even though it
120 is their first exposure to fast-import, and sometimes even to Git. This is
121 an ideal situation, given that most conversion tools are throw-away
122 (use once, and never look back).
125 Parallel Operation
126 ------------------
127 Like 'git-push' or 'git-fetch', imports handled by fast-import are safe to
128 run alongside parallel `git repack -a -d` or `git gc` invocations,
129 or any other Git operation (including 'git-prune', as loose objects
130 are never used by fast-import).
132 fast-import does not lock the branch or tag refs it is actively importing.
133 After the import, during its ref update phase, fast-import tests each
134 existing branch ref to verify the update will be a fast-forward
135 update (the commit stored in the ref is contained in the new
136 history of the commit to be written). If the update is not a
137 fast-forward update, fast-import will skip updating that ref and instead
138 prints a warning message. fast-import will always attempt to update all
139 branch refs, and does not stop on the first failure.
141 Branch updates can be forced with \--force, but its recommended that
142 this only be used on an otherwise quiet repository. Using \--force
143 is not necessary for an initial import into an empty repository.
146 Technical Discussion
147 --------------------
148 fast-import tracks a set of branches in memory. Any branch can be created
149 or modified at any point during the import process by sending a
150 `commit` command on the input stream. This design allows a frontend
151 program to process an unlimited number of branches simultaneously,
152 generating commits in the order they are available from the source
153 data. It also simplifies the frontend programs considerably.
155 fast-import does not use or alter the current working directory, or any
156 file within it. (It does however update the current Git repository,
157 as referenced by `GIT_DIR`.) Therefore an import frontend may use
158 the working directory for its own purposes, such as extracting file
159 revisions from the foreign source. This ignorance of the working
160 directory also allows fast-import to run very quickly, as it does not
161 need to perform any costly file update operations when switching
162 between branches.
164 Input Format
165 ------------
166 With the exception of raw file data (which Git does not interpret)
167 the fast-import input format is text (ASCII) based. This text based
168 format simplifies development and debugging of frontend programs,
169 especially when a higher level language such as Perl, Python or
170 Ruby is being used.
172 fast-import is very strict about its input. Where we say SP below we mean
173 *exactly* one space. Likewise LF means one (and only one) linefeed.
174 Supplying additional whitespace characters will cause unexpected
175 results, such as branch names or file names with leading or trailing
176 spaces in their name, or early termination of fast-import when it encounters
177 unexpected input.
179 Stream Comments
180 ~~~~~~~~~~~~~~~
181 To aid in debugging frontends fast-import ignores any line that
182 begins with `#` (ASCII pound/hash) up to and including the line
183 ending `LF`. A comment line may contain any sequence of bytes
184 that does not contain an LF and therefore may be used to include
185 any detailed debugging information that might be specific to the
186 frontend and useful when inspecting a fast-import data stream.
188 Date Formats
189 ~~~~~~~~~~~~
190 The following date formats are supported. A frontend should select
191 the format it will use for this import by passing the format name
192 in the \--date-format=<fmt> command line option.
194 `raw`::
195 This is the Git native format and is `<time> SP <offutc>`.
196 It is also fast-import's default format, if \--date-format was
197 not specified.
198 +
199 The time of the event is specified by `<time>` as the number of
200 seconds since the UNIX epoch (midnight, Jan 1, 1970, UTC) and is
201 written as an ASCII decimal integer.
202 +
203 The local offset is specified by `<offutc>` as a positive or negative
204 offset from UTC. For example EST (which is 5 hours behind UTC)
205 would be expressed in `<tz>` by ``-0500'' while UTC is ``+0000''.
206 The local offset does not affect `<time>`; it is used only as an
207 advisement to help formatting routines display the timestamp.
208 +
209 If the local offset is not available in the source material, use
210 ``+0000'', or the most common local offset. For example many
211 organizations have a CVS repository which has only ever been accessed
212 by users who are located in the same location and timezone. In this
213 case a reasonable offset from UTC could be assumed.
214 +
215 Unlike the `rfc2822` format, this format is very strict. Any
216 variation in formatting will cause fast-import to reject the value.
218 `rfc2822`::
219 This is the standard email format as described by RFC 2822.
220 +
221 An example value is ``Tue Feb 6 11:22:18 2007 -0500''. The Git
222 parser is accurate, but a little on the lenient side. It is the
223 same parser used by 'git-am' when applying patches
224 received from email.
225 +
226 Some malformed strings may be accepted as valid dates. In some of
227 these cases Git will still be able to obtain the correct date from
228 the malformed string. There are also some types of malformed
229 strings which Git will parse wrong, and yet consider valid.
230 Seriously malformed strings will be rejected.
231 +
232 Unlike the `raw` format above, the timezone/UTC offset information
233 contained in an RFC 2822 date string is used to adjust the date
234 value to UTC prior to storage. Therefore it is important that
235 this information be as accurate as possible.
236 +
237 If the source material uses RFC 2822 style dates,
238 the frontend should let fast-import handle the parsing and conversion
239 (rather than attempting to do it itself) as the Git parser has
240 been well tested in the wild.
241 +
242 Frontends should prefer the `raw` format if the source material
243 already uses UNIX-epoch format, can be coaxed to give dates in that
244 format, or its format is easily convertible to it, as there is no
245 ambiguity in parsing.
247 `now`::
248 Always use the current time and timezone. The literal
249 `now` must always be supplied for `<when>`.
250 +
251 This is a toy format. The current time and timezone of this system
252 is always copied into the identity string at the time it is being
253 created by fast-import. There is no way to specify a different time or
254 timezone.
255 +
256 This particular format is supplied as its short to implement and
257 may be useful to a process that wants to create a new commit
258 right now, without needing to use a working directory or
259 'git-update-index'.
260 +
261 If separate `author` and `committer` commands are used in a `commit`
262 the timestamps may not match, as the system clock will be polled
263 twice (once for each command). The only way to ensure that both
264 author and committer identity information has the same timestamp
265 is to omit `author` (thus copying from `committer`) or to use a
266 date format other than `now`.
268 Commands
269 ~~~~~~~~
270 fast-import accepts several commands to update the current repository
271 and control the current import process. More detailed discussion
272 (with examples) of each command follows later.
274 `commit`::
275 Creates a new branch or updates an existing branch by
276 creating a new commit and updating the branch to point at
277 the newly created commit.
279 `tag`::
280 Creates an annotated tag object from an existing commit or
281 branch. Lightweight tags are not supported by this command,
282 as they are not recommended for recording meaningful points
283 in time.
285 `reset`::
286 Reset an existing branch (or a new branch) to a specific
287 revision. This command must be used to change a branch to
288 a specific revision without making a commit on it.
290 `blob`::
291 Convert raw file data into a blob, for future use in a
292 `commit` command. This command is optional and is not
293 needed to perform an import.
295 `checkpoint`::
296 Forces fast-import to close the current packfile, generate its
297 unique SHA-1 checksum and index, and start a new packfile.
298 This command is optional and is not needed to perform
299 an import.
301 `progress`::
302 Causes fast-import to echo the entire line to its own
303 standard output. This command is optional and is not needed
304 to perform an import.
306 `commit`
307 ~~~~~~~~
308 Create or update a branch with a new commit, recording one logical
309 change to the project.
311 ....
312 'commit' SP <ref> LF
313 mark?
314 ('author' SP <name> SP LT <email> GT SP <when> LF)?
315 'committer' SP <name> SP LT <email> GT SP <when> LF
316 data
317 ('from' SP <committish> LF)?
318 ('merge' SP <committish> LF)?
319 (filemodify | filedelete | filecopy | filerename | filedeleteall)*
320 LF?
321 ....
323 where `<ref>` is the name of the branch to make the commit on.
324 Typically branch names are prefixed with `refs/heads/` in
325 Git, so importing the CVS branch symbol `RELENG-1_0` would use
326 `refs/heads/RELENG-1_0` for the value of `<ref>`. The value of
327 `<ref>` must be a valid refname in Git. As `LF` is not valid in
328 a Git refname, no quoting or escaping syntax is supported here.
330 A `mark` command may optionally appear, requesting fast-import to save a
331 reference to the newly created commit for future use by the frontend
332 (see below for format). It is very common for frontends to mark
333 every commit they create, thereby allowing future branch creation
334 from any imported commit.
336 The `data` command following `committer` must supply the commit
337 message (see below for `data` command syntax). To import an empty
338 commit message use a 0 length data. Commit messages are free-form
339 and are not interpreted by Git. Currently they must be encoded in
340 UTF-8, as fast-import does not permit other encodings to be specified.
342 Zero or more `filemodify`, `filedelete`, `filecopy`, `filerename`
343 and `filedeleteall` commands
344 may be included to update the contents of the branch prior to
345 creating the commit. These commands may be supplied in any order.
346 However it is recommended that a `filedeleteall` command precede
347 all `filemodify`, `filecopy` and `filerename` commands in the same
348 commit, as `filedeleteall`
349 wipes the branch clean (see below).
351 The `LF` after the command is optional (it used to be required).
353 `author`
354 ^^^^^^^^
355 An `author` command may optionally appear, if the author information
356 might differ from the committer information. If `author` is omitted
357 then fast-import will automatically use the committer's information for
358 the author portion of the commit. See below for a description of
359 the fields in `author`, as they are identical to `committer`.
361 `committer`
362 ^^^^^^^^^^^
363 The `committer` command indicates who made this commit, and when
364 they made it.
366 Here `<name>` is the person's display name (for example
367 ``Com M Itter'') and `<email>` is the person's email address
368 (``''). `LT` and `GT` are the literal less-than (\x3c)
369 and greater-than (\x3e) symbols. These are required to delimit
370 the email address from the other fields in the line. Note that
371 `<name>` is free-form and may contain any sequence of bytes, except
372 `LT` and `LF`. It is typically UTF-8 encoded.
374 The time of the change is specified by `<when>` using the date format
375 that was selected by the \--date-format=<fmt> command line option.
376 See ``Date Formats'' above for the set of supported formats, and
377 their syntax.
379 `from`
380 ^^^^^^
381 The `from` command is used to specify the commit to initialize
382 this branch from. This revision will be the first ancestor of the
383 new commit.
385 Omitting the `from` command in the first commit of a new branch
386 will cause fast-import to create that commit with no ancestor. This
387 tends to be desired only for the initial commit of a project.
388 If the frontend creates all files from scratch when making a new
389 branch, a `merge` command may be used instead of `from` to start
390 the commit with an empty tree.
391 Omitting the `from` command on existing branches is usually desired,
392 as the current commit on that branch is automatically assumed to
393 be the first ancestor of the new commit.
395 As `LF` is not valid in a Git refname or SHA-1 expression, no
396 quoting or escaping syntax is supported within `<committish>`.
398 Here `<committish>` is any of the following:
400 * The name of an existing branch already in fast-import's internal branch
401 table. If fast-import doesn't know the name, its treated as a SHA-1
402 expression.
404 * A mark reference, `:<idnum>`, where `<idnum>` is the mark number.
405 +
406 The reason fast-import uses `:` to denote a mark reference is this character
407 is not legal in a Git branch name. The leading `:` makes it easy
408 to distinguish between the mark 42 (`:42`) and the branch 42 (`42`
409 or `refs/heads/42`), or an abbreviated SHA-1 which happened to
410 consist only of base-10 digits.
411 +
412 Marks must be declared (via `mark`) before they can be used.
414 * A complete 40 byte or abbreviated commit SHA-1 in hex.
416 * Any valid Git SHA-1 expression that resolves to a commit. See
417 ``SPECIFYING REVISIONS'' in linkgit:git-rev-parse[1] for details.
419 The special case of restarting an incremental import from the
420 current branch value should be written as:
421 ----
422 from refs/heads/branch^0
423 ----
424 The `{caret}0` suffix is necessary as fast-import does not permit a branch to
425 start from itself, and the branch is created in memory before the
426 `from` command is even read from the input. Adding `{caret}0` will force
427 fast-import to resolve the commit through Git's revision parsing library,
428 rather than its internal branch table, thereby loading in the
429 existing value of the branch.
431 `merge`
432 ^^^^^^^
433 Includes one additional ancestor commit. If the `from` command is
434 omitted when creating a new branch, the first `merge` commit will be
435 the first ancestor of the current commit, and the branch will start
436 out with no files. An unlimited number of `merge` commands per
437 commit are permitted by fast-import, thereby establishing an n-way merge.
438 However Git's other tools never create commits with more than 15
439 additional ancestors (forming a 16-way merge). For this reason
440 it is suggested that frontends do not use more than 15 `merge`
441 commands per commit; 16, if starting a new, empty branch.
443 Here `<committish>` is any of the commit specification expressions
444 also accepted by `from` (see above).
446 `filemodify`
447 ^^^^^^^^^^^^
448 Included in a `commit` command to add a new file or change the
449 content of an existing file. This command has two different means
450 of specifying the content of the file.
452 External data format::
453 The data content for the file was already supplied by a prior
454 `blob` command. The frontend just needs to connect it.
455 +
456 ....
457 'M' SP <mode> SP <dataref> SP <path> LF
458 ....
459 +
460 Here `<dataref>` can be either a mark reference (`:<idnum>`)
461 set by a prior `blob` command, or a full 40-byte SHA-1 of an
462 existing Git blob object.
464 Inline data format::
465 The data content for the file has not been supplied yet.
466 The frontend wants to supply it as part of this modify
467 command.
468 +
469 ....
470 'M' SP <mode> SP 'inline' SP <path> LF
471 data
472 ....
473 +
474 See below for a detailed description of the `data` command.
476 In both formats `<mode>` is the type of file entry, specified
477 in octal. Git only supports the following modes:
479 * `100644` or `644`: A normal (not-executable) file. The majority
480 of files in most projects use this mode. If in doubt, this is
481 what you want.
482 * `100755` or `755`: A normal, but executable, file.
483 * `120000`: A symlink, the content of the file will be the link target.
484 * `160000`: A gitlink, SHA-1 of the object refers to a commit in
485 another repository. Git links can only be specified by SHA or through
486 a commit mark. They are used to implement submodules.
488 In both formats `<path>` is the complete path of the file to be added
489 (if not already existing) or modified (if already existing).
491 A `<path>` string must use UNIX-style directory separators (forward
492 slash `/`), may contain any byte other than `LF`, and must not
493 start with double quote (`"`).
495 If an `LF` or double quote must be encoded into `<path>` shell-style
496 quoting should be used, e.g. `"path/with\n and \" in it"`.
498 The value of `<path>` must be in canonical form. That is it must not:
500 * contain an empty directory component (e.g. `foo//bar` is invalid),
501 * end with a directory separator (e.g. `foo/` is invalid),
502 * start with a directory separator (e.g. `/foo` is invalid),
503 * contain the special component `.` or `..` (e.g. `foo/./bar` and
504 `foo/../bar` are invalid).
506 It is recommended that `<path>` always be encoded using UTF-8.
508 `filedelete`
509 ^^^^^^^^^^^^
510 Included in a `commit` command to remove a file or recursively
511 delete an entire directory from the branch. If the file or directory
512 removal makes its parent directory empty, the parent directory will
513 be automatically removed too. This cascades up the tree until the
514 first non-empty directory or the root is reached.
516 ....
517 'D' SP <path> LF
518 ....
520 here `<path>` is the complete path of the file or subdirectory to
521 be removed from the branch.
522 See `filemodify` above for a detailed description of `<path>`.
524 `filecopy`
525 ^^^^^^^^^^^^
526 Recursively copies an existing file or subdirectory to a different
527 location within the branch. The existing file or directory must
528 exist. If the destination exists it will be completely replaced
529 by the content copied from the source.
531 ....
532 'C' SP <path> SP <path> LF
533 ....
535 here the first `<path>` is the source location and the second
536 `<path>` is the destination. See `filemodify` above for a detailed
537 description of what `<path>` may look like. To use a source path
538 that contains SP the path must be quoted.
540 A `filecopy` command takes effect immediately. Once the source
541 location has been copied to the destination any future commands
542 applied to the source location will not impact the destination of
543 the copy.
545 `filerename`
546 ^^^^^^^^^^^^
547 Renames an existing file or subdirectory to a different location
548 within the branch. The existing file or directory must exist. If
549 the destination exists it will be replaced by the source directory.
551 ....
552 'R' SP <path> SP <path> LF
553 ....
555 here the first `<path>` is the source location and the second
556 `<path>` is the destination. See `filemodify` above for a detailed
557 description of what `<path>` may look like. To use a source path
558 that contains SP the path must be quoted.
560 A `filerename` command takes effect immediately. Once the source
561 location has been renamed to the destination any future commands
562 applied to the source location will create new files there and not
563 impact the destination of the rename.
565 Note that a `filerename` is the same as a `filecopy` followed by a
566 `filedelete` of the source location. There is a slight performance
567 advantage to using `filerename`, but the advantage is so small
568 that it is never worth trying to convert a delete/add pair in
569 source material into a rename for fast-import. This `filerename`
570 command is provided just to simplify frontends that already have
571 rename information and don't want bother with decomposing it into a
572 `filecopy` followed by a `filedelete`.
574 `filedeleteall`
575 ^^^^^^^^^^^^^^^
576 Included in a `commit` command to remove all files (and also all
577 directories) from the branch. This command resets the internal
578 branch structure to have no files in it, allowing the frontend
579 to subsequently add all interesting files from scratch.
581 ....
582 'deleteall' LF
583 ....
585 This command is extremely useful if the frontend does not know
586 (or does not care to know) what files are currently on the branch,
587 and therefore cannot generate the proper `filedelete` commands to
588 update the content.
590 Issuing a `filedeleteall` followed by the needed `filemodify`
591 commands to set the correct content will produce the same results
592 as sending only the needed `filemodify` and `filedelete` commands.
593 The `filedeleteall` approach may however require fast-import to use slightly
594 more memory per active branch (less than 1 MiB for even most large
595 projects); so frontends that can easily obtain only the affected
596 paths for a commit are encouraged to do so.
598 `mark`
599 ~~~~~~
600 Arranges for fast-import to save a reference to the current object, allowing
601 the frontend to recall this object at a future point in time, without
602 knowing its SHA-1. Here the current object is the object creation
603 command the `mark` command appears within. This can be `commit`,
604 `tag`, and `blob`, but `commit` is the most common usage.
606 ....
607 'mark' SP ':' <idnum> LF
608 ....
610 where `<idnum>` is the number assigned by the frontend to this mark.
611 The value of `<idnum>` is expressed as an ASCII decimal integer.
612 The value 0 is reserved and cannot be used as
613 a mark. Only values greater than or equal to 1 may be used as marks.
615 New marks are created automatically. Existing marks can be moved
616 to another object simply by reusing the same `<idnum>` in another
617 `mark` command.
619 `tag`
620 ~~~~~
621 Creates an annotated tag referring to a specific commit. To create
622 lightweight (non-annotated) tags see the `reset` command below.
624 ....
625 'tag' SP <name> LF
626 'from' SP <committish> LF
627 'tagger' SP <name> SP LT <email> GT SP <when> LF
628 data
629 ....
631 where `<name>` is the name of the tag to create.
633 Tag names are automatically prefixed with `refs/tags/` when stored
634 in Git, so importing the CVS branch symbol `RELENG-1_0-FINAL` would
635 use just `RELENG-1_0-FINAL` for `<name>`, and fast-import will write the
636 corresponding ref as `refs/tags/RELENG-1_0-FINAL`.
638 The value of `<name>` must be a valid refname in Git and therefore
639 may contain forward slashes. As `LF` is not valid in a Git refname,
640 no quoting or escaping syntax is supported here.
642 The `from` command is the same as in the `commit` command; see
643 above for details.
645 The `tagger` command uses the same format as `committer` within
646 `commit`; again see above for details.
648 The `data` command following `tagger` must supply the annotated tag
649 message (see below for `data` command syntax). To import an empty
650 tag message use a 0 length data. Tag messages are free-form and are
651 not interpreted by Git. Currently they must be encoded in UTF-8,
652 as fast-import does not permit other encodings to be specified.
654 Signing annotated tags during import from within fast-import is not
655 supported. Trying to include your own PGP/GPG signature is not
656 recommended, as the frontend does not (easily) have access to the
657 complete set of bytes which normally goes into such a signature.
658 If signing is required, create lightweight tags from within fast-import with
659 `reset`, then create the annotated versions of those tags offline
660 with the standard 'git-tag' process.
662 `reset`
663 ~~~~~~~
664 Creates (or recreates) the named branch, optionally starting from
665 a specific revision. The reset command allows a frontend to issue
666 a new `from` command for an existing branch, or to create a new
667 branch from an existing commit without creating a new commit.
669 ....
670 'reset' SP <ref> LF
671 ('from' SP <committish> LF)?
672 LF?
673 ....
675 For a detailed description of `<ref>` and `<committish>` see above
676 under `commit` and `from`.
678 The `LF` after the command is optional (it used to be required).
680 The `reset` command can also be used to create lightweight
681 (non-annotated) tags. For example:
683 ====
684 reset refs/tags/938
685 from :938
686 ====
688 would create the lightweight tag `refs/tags/938` referring to
689 whatever commit mark `:938` references.
691 `blob`
692 ~~~~~~
693 Requests writing one file revision to the packfile. The revision
694 is not connected to any commit; this connection must be formed in
695 a subsequent `commit` command by referencing the blob through an
696 assigned mark.
698 ....
699 'blob' LF
700 mark?
701 data
702 ....
704 The mark command is optional here as some frontends have chosen
705 to generate the Git SHA-1 for the blob on their own, and feed that
706 directly to `commit`. This is typically more work than its worth
707 however, as marks are inexpensive to store and easy to use.
709 `data`
710 ~~~~~~
711 Supplies raw data (for use as blob/file content, commit messages, or
712 annotated tag messages) to fast-import. Data can be supplied using an exact
713 byte count or delimited with a terminating line. Real frontends
714 intended for production-quality conversions should always use the
715 exact byte count format, as it is more robust and performs better.
716 The delimited format is intended primarily for testing fast-import.
718 Comment lines appearing within the `<raw>` part of `data` commands
719 are always taken to be part of the body of the data and are therefore
720 never ignored by fast-import. This makes it safe to import any
721 file/message content whose lines might start with `#`.
723 Exact byte count format::
724 The frontend must specify the number of bytes of data.
725 +
726 ....
727 'data' SP <count> LF
728 <raw> LF?
729 ....
730 +
731 where `<count>` is the exact number of bytes appearing within
732 `<raw>`. The value of `<count>` is expressed as an ASCII decimal
733 integer. The `LF` on either side of `<raw>` is not
734 included in `<count>` and will not be included in the imported data.
735 +
736 The `LF` after `<raw>` is optional (it used to be required) but
737 recommended. Always including it makes debugging a fast-import
738 stream easier as the next command always starts in column 0
739 of the next line, even if `<raw>` did not end with an `LF`.
741 Delimited format::
742 A delimiter string is used to mark the end of the data.
743 fast-import will compute the length by searching for the delimiter.
744 This format is primarily useful for testing and is not
745 recommended for real data.
746 +
747 ....
748 'data' SP '<<' <delim> LF
749 <raw> LF
750 <delim> LF
751 LF?
752 ....
753 +
754 where `<delim>` is the chosen delimiter string. The string `<delim>`
755 must not appear on a line by itself within `<raw>`, as otherwise
756 fast-import will think the data ends earlier than it really does. The `LF`
757 immediately trailing `<raw>` is part of `<raw>`. This is one of
758 the limitations of the delimited format, it is impossible to supply
759 a data chunk which does not have an LF as its last byte.
760 +
761 The `LF` after `<delim> LF` is optional (it used to be required).
763 `checkpoint`
764 ~~~~~~~~~~~~
765 Forces fast-import to close the current packfile, start a new one, and to
766 save out all current branch refs, tags and marks.
768 ....
769 'checkpoint' LF
770 LF?
771 ....
773 Note that fast-import automatically switches packfiles when the current
774 packfile reaches \--max-pack-size, or 4 GiB, whichever limit is
775 smaller. During an automatic packfile switch fast-import does not update
776 the branch refs, tags or marks.
778 As a `checkpoint` can require a significant amount of CPU time and
779 disk IO (to compute the overall pack SHA-1 checksum, generate the
780 corresponding index file, and update the refs) it can easily take
781 several minutes for a single `checkpoint` command to complete.
783 Frontends may choose to issue checkpoints during extremely large
784 and long running imports, or when they need to allow another Git
785 process access to a branch. However given that a 30 GiB Subversion
786 repository can be loaded into Git through fast-import in about 3 hours,
787 explicit checkpointing may not be necessary.
789 The `LF` after the command is optional (it used to be required).
791 `progress`
792 ~~~~~~~~~~
793 Causes fast-import to print the entire `progress` line unmodified to
794 its standard output channel (file descriptor 1) when the command is
795 processed from the input stream. The command otherwise has no impact
796 on the current import, or on any of fast-import's internal state.
798 ....
799 'progress' SP <any> LF
800 LF?
801 ....
803 The `<any>` part of the command may contain any sequence of bytes
804 that does not contain `LF`. The `LF` after the command is optional.
805 Callers may wish to process the output through a tool such as sed to
806 remove the leading part of the line, for example:
808 ====
809 frontend | git fast-import | sed 's/^progress //'
810 ====
812 Placing a `progress` command immediately after a `checkpoint` will
813 inform the reader when the `checkpoint` has been completed and it
814 can safely access the refs that fast-import updated.
816 Crash Reports
817 -------------
818 If fast-import is supplied invalid input it will terminate with a
819 non-zero exit status and create a crash report in the top level of
820 the Git repository it was importing into. Crash reports contain
821 a snapshot of the internal fast-import state as well as the most
822 recent commands that lead up to the crash.
824 All recent commands (including stream comments, file changes and
825 progress commands) are shown in the command history within the crash
826 report, but raw file data and commit messages are excluded from the
827 crash report. This exclusion saves space within the report file
828 and reduces the amount of buffering that fast-import must perform
829 during execution.
831 After writing a crash report fast-import will close the current
832 packfile and export the marks table. This allows the frontend
833 developer to inspect the repository state and resume the import from
834 the point where it crashed. The modified branches and tags are not
835 updated during a crash, as the import did not complete successfully.
836 Branch and tag information can be found in the crash report and
837 must be applied manually if the update is needed.
839 An example crash:
841 ====
842 $ cat >in <<END_OF_INPUT
843 # my very first test commit
844 commit refs/heads/master
845 committer Shawn O. Pearce <spearce> 19283 -0400
846 # who is that guy anyway?
847 data <<EOF
848 this is my commit
849 EOF
850 M 644 inline .gitignore
851 data <<EOF
852 .gitignore
853 EOF
854 M 777 inline bob
857 $ git fast-import <in
858 fatal: Corrupt mode: M 777 inline bob
859 fast-import: dumping crash report to .git/fast_import_crash_8434
861 $ cat .git/fast_import_crash_8434
862 fast-import crash report:
863 fast-import process: 8434
864 parent process : 1391
865 at Sat Sep 1 00:58:12 2007
867 fatal: Corrupt mode: M 777 inline bob
869 Most Recent Commands Before Crash
870 ---------------------------------
871 # my very first test commit
872 commit refs/heads/master
873 committer Shawn O. Pearce <spearce> 19283 -0400
874 # who is that guy anyway?
875 data <<EOF
876 M 644 inline .gitignore
877 data <<EOF
878 * M 777 inline bob
880 Active Branch LRU
881 -----------------
882 active_branches = 1 cur, 5 max
884 pos clock name
885 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
886 1) 0 refs/heads/master
888 Inactive Branches
889 -----------------
890 refs/heads/master:
891 status : active loaded dirty
892 tip commit : 0000000000000000000000000000000000000000
893 old tree : 0000000000000000000000000000000000000000
894 cur tree : 0000000000000000000000000000000000000000
895 commit clock: 0
896 last pack :
899 -------------------
901 ====
903 Tips and Tricks
904 ---------------
905 The following tips and tricks have been collected from various
906 users of fast-import, and are offered here as suggestions.
908 Use One Mark Per Commit
909 ~~~~~~~~~~~~~~~~~~~~~~~
910 When doing a repository conversion, use a unique mark per commit
911 (`mark :<n>`) and supply the \--export-marks option on the command
912 line. fast-import will dump a file which lists every mark and the Git
913 object SHA-1 that corresponds to it. If the frontend can tie
914 the marks back to the source repository, it is easy to verify the
915 accuracy and completeness of the import by comparing each Git
916 commit to the corresponding source revision.
918 Coming from a system such as Perforce or Subversion this should be
919 quite simple, as the fast-import mark can also be the Perforce changeset
920 number or the Subversion revision number.
922 Freely Skip Around Branches
923 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
924 Don't bother trying to optimize the frontend to stick to one branch
925 at a time during an import. Although doing so might be slightly
926 faster for fast-import, it tends to increase the complexity of the frontend
927 code considerably.
929 The branch LRU builtin to fast-import tends to behave very well, and the
930 cost of activating an inactive branch is so low that bouncing around
931 between branches has virtually no impact on import performance.
933 Handling Renames
934 ~~~~~~~~~~~~~~~~
935 When importing a renamed file or directory, simply delete the old
936 name(s) and modify the new name(s) during the corresponding commit.
937 Git performs rename detection after-the-fact, rather than explicitly
938 during a commit.
940 Use Tag Fixup Branches
941 ~~~~~~~~~~~~~~~~~~~~~~
942 Some other SCM systems let the user create a tag from multiple
943 files which are not from the same commit/changeset. Or to create
944 tags which are a subset of the files available in the repository.
946 Importing these tags as-is in Git is impossible without making at
947 least one commit which ``fixes up'' the files to match the content
948 of the tag. Use fast-import's `reset` command to reset a dummy branch
949 outside of your normal branch space to the base commit for the tag,
950 then commit one or more file fixup commits, and finally tag the
951 dummy branch.
953 For example since all normal branches are stored under `refs/heads/`
954 name the tag fixup branch `TAG_FIXUP`. This way it is impossible for
955 the fixup branch used by the importer to have namespace conflicts
956 with real branches imported from the source (the name `TAG_FIXUP`
957 is not `refs/heads/TAG_FIXUP`).
959 When committing fixups, consider using `merge` to connect the
960 commit(s) which are supplying file revisions to the fixup branch.
961 Doing so will allow tools such as 'git-blame' to track
962 through the real commit history and properly annotate the source
963 files.
965 After fast-import terminates the frontend will need to do `rm .git/TAG_FIXUP`
966 to remove the dummy branch.
968 Import Now, Repack Later
969 ~~~~~~~~~~~~~~~~~~~~~~~~
970 As soon as fast-import completes the Git repository is completely valid
971 and ready for use. Typically this takes only a very short time,
972 even for considerably large projects (100,000+ commits).
974 However repacking the repository is necessary to improve data
975 locality and access performance. It can also take hours on extremely
976 large projects (especially if -f and a large \--window parameter is
977 used). Since repacking is safe to run alongside readers and writers,
978 run the repack in the background and let it finish when it finishes.
979 There is no reason to wait to explore your new Git project!
981 If you choose to wait for the repack, don't try to run benchmarks
982 or performance tests until repacking is completed. fast-import outputs
983 suboptimal packfiles that are simply never seen in real use
984 situations.
986 Repacking Historical Data
987 ~~~~~~~~~~~~~~~~~~~~~~~~~
988 If you are repacking very old imported data (e.g. older than the
989 last year), consider expending some extra CPU time and supplying
990 \--window=50 (or higher) when you run 'git-repack'.
991 This will take longer, but will also produce a smaller packfile.
992 You only need to expend the effort once, and everyone using your
993 project will benefit from the smaller repository.
995 Include Some Progress Messages
996 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
997 Every once in a while have your frontend emit a `progress` message
998 to fast-import. The contents of the messages are entirely free-form,
999 so one suggestion would be to output the current month and year
1000 each time the current commit date moves into the next month.
1001 Your users will feel better knowing how much of the data stream
1002 has been processed.
1005 Packfile Optimization
1006 ---------------------
1007 When packing a blob fast-import always attempts to deltify against the last
1008 blob written. Unless specifically arranged for by the frontend,
1009 this will probably not be a prior version of the same file, so the
1010 generated delta will not be the smallest possible. The resulting
1011 packfile will be compressed, but will not be optimal.
1013 Frontends which have efficient access to all revisions of a
1014 single file (for example reading an RCS/CVS ,v file) can choose
1015 to supply all revisions of that file as a sequence of consecutive
1016 `blob` commands. This allows fast-import to deltify the different file
1017 revisions against each other, saving space in the final packfile.
1018 Marks can be used to later identify individual file revisions during
1019 a sequence of `commit` commands.
1021 The packfile(s) created by fast-import do not encourage good disk access
1022 patterns. This is caused by fast-import writing the data in the order
1023 it is received on standard input, while Git typically organizes
1024 data within packfiles to make the most recent (current tip) data
1025 appear before historical data. Git also clusters commits together,
1026 speeding up revision traversal through better cache locality.
1028 For this reason it is strongly recommended that users repack the
1029 repository with `git repack -a -d` after fast-import completes, allowing
1030 Git to reorganize the packfiles for faster data access. If blob
1031 deltas are suboptimal (see above) then also adding the `-f` option
1032 to force recomputation of all deltas can significantly reduce the
1033 final packfile size (30-50% smaller can be quite typical).
1036 Memory Utilization
1037 ------------------
1038 There are a number of factors which affect how much memory fast-import
1039 requires to perform an import. Like critical sections of core
1040 Git, fast-import uses its own memory allocators to amortize any overheads
1041 associated with malloc. In practice fast-import tends to amortize any
1042 malloc overheads to 0, due to its use of large block allocations.
1044 per object
1045 ~~~~~~~~~~
1046 fast-import maintains an in-memory structure for every object written in
1047 this execution. On a 32 bit system the structure is 32 bytes,
1048 on a 64 bit system the structure is 40 bytes (due to the larger
1049 pointer sizes). Objects in the table are not deallocated until
1050 fast-import terminates. Importing 2 million objects on a 32 bit system
1051 will require approximately 64 MiB of memory.
1053 The object table is actually a hashtable keyed on the object name
1054 (the unique SHA-1). This storage configuration allows fast-import to reuse
1055 an existing or already written object and avoid writing duplicates
1056 to the output packfile. Duplicate blobs are surprisingly common
1057 in an import, typically due to branch merges in the source.
1059 per mark
1060 ~~~~~~~~
1061 Marks are stored in a sparse array, using 1 pointer (4 bytes or 8
1062 bytes, depending on pointer size) per mark. Although the array
1063 is sparse, frontends are still strongly encouraged to use marks
1064 between 1 and n, where n is the total number of marks required for
1065 this import.
1067 per branch
1068 ~~~~~~~~~~
1069 Branches are classified as active and inactive. The memory usage
1070 of the two classes is significantly different.
1072 Inactive branches are stored in a structure which uses 96 or 120
1073 bytes (32 bit or 64 bit systems, respectively), plus the length of
1074 the branch name (typically under 200 bytes), per branch. fast-import will
1075 easily handle as many as 10,000 inactive branches in under 2 MiB
1076 of memory.
1078 Active branches have the same overhead as inactive branches, but
1079 also contain copies of every tree that has been recently modified on
1080 that branch. If subtree `include` has not been modified since the
1081 branch became active, its contents will not be loaded into memory,
1082 but if subtree `src` has been modified by a commit since the branch
1083 became active, then its contents will be loaded in memory.
1085 As active branches store metadata about the files contained on that
1086 branch, their in-memory storage size can grow to a considerable size
1087 (see below).
1089 fast-import automatically moves active branches to inactive status based on
1090 a simple least-recently-used algorithm. The LRU chain is updated on
1091 each `commit` command. The maximum number of active branches can be
1092 increased or decreased on the command line with \--active-branches=.
1094 per active tree
1095 ~~~~~~~~~~~~~~~
1096 Trees (aka directories) use just 12 bytes of memory on top of the
1097 memory required for their entries (see ``per active file'' below).
1098 The cost of a tree is virtually 0, as its overhead amortizes out
1099 over the individual file entries.
1101 per active file entry
1102 ~~~~~~~~~~~~~~~~~~~~~
1103 Files (and pointers to subtrees) within active trees require 52 or 64
1104 bytes (32/64 bit platforms) per entry. To conserve space, file and
1105 tree names are pooled in a common string table, allowing the filename
1106 ``Makefile'' to use just 16 bytes (after including the string header
1107 overhead) no matter how many times it occurs within the project.
1109 The active branch LRU, when coupled with the filename string pool
1110 and lazy loading of subtrees, allows fast-import to efficiently import
1111 projects with 2,000+ branches and 45,114+ files in a very limited
1112 memory footprint (less than 2.7 MiB per active branch).
1115 Author
1116 ------
1117 Written by Shawn O. Pearce <>.
1119 Documentation
1120 --------------
1121 Documentation by Shawn O. Pearce <>.
1123 GIT
1124 ---
1125 Part of the linkgit:git[1] suite