6 git-fast-import - Backend for fast Git data importers
12 frontend | 'git fast-import' [<options>]
16 This program is usually not what the end user wants to run directly.
17 Most end users want to use one of the existing frontend programs,
18 which parses a specific type of foreign source and feeds the contents
19 stored there to 'git fast-import'.
21 fast-import reads a mixed command/data stream from standard input and
22 writes one or more packfiles directly into the current repository.
23 When EOF is received on standard input, fast import writes out
24 updated branch and tag refs, fully updating the current repository
25 with the newly imported data.
27 The fast-import backend itself can import into an empty repository (one that
28 has already been initialized by 'git init') or incrementally
29 update an existing populated repository. Whether or not incremental
30 imports are supported from a particular foreign source depends on
31 the frontend program in use.
38 Force updating modified existing branches, even if doing
39 so would cause commits to be lost (as the new commit does
40 not contain the old commit).
43 Disable the output shown by --stats, making fast-import usually
44 be silent when it is successful. However, if the import stream
45 has directives intended to show user output (e.g. `progress`
46 directives), the corresponding messages will still be shown.
49 Display some basic statistics about the objects fast-import has
50 created, the packfiles they were stored into, and the
51 memory used by fast-import during this run. Showing this output
52 is currently the default, but can be disabled with --quiet.
58 Write responses to `get-mark`, `cat-blob`, and `ls` queries to the
59 file descriptor <fd> instead of `stdout`. Allows `progress`
60 output intended for the end-user to be separated from other
64 Specify the type of dates the frontend will supply to
65 fast-import within `author`, `committer` and `tagger` commands.
66 See ``Date Formats'' below for details about which formats
67 are supported, and their syntax.
70 Terminate with error if there is no `done` command at the end of
71 the stream. This option might be useful for detecting errors
72 that cause the frontend to terminate before it has started to
75 Locations of Marks Files
76 ~~~~~~~~~~~~~~~~~~~~~~~~
78 --export-marks=<file>::
79 Dumps the internal marks table to <file> when complete.
80 Marks are written one per line as `:markid SHA-1`.
81 Frontends can use this file to validate imports after they
82 have been completed, or to save the marks table across
83 incremental runs. As <file> is only opened and truncated
84 at checkpoint (or completion) the same path can also be
85 safely given to --import-marks.
87 --import-marks=<file>::
88 Before processing any input, load the marks specified in
89 <file>. The input file must exist, must be readable, and
90 must use the same format as produced by --export-marks.
91 Multiple options may be supplied to import more than one
92 set of marks. If a mark is defined to different values,
95 --import-marks-if-exists=<file>::
96 Like --import-marks but instead of erroring out, silently
97 skips the file if it does not exist.
99 --[no-]relative-marks::
100 After specifying --relative-marks the paths specified
101 with --import-marks= and --export-marks= are relative
102 to an internal directory in the current repository.
103 In git-fast-import this means that the paths are relative
104 to the .git/info/fast-import directory. However, other
105 importers may use a different location.
107 Relative and non-relative marks may be combined by interweaving
108 --(no-)-relative-marks with the --(import|export)-marks= options.
110 Performance and Compression Tuning
111 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
113 --active-branches=<n>::
114 Maximum number of branches to maintain active at once.
115 See ``Memory Utilization'' below for details. Default is 5.
117 --big-file-threshold=<n>::
118 Maximum size of a blob that fast-import will attempt to
119 create a delta for, expressed in bytes. The default is 512m
120 (512 MiB). Some importers may wish to lower this on systems
121 with constrained memory.
124 Maximum delta depth, for blob and tree deltification.
127 --export-pack-edges=<file>::
128 After creating a packfile, print a line of data to
129 <file> listing the filename of the packfile and the last
130 commit on each branch that was written to that packfile.
131 This information may be useful after importing projects
132 whose total object set exceeds the 4 GiB packfile limit,
133 as these commits can be used as edge points during calls
134 to 'git pack-objects'.
136 --max-pack-size=<n>::
137 Maximum size of each output packfile.
138 The default is unlimited.
140 fastimport.unpackLimit::
141 See linkgit:git-config[1]
145 The design of fast-import allows it to import large projects in a minimum
146 amount of memory usage and processing time. Assuming the frontend
147 is able to keep up with fast-import and feed it a constant stream of data,
148 import times for projects holding 10+ years of history and containing
149 100,000+ individual commits are generally completed in just 1-2
150 hours on quite modest (~$2,000 USD) hardware.
152 Most bottlenecks appear to be in foreign source data access (the
153 source just cannot extract revisions fast enough) or disk IO (fast-import
154 writes as fast as the disk will take the data). Imports will run
155 faster if the source data is stored on a different drive than the
156 destination Git repository (due to less IO contention).
161 A typical frontend for fast-import tends to weigh in at approximately 200
162 lines of Perl/Python/Ruby code. Most developers have been able to
163 create working importers in just a couple of hours, even though it
164 is their first exposure to fast-import, and sometimes even to Git. This is
165 an ideal situation, given that most conversion tools are throw-away
166 (use once, and never look back).
171 Like 'git push' or 'git fetch', imports handled by fast-import are safe to
172 run alongside parallel `git repack -a -d` or `git gc` invocations,
173 or any other Git operation (including 'git prune', as loose objects
174 are never used by fast-import).
176 fast-import does not lock the branch or tag refs it is actively importing.
177 After the import, during its ref update phase, fast-import tests each
178 existing branch ref to verify the update will be a fast-forward
179 update (the commit stored in the ref is contained in the new
180 history of the commit to be written). If the update is not a
181 fast-forward update, fast-import will skip updating that ref and instead
182 prints a warning message. fast-import will always attempt to update all
183 branch refs, and does not stop on the first failure.
185 Branch updates can be forced with --force, but it's recommended that
186 this only be used on an otherwise quiet repository. Using --force
187 is not necessary for an initial import into an empty repository.
192 fast-import tracks a set of branches in memory. Any branch can be created
193 or modified at any point during the import process by sending a
194 `commit` command on the input stream. This design allows a frontend
195 program to process an unlimited number of branches simultaneously,
196 generating commits in the order they are available from the source
197 data. It also simplifies the frontend programs considerably.
199 fast-import does not use or alter the current working directory, or any
200 file within it. (It does however update the current Git repository,
201 as referenced by `GIT_DIR`.) Therefore an import frontend may use
202 the working directory for its own purposes, such as extracting file
203 revisions from the foreign source. This ignorance of the working
204 directory also allows fast-import to run very quickly, as it does not
205 need to perform any costly file update operations when switching
210 With the exception of raw file data (which Git does not interpret)
211 the fast-import input format is text (ASCII) based. This text based
212 format simplifies development and debugging of frontend programs,
213 especially when a higher level language such as Perl, Python or
216 fast-import is very strict about its input. Where we say SP below we mean
217 *exactly* one space. Likewise LF means one (and only one) linefeed
218 and HT one (and only one) horizontal tab.
219 Supplying additional whitespace characters will cause unexpected
220 results, such as branch names or file names with leading or trailing
221 spaces in their name, or early termination of fast-import when it encounters
226 To aid in debugging frontends fast-import ignores any line that
227 begins with `#` (ASCII pound/hash) up to and including the line
228 ending `LF`. A comment line may contain any sequence of bytes
229 that does not contain an LF and therefore may be used to include
230 any detailed debugging information that might be specific to the
231 frontend and useful when inspecting a fast-import data stream.
235 The following date formats are supported. A frontend should select
236 the format it will use for this import by passing the format name
237 in the --date-format=<fmt> command-line option.
240 This is the Git native format and is `<time> SP <offutc>`.
241 It is also fast-import's default format, if --date-format was
244 The time of the event is specified by `<time>` as the number of
245 seconds since the UNIX epoch (midnight, Jan 1, 1970, UTC) and is
246 written as an ASCII decimal integer.
248 The local offset is specified by `<offutc>` as a positive or negative
249 offset from UTC. For example EST (which is 5 hours behind UTC)
250 would be expressed in `<tz>` by ``-0500'' while UTC is ``+0000''.
251 The local offset does not affect `<time>`; it is used only as an
252 advisement to help formatting routines display the timestamp.
254 If the local offset is not available in the source material, use
255 ``+0000'', or the most common local offset. For example many
256 organizations have a CVS repository which has only ever been accessed
257 by users who are located in the same location and time zone. In this
258 case a reasonable offset from UTC could be assumed.
260 Unlike the `rfc2822` format, this format is very strict. Any
261 variation in formatting will cause fast-import to reject the value.
264 This is the standard email format as described by RFC 2822.
266 An example value is ``Tue Feb 6 11:22:18 2007 -0500''. The Git
267 parser is accurate, but a little on the lenient side. It is the
268 same parser used by 'git am' when applying patches
271 Some malformed strings may be accepted as valid dates. In some of
272 these cases Git will still be able to obtain the correct date from
273 the malformed string. There are also some types of malformed
274 strings which Git will parse wrong, and yet consider valid.
275 Seriously malformed strings will be rejected.
277 Unlike the `raw` format above, the time zone/UTC offset information
278 contained in an RFC 2822 date string is used to adjust the date
279 value to UTC prior to storage. Therefore it is important that
280 this information be as accurate as possible.
282 If the source material uses RFC 2822 style dates,
283 the frontend should let fast-import handle the parsing and conversion
284 (rather than attempting to do it itself) as the Git parser has
285 been well tested in the wild.
287 Frontends should prefer the `raw` format if the source material
288 already uses UNIX-epoch format, can be coaxed to give dates in that
289 format, or its format is easily convertible to it, as there is no
290 ambiguity in parsing.
293 Always use the current time and time zone. The literal
294 `now` must always be supplied for `<when>`.
296 This is a toy format. The current time and time zone of this system
297 is always copied into the identity string at the time it is being
298 created by fast-import. There is no way to specify a different time or
301 This particular format is supplied as it's short to implement and
302 may be useful to a process that wants to create a new commit
303 right now, without needing to use a working directory or
306 If separate `author` and `committer` commands are used in a `commit`
307 the timestamps may not match, as the system clock will be polled
308 twice (once for each command). The only way to ensure that both
309 author and committer identity information has the same timestamp
310 is to omit `author` (thus copying from `committer`) or to use a
311 date format other than `now`.
315 fast-import accepts several commands to update the current repository
316 and control the current import process. More detailed discussion
317 (with examples) of each command follows later.
320 Creates a new branch or updates an existing branch by
321 creating a new commit and updating the branch to point at
322 the newly created commit.
325 Creates an annotated tag object from an existing commit or
326 branch. Lightweight tags are not supported by this command,
327 as they are not recommended for recording meaningful points
331 Reset an existing branch (or a new branch) to a specific
332 revision. This command must be used to change a branch to
333 a specific revision without making a commit on it.
336 Convert raw file data into a blob, for future use in a
337 `commit` command. This command is optional and is not
338 needed to perform an import.
341 Record that a mark refers to a given object without first
342 creating any new object. Using --import-marks and referring
343 to missing marks will cause fast-import to fail, so aliases
344 can provide a way to set otherwise pruned commits to a valid
345 value (e.g. the nearest non-pruned ancestor).
348 Forces fast-import to close the current packfile, generate its
349 unique SHA-1 checksum and index, and start a new packfile.
350 This command is optional and is not needed to perform
354 Causes fast-import to echo the entire line to its own
355 standard output. This command is optional and is not needed
356 to perform an import.
359 Marks the end of the stream. This command is optional
360 unless the `done` feature was requested using the
361 `--done` command-line option or `feature done` command.
364 Causes fast-import to print the SHA-1 corresponding to a mark
365 to the file descriptor set with `--cat-blob-fd`, or `stdout` if
369 Causes fast-import to print a blob in 'cat-file --batch'
370 format to the file descriptor set with `--cat-blob-fd` or
371 `stdout` if unspecified.
374 Causes fast-import to print a line describing a directory
375 entry in 'ls-tree' format to the file descriptor set with
376 `--cat-blob-fd` or `stdout` if unspecified.
379 Enable the specified feature. This requires that fast-import
380 supports the specified feature, and aborts if it does not.
383 Specify any of the options listed under OPTIONS that do not
384 change stream semantic to suit the frontend's needs. This
385 command is optional and is not needed to perform an import.
389 Create or update a branch with a new commit, recording one logical
390 change to the project.
396 ('author' (SP <name>)? SP LT <email> GT SP <when> LF)?
397 'committer' (SP <name>)? SP LT <email> GT SP <when> LF
398 ('encoding' SP <encoding>)?
400 ('from' SP <commit-ish> LF)?
401 ('merge' SP <commit-ish> LF)*
402 (filemodify | filedelete | filecopy | filerename | filedeleteall | notemodify)*
406 where `<ref>` is the name of the branch to make the commit on.
407 Typically branch names are prefixed with `refs/heads/` in
408 Git, so importing the CVS branch symbol `RELENG-1_0` would use
409 `refs/heads/RELENG-1_0` for the value of `<ref>`. The value of
410 `<ref>` must be a valid refname in Git. As `LF` is not valid in
411 a Git refname, no quoting or escaping syntax is supported here.
413 A `mark` command may optionally appear, requesting fast-import to save a
414 reference to the newly created commit for future use by the frontend
415 (see below for format). It is very common for frontends to mark
416 every commit they create, thereby allowing future branch creation
417 from any imported commit.
419 The `data` command following `committer` must supply the commit
420 message (see below for `data` command syntax). To import an empty
421 commit message use a 0 length data. Commit messages are free-form
422 and are not interpreted by Git. Currently they must be encoded in
423 UTF-8, as fast-import does not permit other encodings to be specified.
425 Zero or more `filemodify`, `filedelete`, `filecopy`, `filerename`,
426 `filedeleteall` and `notemodify` commands
427 may be included to update the contents of the branch prior to
428 creating the commit. These commands may be supplied in any order.
429 However it is recommended that a `filedeleteall` command precede
430 all `filemodify`, `filecopy`, `filerename` and `notemodify` commands in
431 the same commit, as `filedeleteall` wipes the branch clean (see below).
433 The `LF` after the command is optional (it used to be required). Note
434 that for reasons of backward compatibility, if the commit ends with a
435 `data` command (i.e. it has no `from`, `merge`, `filemodify`,
436 `filedelete`, `filecopy`, `filerename`, `filedeleteall` or
437 `notemodify` commands) then two `LF` commands may appear at the end of
438 the command instead of just one.
442 An `author` command may optionally appear, if the author information
443 might differ from the committer information. If `author` is omitted
444 then fast-import will automatically use the committer's information for
445 the author portion of the commit. See below for a description of
446 the fields in `author`, as they are identical to `committer`.
450 The `committer` command indicates who made this commit, and when
453 Here `<name>` is the person's display name (for example
454 ``Com M Itter'') and `<email>` is the person's email address
455 (``\cm@example.com''). `LT` and `GT` are the literal less-than (\x3c)
456 and greater-than (\x3e) symbols. These are required to delimit
457 the email address from the other fields in the line. Note that
458 `<name>` and `<email>` are free-form and may contain any sequence
459 of bytes, except `LT`, `GT` and `LF`. `<name>` is typically UTF-8 encoded.
461 The time of the change is specified by `<when>` using the date format
462 that was selected by the --date-format=<fmt> command-line option.
463 See ``Date Formats'' above for the set of supported formats, and
468 The optional `encoding` command indicates the encoding of the commit
469 message. Most commits are UTF-8 and the encoding is omitted, but this
470 allows importing commit messages into git without first reencoding them.
474 The `from` command is used to specify the commit to initialize
475 this branch from. This revision will be the first ancestor of the
476 new commit. The state of the tree built at this commit will begin
477 with the state at the `from` commit, and be altered by the content
478 modifications in this commit.
480 Omitting the `from` command in the first commit of a new branch
481 will cause fast-import to create that commit with no ancestor. This
482 tends to be desired only for the initial commit of a project.
483 If the frontend creates all files from scratch when making a new
484 branch, a `merge` command may be used instead of `from` to start
485 the commit with an empty tree.
486 Omitting the `from` command on existing branches is usually desired,
487 as the current commit on that branch is automatically assumed to
488 be the first ancestor of the new commit.
490 As `LF` is not valid in a Git refname or SHA-1 expression, no
491 quoting or escaping syntax is supported within `<commit-ish>`.
493 Here `<commit-ish>` is any of the following:
495 * The name of an existing branch already in fast-import's internal branch
496 table. If fast-import doesn't know the name, it's treated as a SHA-1
499 * A mark reference, `:<idnum>`, where `<idnum>` is the mark number.
501 The reason fast-import uses `:` to denote a mark reference is this character
502 is not legal in a Git branch name. The leading `:` makes it easy
503 to distinguish between the mark 42 (`:42`) and the branch 42 (`42`
504 or `refs/heads/42`), or an abbreviated SHA-1 which happened to
505 consist only of base-10 digits.
507 Marks must be declared (via `mark`) before they can be used.
509 * A complete 40 byte or abbreviated commit SHA-1 in hex.
511 * Any valid Git SHA-1 expression that resolves to a commit. See
512 ``SPECIFYING REVISIONS'' in linkgit:gitrevisions[7] for details.
514 * The special null SHA-1 (40 zeros) specifies that the branch is to be
517 The special case of restarting an incremental import from the
518 current branch value should be written as:
520 from refs/heads/branch^0
522 The `^0` suffix is necessary as fast-import does not permit a branch to
523 start from itself, and the branch is created in memory before the
524 `from` command is even read from the input. Adding `^0` will force
525 fast-import to resolve the commit through Git's revision parsing library,
526 rather than its internal branch table, thereby loading in the
527 existing value of the branch.
531 Includes one additional ancestor commit. The additional ancestry
532 link does not change the way the tree state is built at this commit.
533 If the `from` command is
534 omitted when creating a new branch, the first `merge` commit will be
535 the first ancestor of the current commit, and the branch will start
536 out with no files. An unlimited number of `merge` commands per
537 commit are permitted by fast-import, thereby establishing an n-way merge.
539 Here `<commit-ish>` is any of the commit specification expressions
540 also accepted by `from` (see above).
544 Included in a `commit` command to add a new file or change the
545 content of an existing file. This command has two different means
546 of specifying the content of the file.
548 External data format::
549 The data content for the file was already supplied by a prior
550 `blob` command. The frontend just needs to connect it.
553 'M' SP <mode> SP <dataref> SP <path> LF
556 Here usually `<dataref>` must be either a mark reference (`:<idnum>`)
557 set by a prior `blob` command, or a full 40-byte SHA-1 of an
558 existing Git blob object. If `<mode>` is `040000`` then
559 `<dataref>` must be the full 40-byte SHA-1 of an existing
560 Git tree object or a mark reference set with `--import-marks`.
563 The data content for the file has not been supplied yet.
564 The frontend wants to supply it as part of this modify
568 'M' SP <mode> SP 'inline' SP <path> LF
572 See below for a detailed description of the `data` command.
574 In both formats `<mode>` is the type of file entry, specified
575 in octal. Git only supports the following modes:
577 * `100644` or `644`: A normal (not-executable) file. The majority
578 of files in most projects use this mode. If in doubt, this is
580 * `100755` or `755`: A normal, but executable, file.
581 * `120000`: A symlink, the content of the file will be the link target.
582 * `160000`: A gitlink, SHA-1 of the object refers to a commit in
583 another repository. Git links can only be specified by SHA or through
584 a commit mark. They are used to implement submodules.
585 * `040000`: A subdirectory. Subdirectories can only be specified by
586 SHA or through a tree mark set with `--import-marks`.
588 In both formats `<path>` is the complete path of the file to be added
589 (if not already existing) or modified (if already existing).
591 A `<path>` string must use UNIX-style directory separators (forward
592 slash `/`), may contain any byte other than `LF`, and must not
593 start with double quote (`"`).
595 A path can use C-style string quoting; this is accepted in all cases
596 and mandatory if the filename starts with double quote or contains
597 `LF`. In C-style quoting, the complete name should be surrounded with
598 double quotes, and any `LF`, backslash, or double quote characters
599 must be escaped by preceding them with a backslash (e.g.,
600 `"path/with\n, \\ and \" in it"`).
602 The value of `<path>` must be in canonical form. That is it must not:
604 * contain an empty directory component (e.g. `foo//bar` is invalid),
605 * end with a directory separator (e.g. `foo/` is invalid),
606 * start with a directory separator (e.g. `/foo` is invalid),
607 * contain the special component `.` or `..` (e.g. `foo/./bar` and
608 `foo/../bar` are invalid).
610 The root of the tree can be represented by an empty string as `<path>`.
612 It is recommended that `<path>` always be encoded using UTF-8.
616 Included in a `commit` command to remove a file or recursively
617 delete an entire directory from the branch. If the file or directory
618 removal makes its parent directory empty, the parent directory will
619 be automatically removed too. This cascades up the tree until the
620 first non-empty directory or the root is reached.
626 here `<path>` is the complete path of the file or subdirectory to
627 be removed from the branch.
628 See `filemodify` above for a detailed description of `<path>`.
632 Recursively copies an existing file or subdirectory to a different
633 location within the branch. The existing file or directory must
634 exist. If the destination exists it will be completely replaced
635 by the content copied from the source.
638 'C' SP <path> SP <path> LF
641 here the first `<path>` is the source location and the second
642 `<path>` is the destination. See `filemodify` above for a detailed
643 description of what `<path>` may look like. To use a source path
644 that contains SP the path must be quoted.
646 A `filecopy` command takes effect immediately. Once the source
647 location has been copied to the destination any future commands
648 applied to the source location will not impact the destination of
653 Renames an existing file or subdirectory to a different location
654 within the branch. The existing file or directory must exist. If
655 the destination exists it will be replaced by the source directory.
658 'R' SP <path> SP <path> LF
661 here the first `<path>` is the source location and the second
662 `<path>` is the destination. See `filemodify` above for a detailed
663 description of what `<path>` may look like. To use a source path
664 that contains SP the path must be quoted.
666 A `filerename` command takes effect immediately. Once the source
667 location has been renamed to the destination any future commands
668 applied to the source location will create new files there and not
669 impact the destination of the rename.
671 Note that a `filerename` is the same as a `filecopy` followed by a
672 `filedelete` of the source location. There is a slight performance
673 advantage to using `filerename`, but the advantage is so small
674 that it is never worth trying to convert a delete/add pair in
675 source material into a rename for fast-import. This `filerename`
676 command is provided just to simplify frontends that already have
677 rename information and don't want bother with decomposing it into a
678 `filecopy` followed by a `filedelete`.
682 Included in a `commit` command to remove all files (and also all
683 directories) from the branch. This command resets the internal
684 branch structure to have no files in it, allowing the frontend
685 to subsequently add all interesting files from scratch.
691 This command is extremely useful if the frontend does not know
692 (or does not care to know) what files are currently on the branch,
693 and therefore cannot generate the proper `filedelete` commands to
696 Issuing a `filedeleteall` followed by the needed `filemodify`
697 commands to set the correct content will produce the same results
698 as sending only the needed `filemodify` and `filedelete` commands.
699 The `filedeleteall` approach may however require fast-import to use slightly
700 more memory per active branch (less than 1 MiB for even most large
701 projects); so frontends that can easily obtain only the affected
702 paths for a commit are encouraged to do so.
706 Included in a `commit` `<notes_ref>` command to add a new note
707 annotating a `<commit-ish>` or change this annotation contents.
708 Internally it is similar to filemodify 100644 on `<commit-ish>`
709 path (maybe split into subdirectories). It's not advised to
710 use any other commands to write to the `<notes_ref>` tree except
711 `filedeleteall` to delete all existing notes in this tree.
712 This command has two different means of specifying the content
715 External data format::
716 The data content for the note was already supplied by a prior
717 `blob` command. The frontend just needs to connect it to the
718 commit that is to be annotated.
721 'N' SP <dataref> SP <commit-ish> LF
724 Here `<dataref>` can be either a mark reference (`:<idnum>`)
725 set by a prior `blob` command, or a full 40-byte SHA-1 of an
726 existing Git blob object.
729 The data content for the note has not been supplied yet.
730 The frontend wants to supply it as part of this modify
734 'N' SP 'inline' SP <commit-ish> LF
738 See below for a detailed description of the `data` command.
740 In both formats `<commit-ish>` is any of the commit specification
741 expressions also accepted by `from` (see above).
745 Arranges for fast-import to save a reference to the current object, allowing
746 the frontend to recall this object at a future point in time, without
747 knowing its SHA-1. Here the current object is the object creation
748 command the `mark` command appears within. This can be `commit`,
749 `tag`, and `blob`, but `commit` is the most common usage.
752 'mark' SP ':' <idnum> LF
755 where `<idnum>` is the number assigned by the frontend to this mark.
756 The value of `<idnum>` is expressed as an ASCII decimal integer.
757 The value 0 is reserved and cannot be used as
758 a mark. Only values greater than or equal to 1 may be used as marks.
760 New marks are created automatically. Existing marks can be moved
761 to another object simply by reusing the same `<idnum>` in another
766 Provides the name of the object in the original source control system.
767 fast-import will simply ignore this directive, but filter processes
768 which operate on and modify the stream before feeding to fast-import
769 may have uses for this information
772 'original-oid' SP <object-identifier> LF
775 where `<object-identifer>` is any string not containing LF.
779 Creates an annotated tag referring to a specific commit. To create
780 lightweight (non-annotated) tags see the `reset` command below.
785 'from' SP <commit-ish> LF
787 'tagger' (SP <name>)? SP LT <email> GT SP <when> LF
791 where `<name>` is the name of the tag to create.
793 Tag names are automatically prefixed with `refs/tags/` when stored
794 in Git, so importing the CVS branch symbol `RELENG-1_0-FINAL` would
795 use just `RELENG-1_0-FINAL` for `<name>`, and fast-import will write the
796 corresponding ref as `refs/tags/RELENG-1_0-FINAL`.
798 The value of `<name>` must be a valid refname in Git and therefore
799 may contain forward slashes. As `LF` is not valid in a Git refname,
800 no quoting or escaping syntax is supported here.
802 The `from` command is the same as in the `commit` command; see
805 The `tagger` command uses the same format as `committer` within
806 `commit`; again see above for details.
808 The `data` command following `tagger` must supply the annotated tag
809 message (see below for `data` command syntax). To import an empty
810 tag message use a 0 length data. Tag messages are free-form and are
811 not interpreted by Git. Currently they must be encoded in UTF-8,
812 as fast-import does not permit other encodings to be specified.
814 Signing annotated tags during import from within fast-import is not
815 supported. Trying to include your own PGP/GPG signature is not
816 recommended, as the frontend does not (easily) have access to the
817 complete set of bytes which normally goes into such a signature.
818 If signing is required, create lightweight tags from within fast-import with
819 `reset`, then create the annotated versions of those tags offline
820 with the standard 'git tag' process.
824 Creates (or recreates) the named branch, optionally starting from
825 a specific revision. The reset command allows a frontend to issue
826 a new `from` command for an existing branch, or to create a new
827 branch from an existing commit without creating a new commit.
831 ('from' SP <commit-ish> LF)?
835 For a detailed description of `<ref>` and `<commit-ish>` see above
836 under `commit` and `from`.
838 The `LF` after the command is optional (it used to be required).
840 The `reset` command can also be used to create lightweight
841 (non-annotated) tags. For example:
848 would create the lightweight tag `refs/tags/938` referring to
849 whatever commit mark `:938` references.
853 Requests writing one file revision to the packfile. The revision
854 is not connected to any commit; this connection must be formed in
855 a subsequent `commit` command by referencing the blob through an
865 The mark command is optional here as some frontends have chosen
866 to generate the Git SHA-1 for the blob on their own, and feed that
867 directly to `commit`. This is typically more work than it's worth
868 however, as marks are inexpensive to store and easy to use.
872 Supplies raw data (for use as blob/file content, commit messages, or
873 annotated tag messages) to fast-import. Data can be supplied using an exact
874 byte count or delimited with a terminating line. Real frontends
875 intended for production-quality conversions should always use the
876 exact byte count format, as it is more robust and performs better.
877 The delimited format is intended primarily for testing fast-import.
879 Comment lines appearing within the `<raw>` part of `data` commands
880 are always taken to be part of the body of the data and are therefore
881 never ignored by fast-import. This makes it safe to import any
882 file/message content whose lines might start with `#`.
884 Exact byte count format::
885 The frontend must specify the number of bytes of data.
892 where `<count>` is the exact number of bytes appearing within
893 `<raw>`. The value of `<count>` is expressed as an ASCII decimal
894 integer. The `LF` on either side of `<raw>` is not
895 included in `<count>` and will not be included in the imported data.
897 The `LF` after `<raw>` is optional (it used to be required) but
898 recommended. Always including it makes debugging a fast-import
899 stream easier as the next command always starts in column 0
900 of the next line, even if `<raw>` did not end with an `LF`.
903 A delimiter string is used to mark the end of the data.
904 fast-import will compute the length by searching for the delimiter.
905 This format is primarily useful for testing and is not
906 recommended for real data.
909 'data' SP '<<' <delim> LF
915 where `<delim>` is the chosen delimiter string. The string `<delim>`
916 must not appear on a line by itself within `<raw>`, as otherwise
917 fast-import will think the data ends earlier than it really does. The `LF`
918 immediately trailing `<raw>` is part of `<raw>`. This is one of
919 the limitations of the delimited format, it is impossible to supply
920 a data chunk which does not have an LF as its last byte.
922 The `LF` after `<delim> LF` is optional (it used to be required).
926 Record that a mark refers to a given object without first creating any
932 'to' SP <commit-ish> LF
936 For a detailed description of `<commit-ish>` see above under `from`.
941 Forces fast-import to close the current packfile, start a new one, and to
942 save out all current branch refs, tags and marks.
949 Note that fast-import automatically switches packfiles when the current
950 packfile reaches --max-pack-size, or 4 GiB, whichever limit is
951 smaller. During an automatic packfile switch fast-import does not update
952 the branch refs, tags or marks.
954 As a `checkpoint` can require a significant amount of CPU time and
955 disk IO (to compute the overall pack SHA-1 checksum, generate the
956 corresponding index file, and update the refs) it can easily take
957 several minutes for a single `checkpoint` command to complete.
959 Frontends may choose to issue checkpoints during extremely large
960 and long running imports, or when they need to allow another Git
961 process access to a branch. However given that a 30 GiB Subversion
962 repository can be loaded into Git through fast-import in about 3 hours,
963 explicit checkpointing may not be necessary.
965 The `LF` after the command is optional (it used to be required).
969 Causes fast-import to print the entire `progress` line unmodified to
970 its standard output channel (file descriptor 1) when the command is
971 processed from the input stream. The command otherwise has no impact
972 on the current import, or on any of fast-import's internal state.
975 'progress' SP <any> LF
979 The `<any>` part of the command may contain any sequence of bytes
980 that does not contain `LF`. The `LF` after the command is optional.
981 Callers may wish to process the output through a tool such as sed to
982 remove the leading part of the line, for example:
985 frontend | git fast-import | sed 's/^progress //'
988 Placing a `progress` command immediately after a `checkpoint` will
989 inform the reader when the `checkpoint` has been completed and it
990 can safely access the refs that fast-import updated.
994 Causes fast-import to print the SHA-1 corresponding to a mark to
995 stdout or to the file descriptor previously arranged with the
996 `--cat-blob-fd` argument. The command otherwise has no impact on the
997 current import; its purpose is to retrieve SHA-1s that later commits
998 might want to refer to in their commit messages.
1001 'get-mark' SP ':' <idnum> LF
1004 See ``Responses To Commands'' below for details about how to read
1009 Causes fast-import to print a blob to a file descriptor previously
1010 arranged with the `--cat-blob-fd` argument. The command otherwise
1011 has no impact on the current import; its main purpose is to
1012 retrieve blobs that may be in fast-import's memory but not
1013 accessible from the target repository.
1016 'cat-blob' SP <dataref> LF
1019 The `<dataref>` can be either a mark reference (`:<idnum>`)
1020 set previously or a full 40-byte SHA-1 of a Git blob, preexisting or
1021 ready to be written.
1023 Output uses the same format as `git cat-file --batch`:
1026 <sha1> SP 'blob' SP <size> LF
1030 This command can be used where a `filemodify` directive can appear,
1031 allowing it to be used in the middle of a commit. For a `filemodify`
1032 using an inline directive, it can also appear right before the `data`
1035 See ``Responses To Commands'' below for details about how to read
1040 Prints information about the object at a path to a file descriptor
1041 previously arranged with the `--cat-blob-fd` argument. This allows
1042 printing a blob from the active commit (with `cat-blob`) or copying a
1043 blob or tree from a previous commit for use in the current one (with
1046 The `ls` command can also be used where a `filemodify` directive can
1047 appear, allowing it to be used in the middle of a commit.
1049 Reading from the active commit::
1050 This form can only be used in the middle of a `commit`.
1051 The path names a directory entry within fast-import's
1052 active commit. The path must be quoted in this case.
1058 Reading from a named tree::
1059 The `<dataref>` can be a mark reference (`:<idnum>`) or the
1060 full 40-byte SHA-1 of a Git tag, commit, or tree object,
1061 preexisting or waiting to be written.
1062 The path is relative to the top level of the tree
1063 named by `<dataref>`.
1066 'ls' SP <dataref> SP <path> LF
1069 See `filemodify` above for a detailed description of `<path>`.
1071 Output uses the same format as `git ls-tree <tree> -- <path>`:
1074 <mode> SP ('blob' | 'tree' | 'commit') SP <dataref> HT <path> LF
1077 The <dataref> represents the blob, tree, or commit object at <path>
1078 and can be used in later 'get-mark', 'cat-blob', 'filemodify', or
1081 If there is no file or subtree at that path, 'git fast-import' will
1085 missing SP <path> LF
1088 See ``Responses To Commands'' below for details about how to read
1093 Require that fast-import supports the specified feature, or abort if
1097 'feature' SP <feature> ('=' <argument>)? LF
1100 The <feature> part of the command may be any one of the following:
1107 Act as though the corresponding command-line option with
1108 a leading `--` was passed on the command line
1109 (see OPTIONS, above).
1112 import-marks-if-exists::
1113 Like --import-marks except in two respects: first, only one
1114 "feature import-marks" or "feature import-marks-if-exists"
1115 command is allowed per stream; second, an --import-marks=
1116 or --import-marks-if-exists command-line option overrides
1117 any of these "feature" commands in the stream; third,
1118 "feature import-marks-if-exists" like a corresponding
1119 command-line option silently skips a nonexistent file.
1124 Require that the backend support the 'get-mark', 'cat-blob',
1125 or 'ls' command respectively.
1126 Versions of fast-import not supporting the specified command
1127 will exit with a message indicating so.
1128 This lets the import error out early with a clear message,
1129 rather than wasting time on the early part of an import
1130 before the unsupported command is detected.
1133 Require that the backend support the 'notemodify' (N)
1134 subcommand to the 'commit' command.
1135 Versions of fast-import not supporting notes will exit
1136 with a message indicating so.
1139 Error out if the stream ends without a 'done' command.
1140 Without this feature, errors causing the frontend to end
1141 abruptly at a convenient point in the stream can go
1142 undetected. This may occur, for example, if an import
1143 front end dies in mid-operation without emitting SIGTERM
1144 or SIGKILL at its subordinate git fast-import instance.
1148 Processes the specified option so that git fast-import behaves in a
1149 way that suits the frontend's needs.
1150 Note that options specified by the frontend are overridden by any
1151 options the user may specify to git fast-import itself.
1154 'option' SP <option> LF
1157 The `<option>` part of the command may contain any of the options
1158 listed in the OPTIONS section that do not change import semantics,
1159 without the leading `--` and is treated in the same way.
1161 Option commands must be the first commands on the input (not counting
1162 feature commands), to give an option command after any non-option
1163 command is an error.
1165 The following command-line options change import semantics and may therefore
1166 not be passed as option:
1176 If the `done` feature is not in use, treated as if EOF was read.
1177 This can be used to tell fast-import to finish early.
1179 If the `--done` command-line option or `feature done` command is
1180 in use, the `done` command is mandatory and marks the end of the
1183 RESPONSES TO COMMANDS
1184 ---------------------
1185 New objects written by fast-import are not available immediately.
1186 Most fast-import commands have no visible effect until the next
1187 checkpoint (or completion). The frontend can send commands to
1188 fill fast-import's input pipe without worrying about how quickly
1189 they will take effect, which improves performance by simplifying
1192 For some frontends, though, it is useful to be able to read back
1193 data from the current repository as it is being updated (for
1194 example when the source material describes objects in terms of
1195 patches to be applied to previously imported objects). This can
1196 be accomplished by connecting the frontend and fast-import via
1197 bidirectional pipes:
1200 mkfifo fast-import-output
1201 frontend <fast-import-output |
1202 git fast-import >fast-import-output
1205 A frontend set up this way can use `progress`, `get-mark`, `ls`, and
1206 `cat-blob` commands to read information from the import in progress.
1208 To avoid deadlock, such frontends must completely consume any
1209 pending output from `progress`, `ls`, `get-mark`, and `cat-blob` before
1210 performing writes to fast-import that might block.
1214 If fast-import is supplied invalid input it will terminate with a
1215 non-zero exit status and create a crash report in the top level of
1216 the Git repository it was importing into. Crash reports contain
1217 a snapshot of the internal fast-import state as well as the most
1218 recent commands that lead up to the crash.
1220 All recent commands (including stream comments, file changes and
1221 progress commands) are shown in the command history within the crash
1222 report, but raw file data and commit messages are excluded from the
1223 crash report. This exclusion saves space within the report file
1224 and reduces the amount of buffering that fast-import must perform
1227 After writing a crash report fast-import will close the current
1228 packfile and export the marks table. This allows the frontend
1229 developer to inspect the repository state and resume the import from
1230 the point where it crashed. The modified branches and tags are not
1231 updated during a crash, as the import did not complete successfully.
1232 Branch and tag information can be found in the crash report and
1233 must be applied manually if the update is needed.
1238 $ cat >in <<END_OF_INPUT
1239 # my very first test commit
1240 commit refs/heads/master
1241 committer Shawn O. Pearce <spearce> 19283 -0400
1242 # who is that guy anyway?
1246 M 644 inline .gitignore
1253 $ git fast-import <in
1254 fatal: Corrupt mode: M 777 inline bob
1255 fast-import: dumping crash report to .git/fast_import_crash_8434
1257 $ cat .git/fast_import_crash_8434
1258 fast-import crash report:
1259 fast-import process: 8434
1260 parent process : 1391
1261 at Sat Sep 1 00:58:12 2007
1263 fatal: Corrupt mode: M 777 inline bob
1265 Most Recent Commands Before Crash
1266 ---------------------------------
1267 # my very first test commit
1268 commit refs/heads/master
1269 committer Shawn O. Pearce <spearce> 19283 -0400
1270 # who is that guy anyway?
1272 M 644 inline .gitignore
1278 active_branches = 1 cur, 5 max
1281 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1282 1) 0 refs/heads/master
1287 status : active loaded dirty
1288 tip commit : 0000000000000000000000000000000000000000
1289 old tree : 0000000000000000000000000000000000000000
1290 cur tree : 0000000000000000000000000000000000000000
1301 The following tips and tricks have been collected from various
1302 users of fast-import, and are offered here as suggestions.
1304 Use One Mark Per Commit
1305 ~~~~~~~~~~~~~~~~~~~~~~~
1306 When doing a repository conversion, use a unique mark per commit
1307 (`mark :<n>`) and supply the --export-marks option on the command
1308 line. fast-import will dump a file which lists every mark and the Git
1309 object SHA-1 that corresponds to it. If the frontend can tie
1310 the marks back to the source repository, it is easy to verify the
1311 accuracy and completeness of the import by comparing each Git
1312 commit to the corresponding source revision.
1314 Coming from a system such as Perforce or Subversion this should be
1315 quite simple, as the fast-import mark can also be the Perforce changeset
1316 number or the Subversion revision number.
1318 Freely Skip Around Branches
1319 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
1320 Don't bother trying to optimize the frontend to stick to one branch
1321 at a time during an import. Although doing so might be slightly
1322 faster for fast-import, it tends to increase the complexity of the frontend
1325 The branch LRU builtin to fast-import tends to behave very well, and the
1326 cost of activating an inactive branch is so low that bouncing around
1327 between branches has virtually no impact on import performance.
1331 When importing a renamed file or directory, simply delete the old
1332 name(s) and modify the new name(s) during the corresponding commit.
1333 Git performs rename detection after-the-fact, rather than explicitly
1336 Use Tag Fixup Branches
1337 ~~~~~~~~~~~~~~~~~~~~~~
1338 Some other SCM systems let the user create a tag from multiple
1339 files which are not from the same commit/changeset. Or to create
1340 tags which are a subset of the files available in the repository.
1342 Importing these tags as-is in Git is impossible without making at
1343 least one commit which ``fixes up'' the files to match the content
1344 of the tag. Use fast-import's `reset` command to reset a dummy branch
1345 outside of your normal branch space to the base commit for the tag,
1346 then commit one or more file fixup commits, and finally tag the
1349 For example since all normal branches are stored under `refs/heads/`
1350 name the tag fixup branch `TAG_FIXUP`. This way it is impossible for
1351 the fixup branch used by the importer to have namespace conflicts
1352 with real branches imported from the source (the name `TAG_FIXUP`
1353 is not `refs/heads/TAG_FIXUP`).
1355 When committing fixups, consider using `merge` to connect the
1356 commit(s) which are supplying file revisions to the fixup branch.
1357 Doing so will allow tools such as 'git blame' to track
1358 through the real commit history and properly annotate the source
1361 After fast-import terminates the frontend will need to do `rm .git/TAG_FIXUP`
1362 to remove the dummy branch.
1364 Import Now, Repack Later
1365 ~~~~~~~~~~~~~~~~~~~~~~~~
1366 As soon as fast-import completes the Git repository is completely valid
1367 and ready for use. Typically this takes only a very short time,
1368 even for considerably large projects (100,000+ commits).
1370 However repacking the repository is necessary to improve data
1371 locality and access performance. It can also take hours on extremely
1372 large projects (especially if -f and a large --window parameter is
1373 used). Since repacking is safe to run alongside readers and writers,
1374 run the repack in the background and let it finish when it finishes.
1375 There is no reason to wait to explore your new Git project!
1377 If you choose to wait for the repack, don't try to run benchmarks
1378 or performance tests until repacking is completed. fast-import outputs
1379 suboptimal packfiles that are simply never seen in real use
1382 Repacking Historical Data
1383 ~~~~~~~~~~~~~~~~~~~~~~~~~
1384 If you are repacking very old imported data (e.g. older than the
1385 last year), consider expending some extra CPU time and supplying
1386 --window=50 (or higher) when you run 'git repack'.
1387 This will take longer, but will also produce a smaller packfile.
1388 You only need to expend the effort once, and everyone using your
1389 project will benefit from the smaller repository.
1391 Include Some Progress Messages
1392 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1393 Every once in a while have your frontend emit a `progress` message
1394 to fast-import. The contents of the messages are entirely free-form,
1395 so one suggestion would be to output the current month and year
1396 each time the current commit date moves into the next month.
1397 Your users will feel better knowing how much of the data stream
1401 PACKFILE OPTIMIZATION
1402 ---------------------
1403 When packing a blob fast-import always attempts to deltify against the last
1404 blob written. Unless specifically arranged for by the frontend,
1405 this will probably not be a prior version of the same file, so the
1406 generated delta will not be the smallest possible. The resulting
1407 packfile will be compressed, but will not be optimal.
1409 Frontends which have efficient access to all revisions of a
1410 single file (for example reading an RCS/CVS ,v file) can choose
1411 to supply all revisions of that file as a sequence of consecutive
1412 `blob` commands. This allows fast-import to deltify the different file
1413 revisions against each other, saving space in the final packfile.
1414 Marks can be used to later identify individual file revisions during
1415 a sequence of `commit` commands.
1417 The packfile(s) created by fast-import do not encourage good disk access
1418 patterns. This is caused by fast-import writing the data in the order
1419 it is received on standard input, while Git typically organizes
1420 data within packfiles to make the most recent (current tip) data
1421 appear before historical data. Git also clusters commits together,
1422 speeding up revision traversal through better cache locality.
1424 For this reason it is strongly recommended that users repack the
1425 repository with `git repack -a -d` after fast-import completes, allowing
1426 Git to reorganize the packfiles for faster data access. If blob
1427 deltas are suboptimal (see above) then also adding the `-f` option
1428 to force recomputation of all deltas can significantly reduce the
1429 final packfile size (30-50% smaller can be quite typical).
1431 Instead of running `git repack` you can also run `git gc
1432 --aggressive`, which will also optimize other things after an import
1433 (e.g. pack loose refs). As noted in the "AGGRESSIVE" section in
1434 linkgit:git-gc[1] the `--aggressive` option will find new deltas with
1435 the `-f` option to linkgit:git-repack[1]. For the reasons elaborated
1436 on above using `--aggressive` after a fast-import is one of the few
1437 cases where it's known to be worthwhile.
1441 There are a number of factors which affect how much memory fast-import
1442 requires to perform an import. Like critical sections of core
1443 Git, fast-import uses its own memory allocators to amortize any overheads
1444 associated with malloc. In practice fast-import tends to amortize any
1445 malloc overheads to 0, due to its use of large block allocations.
1449 fast-import maintains an in-memory structure for every object written in
1450 this execution. On a 32 bit system the structure is 32 bytes,
1451 on a 64 bit system the structure is 40 bytes (due to the larger
1452 pointer sizes). Objects in the table are not deallocated until
1453 fast-import terminates. Importing 2 million objects on a 32 bit system
1454 will require approximately 64 MiB of memory.
1456 The object table is actually a hashtable keyed on the object name
1457 (the unique SHA-1). This storage configuration allows fast-import to reuse
1458 an existing or already written object and avoid writing duplicates
1459 to the output packfile. Duplicate blobs are surprisingly common
1460 in an import, typically due to branch merges in the source.
1464 Marks are stored in a sparse array, using 1 pointer (4 bytes or 8
1465 bytes, depending on pointer size) per mark. Although the array
1466 is sparse, frontends are still strongly encouraged to use marks
1467 between 1 and n, where n is the total number of marks required for
1472 Branches are classified as active and inactive. The memory usage
1473 of the two classes is significantly different.
1475 Inactive branches are stored in a structure which uses 96 or 120
1476 bytes (32 bit or 64 bit systems, respectively), plus the length of
1477 the branch name (typically under 200 bytes), per branch. fast-import will
1478 easily handle as many as 10,000 inactive branches in under 2 MiB
1481 Active branches have the same overhead as inactive branches, but
1482 also contain copies of every tree that has been recently modified on
1483 that branch. If subtree `include` has not been modified since the
1484 branch became active, its contents will not be loaded into memory,
1485 but if subtree `src` has been modified by a commit since the branch
1486 became active, then its contents will be loaded in memory.
1488 As active branches store metadata about the files contained on that
1489 branch, their in-memory storage size can grow to a considerable size
1492 fast-import automatically moves active branches to inactive status based on
1493 a simple least-recently-used algorithm. The LRU chain is updated on
1494 each `commit` command. The maximum number of active branches can be
1495 increased or decreased on the command line with --active-branches=.
1499 Trees (aka directories) use just 12 bytes of memory on top of the
1500 memory required for their entries (see ``per active file'' below).
1501 The cost of a tree is virtually 0, as its overhead amortizes out
1502 over the individual file entries.
1504 per active file entry
1505 ~~~~~~~~~~~~~~~~~~~~~
1506 Files (and pointers to subtrees) within active trees require 52 or 64
1507 bytes (32/64 bit platforms) per entry. To conserve space, file and
1508 tree names are pooled in a common string table, allowing the filename
1509 ``Makefile'' to use just 16 bytes (after including the string header
1510 overhead) no matter how many times it occurs within the project.
1512 The active branch LRU, when coupled with the filename string pool
1513 and lazy loading of subtrees, allows fast-import to efficiently import
1514 projects with 2,000+ branches and 45,114+ files in a very limited
1515 memory footprint (less than 2.7 MiB per active branch).
1519 Sending *SIGUSR1* to the 'git fast-import' process ends the current
1520 packfile early, simulating a `checkpoint` command. The impatient
1521 operator can use this facility to peek at the objects and refs from an
1522 import in progress, at the cost of some added running time and worse
1527 linkgit:git-fast-export[1]
1531 Part of the linkgit:git[1] suite
projects / thirdparty / git.git / blob