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.
37 Specify the type of dates the frontend will supply to
38 fast-import within `author`, `committer` and `tagger` commands.
39 See ``Date Formats'' below for details about which formats
40 are supported, and their syntax.
43 Force updating modified existing branches, even if doing
44 so would cause commits to be lost (as the new commit does
45 not contain the old commit).
48 Maximum size of each output packfile.
49 The default is unlimited.
51 --big-file-threshold=<n>::
52 Maximum size of a blob that fast-import will attempt to
53 create a delta for, expressed in bytes. The default is 512m
54 (512 MiB). Some importers may wish to lower this on systems
55 with constrained memory.
58 Maximum delta depth, for blob and tree deltification.
61 --active-branches=<n>::
62 Maximum number of branches to maintain active at once.
63 See ``Memory Utilization'' below for details. Default is 5.
65 --export-marks=<file>::
66 Dumps the internal marks table to <file> when complete.
67 Marks are written one per line as `:markid SHA-1`.
68 Frontends can use this file to validate imports after they
69 have been completed, or to save the marks table across
70 incremental runs. As <file> is only opened and truncated
71 at checkpoint (or completion) the same path can also be
72 safely given to \--import-marks.
74 --import-marks=<file>::
75 Before processing any input, load the marks specified in
76 <file>. The input file must exist, must be readable, and
77 must use the same format as produced by \--export-marks.
78 Multiple options may be supplied to import more than one
79 set of marks. If a mark is defined to different values,
82 --import-marks-if-exists=<file>::
83 Like --import-marks but instead of erroring out, silently
84 skips the file if it does not exist.
87 After specifying --relative-marks the paths specified
88 with --import-marks= and --export-marks= are relative
89 to an internal directory in the current repository.
90 In git-fast-import this means that the paths are relative
91 to the .git/info/fast-import directory. However, other
92 importers may use a different location.
95 Negates a previous --relative-marks. Allows for combining
96 relative and non-relative marks by interweaving
97 --(no-)-relative-marks with the --(import|export)-marks=
101 Specify the file descriptor that will be written to
102 when the `cat-blob` command is encountered in the stream.
103 The default behaviour is to write to `stdout`.
105 --export-pack-edges=<file>::
106 After creating a packfile, print a line of data to
107 <file> listing the filename of the packfile and the last
108 commit on each branch that was written to that packfile.
109 This information may be useful after importing projects
110 whose total object set exceeds the 4 GiB packfile limit,
111 as these commits can be used as edge points during calls
112 to 'git pack-objects'.
115 Disable all non-fatal output, making fast-import silent when it
116 is successful. This option disables the output shown by
120 Display some basic statistics about the objects fast-import has
121 created, the packfiles they were stored into, and the
122 memory used by fast-import during this run. Showing this output
123 is currently the default, but can be disabled with \--quiet.
128 The design of fast-import allows it to import large projects in a minimum
129 amount of memory usage and processing time. Assuming the frontend
130 is able to keep up with fast-import and feed it a constant stream of data,
131 import times for projects holding 10+ years of history and containing
132 100,000+ individual commits are generally completed in just 1-2
133 hours on quite modest (~$2,000 USD) hardware.
135 Most bottlenecks appear to be in foreign source data access (the
136 source just cannot extract revisions fast enough) or disk IO (fast-import
137 writes as fast as the disk will take the data). Imports will run
138 faster if the source data is stored on a different drive than the
139 destination Git repository (due to less IO contention).
144 A typical frontend for fast-import tends to weigh in at approximately 200
145 lines of Perl/Python/Ruby code. Most developers have been able to
146 create working importers in just a couple of hours, even though it
147 is their first exposure to fast-import, and sometimes even to Git. This is
148 an ideal situation, given that most conversion tools are throw-away
149 (use once, and never look back).
154 Like 'git push' or 'git fetch', imports handled by fast-import are safe to
155 run alongside parallel `git repack -a -d` or `git gc` invocations,
156 or any other Git operation (including 'git prune', as loose objects
157 are never used by fast-import).
159 fast-import does not lock the branch or tag refs it is actively importing.
160 After the import, during its ref update phase, fast-import tests each
161 existing branch ref to verify the update will be a fast-forward
162 update (the commit stored in the ref is contained in the new
163 history of the commit to be written). If the update is not a
164 fast-forward update, fast-import will skip updating that ref and instead
165 prints a warning message. fast-import will always attempt to update all
166 branch refs, and does not stop on the first failure.
168 Branch updates can be forced with \--force, but it's recommended that
169 this only be used on an otherwise quiet repository. Using \--force
170 is not necessary for an initial import into an empty repository.
175 fast-import tracks a set of branches in memory. Any branch can be created
176 or modified at any point during the import process by sending a
177 `commit` command on the input stream. This design allows a frontend
178 program to process an unlimited number of branches simultaneously,
179 generating commits in the order they are available from the source
180 data. It also simplifies the frontend programs considerably.
182 fast-import does not use or alter the current working directory, or any
183 file within it. (It does however update the current Git repository,
184 as referenced by `GIT_DIR`.) Therefore an import frontend may use
185 the working directory for its own purposes, such as extracting file
186 revisions from the foreign source. This ignorance of the working
187 directory also allows fast-import to run very quickly, as it does not
188 need to perform any costly file update operations when switching
193 With the exception of raw file data (which Git does not interpret)
194 the fast-import input format is text (ASCII) based. This text based
195 format simplifies development and debugging of frontend programs,
196 especially when a higher level language such as Perl, Python or
199 fast-import is very strict about its input. Where we say SP below we mean
200 *exactly* one space. Likewise LF means one (and only one) linefeed
201 and HT one (and only one) horizontal tab.
202 Supplying additional whitespace characters will cause unexpected
203 results, such as branch names or file names with leading or trailing
204 spaces in their name, or early termination of fast-import when it encounters
209 To aid in debugging frontends fast-import ignores any line that
210 begins with `#` (ASCII pound/hash) up to and including the line
211 ending `LF`. A comment line may contain any sequence of bytes
212 that does not contain an LF and therefore may be used to include
213 any detailed debugging information that might be specific to the
214 frontend and useful when inspecting a fast-import data stream.
218 The following date formats are supported. A frontend should select
219 the format it will use for this import by passing the format name
220 in the \--date-format=<fmt> command line option.
223 This is the Git native format and is `<time> SP <offutc>`.
224 It is also fast-import's default format, if \--date-format was
227 The time of the event is specified by `<time>` as the number of
228 seconds since the UNIX epoch (midnight, Jan 1, 1970, UTC) and is
229 written as an ASCII decimal integer.
231 The local offset is specified by `<offutc>` as a positive or negative
232 offset from UTC. For example EST (which is 5 hours behind UTC)
233 would be expressed in `<tz>` by ``-0500'' while UTC is ``+0000''.
234 The local offset does not affect `<time>`; it is used only as an
235 advisement to help formatting routines display the timestamp.
237 If the local offset is not available in the source material, use
238 ``+0000'', or the most common local offset. For example many
239 organizations have a CVS repository which has only ever been accessed
240 by users who are located in the same location and timezone. In this
241 case a reasonable offset from UTC could be assumed.
243 Unlike the `rfc2822` format, this format is very strict. Any
244 variation in formatting will cause fast-import to reject the value.
247 This is the standard email format as described by RFC 2822.
249 An example value is ``Tue Feb 6 11:22:18 2007 -0500''. The Git
250 parser is accurate, but a little on the lenient side. It is the
251 same parser used by 'git am' when applying patches
254 Some malformed strings may be accepted as valid dates. In some of
255 these cases Git will still be able to obtain the correct date from
256 the malformed string. There are also some types of malformed
257 strings which Git will parse wrong, and yet consider valid.
258 Seriously malformed strings will be rejected.
260 Unlike the `raw` format above, the timezone/UTC offset information
261 contained in an RFC 2822 date string is used to adjust the date
262 value to UTC prior to storage. Therefore it is important that
263 this information be as accurate as possible.
265 If the source material uses RFC 2822 style dates,
266 the frontend should let fast-import handle the parsing and conversion
267 (rather than attempting to do it itself) as the Git parser has
268 been well tested in the wild.
270 Frontends should prefer the `raw` format if the source material
271 already uses UNIX-epoch format, can be coaxed to give dates in that
272 format, or its format is easily convertible to it, as there is no
273 ambiguity in parsing.
276 Always use the current time and timezone. The literal
277 `now` must always be supplied for `<when>`.
279 This is a toy format. The current time and timezone of this system
280 is always copied into the identity string at the time it is being
281 created by fast-import. There is no way to specify a different time or
284 This particular format is supplied as it's short to implement and
285 may be useful to a process that wants to create a new commit
286 right now, without needing to use a working directory or
289 If separate `author` and `committer` commands are used in a `commit`
290 the timestamps may not match, as the system clock will be polled
291 twice (once for each command). The only way to ensure that both
292 author and committer identity information has the same timestamp
293 is to omit `author` (thus copying from `committer`) or to use a
294 date format other than `now`.
298 fast-import accepts several commands to update the current repository
299 and control the current import process. More detailed discussion
300 (with examples) of each command follows later.
303 Creates a new branch or updates an existing branch by
304 creating a new commit and updating the branch to point at
305 the newly created commit.
308 Creates an annotated tag object from an existing commit or
309 branch. Lightweight tags are not supported by this command,
310 as they are not recommended for recording meaningful points
314 Reset an existing branch (or a new branch) to a specific
315 revision. This command must be used to change a branch to
316 a specific revision without making a commit on it.
319 Convert raw file data into a blob, for future use in a
320 `commit` command. This command is optional and is not
321 needed to perform an import.
324 Forces fast-import to close the current packfile, generate its
325 unique SHA-1 checksum and index, and start a new packfile.
326 This command is optional and is not needed to perform
330 Causes fast-import to echo the entire line to its own
331 standard output. This command is optional and is not needed
332 to perform an import.
335 Causes fast-import to print a blob in 'cat-file --batch'
336 format to the file descriptor set with `--cat-blob-fd` or
337 `stdout` if unspecified.
340 Causes fast-import to print a line describing a directory
341 entry in 'ls-tree' format to the file descriptor set with
342 `--cat-blob-fd` or `stdout` if unspecified.
345 Require that fast-import supports the specified feature, or
346 abort if it does not.
349 Specify any of the options listed under OPTIONS that do not
350 change stream semantic to suit the frontend's needs. This
351 command is optional and is not needed to perform an import.
355 Create or update a branch with a new commit, recording one logical
356 change to the project.
361 ('author' (SP <name>)? SP LT <email> GT SP <when> LF)?
362 'committer' (SP <name>)? SP LT <email> GT SP <when> LF
364 ('from' SP <committish> LF)?
365 ('merge' SP <committish> LF)?
366 (filemodify | filedelete | filecopy | filerename | filedeleteall | notemodify)*
370 where `<ref>` is the name of the branch to make the commit on.
371 Typically branch names are prefixed with `refs/heads/` in
372 Git, so importing the CVS branch symbol `RELENG-1_0` would use
373 `refs/heads/RELENG-1_0` for the value of `<ref>`. The value of
374 `<ref>` must be a valid refname in Git. As `LF` is not valid in
375 a Git refname, no quoting or escaping syntax is supported here.
377 A `mark` command may optionally appear, requesting fast-import to save a
378 reference to the newly created commit for future use by the frontend
379 (see below for format). It is very common for frontends to mark
380 every commit they create, thereby allowing future branch creation
381 from any imported commit.
383 The `data` command following `committer` must supply the commit
384 message (see below for `data` command syntax). To import an empty
385 commit message use a 0 length data. Commit messages are free-form
386 and are not interpreted by Git. Currently they must be encoded in
387 UTF-8, as fast-import does not permit other encodings to be specified.
389 Zero or more `filemodify`, `filedelete`, `filecopy`, `filerename`,
390 `filedeleteall` and `notemodify` commands
391 may be included to update the contents of the branch prior to
392 creating the commit. These commands may be supplied in any order.
393 However it is recommended that a `filedeleteall` command precede
394 all `filemodify`, `filecopy`, `filerename` and `notemodify` commands in
395 the same commit, as `filedeleteall` wipes the branch clean (see below).
397 The `LF` after the command is optional (it used to be required).
401 An `author` command may optionally appear, if the author information
402 might differ from the committer information. If `author` is omitted
403 then fast-import will automatically use the committer's information for
404 the author portion of the commit. See below for a description of
405 the fields in `author`, as they are identical to `committer`.
409 The `committer` command indicates who made this commit, and when
412 Here `<name>` is the person's display name (for example
413 ``Com M Itter'') and `<email>` is the person's email address
414 (``cm@example.com''). `LT` and `GT` are the literal less-than (\x3c)
415 and greater-than (\x3e) symbols. These are required to delimit
416 the email address from the other fields in the line. Note that
417 `<name>` is free-form and may contain any sequence of bytes, except
418 `LT` and `LF`. It is typically UTF-8 encoded.
420 The time of the change is specified by `<when>` using the date format
421 that was selected by the \--date-format=<fmt> command line option.
422 See ``Date Formats'' above for the set of supported formats, and
427 The `from` command is used to specify the commit to initialize
428 this branch from. This revision will be the first ancestor of the
431 Omitting the `from` command in the first commit of a new branch
432 will cause fast-import to create that commit with no ancestor. This
433 tends to be desired only for the initial commit of a project.
434 If the frontend creates all files from scratch when making a new
435 branch, a `merge` command may be used instead of `from` to start
436 the commit with an empty tree.
437 Omitting the `from` command on existing branches is usually desired,
438 as the current commit on that branch is automatically assumed to
439 be the first ancestor of the new commit.
441 As `LF` is not valid in a Git refname or SHA-1 expression, no
442 quoting or escaping syntax is supported within `<committish>`.
444 Here `<committish>` is any of the following:
446 * The name of an existing branch already in fast-import's internal branch
447 table. If fast-import doesn't know the name, it's treated as a SHA-1
450 * A mark reference, `:<idnum>`, where `<idnum>` is the mark number.
452 The reason fast-import uses `:` to denote a mark reference is this character
453 is not legal in a Git branch name. The leading `:` makes it easy
454 to distinguish between the mark 42 (`:42`) and the branch 42 (`42`
455 or `refs/heads/42`), or an abbreviated SHA-1 which happened to
456 consist only of base-10 digits.
458 Marks must be declared (via `mark`) before they can be used.
460 * A complete 40 byte or abbreviated commit SHA-1 in hex.
462 * Any valid Git SHA-1 expression that resolves to a commit. See
463 ``SPECIFYING REVISIONS'' in linkgit:gitrevisions[7] for details.
465 The special case of restarting an incremental import from the
466 current branch value should be written as:
468 from refs/heads/branch^0
470 The `{caret}0` suffix is necessary as fast-import does not permit a branch to
471 start from itself, and the branch is created in memory before the
472 `from` command is even read from the input. Adding `{caret}0` will force
473 fast-import to resolve the commit through Git's revision parsing library,
474 rather than its internal branch table, thereby loading in the
475 existing value of the branch.
479 Includes one additional ancestor commit. If the `from` command is
480 omitted when creating a new branch, the first `merge` commit will be
481 the first ancestor of the current commit, and the branch will start
482 out with no files. An unlimited number of `merge` commands per
483 commit are permitted by fast-import, thereby establishing an n-way merge.
484 However Git's other tools never create commits with more than 15
485 additional ancestors (forming a 16-way merge). For this reason
486 it is suggested that frontends do not use more than 15 `merge`
487 commands per commit; 16, if starting a new, empty branch.
489 Here `<committish>` is any of the commit specification expressions
490 also accepted by `from` (see above).
494 Included in a `commit` command to add a new file or change the
495 content of an existing file. This command has two different means
496 of specifying the content of the file.
498 External data format::
499 The data content for the file was already supplied by a prior
500 `blob` command. The frontend just needs to connect it.
503 'M' SP <mode> SP <dataref> SP <path> LF
506 Here usually `<dataref>` must be either a mark reference (`:<idnum>`)
507 set by a prior `blob` command, or a full 40-byte SHA-1 of an
508 existing Git blob object. If `<mode>` is `040000`` then
509 `<dataref>` must be the full 40-byte SHA-1 of an existing
510 Git tree object or a mark reference set with `--import-marks`.
513 The data content for the file has not been supplied yet.
514 The frontend wants to supply it as part of this modify
518 'M' SP <mode> SP 'inline' SP <path> LF
522 See below for a detailed description of the `data` command.
524 In both formats `<mode>` is the type of file entry, specified
525 in octal. Git only supports the following modes:
527 * `100644` or `644`: A normal (not-executable) file. The majority
528 of files in most projects use this mode. If in doubt, this is
530 * `100755` or `755`: A normal, but executable, file.
531 * `120000`: A symlink, the content of the file will be the link target.
532 * `160000`: A gitlink, SHA-1 of the object refers to a commit in
533 another repository. Git links can only be specified by SHA or through
534 a commit mark. They are used to implement submodules.
535 * `040000`: A subdirectory. Subdirectories can only be specified by
536 SHA or through a tree mark set with `--import-marks`.
538 In both formats `<path>` is the complete path of the file to be added
539 (if not already existing) or modified (if already existing).
541 A `<path>` string must use UNIX-style directory separators (forward
542 slash `/`), may contain any byte other than `LF`, and must not
543 start with double quote (`"`).
545 If an `LF` or double quote must be encoded into `<path>` shell-style
546 quoting should be used, e.g. `"path/with\n and \" in it"`.
548 The value of `<path>` must be in canonical form. That is it must not:
550 * contain an empty directory component (e.g. `foo//bar` is invalid),
551 * end with a directory separator (e.g. `foo/` is invalid),
552 * start with a directory separator (e.g. `/foo` is invalid),
553 * contain the special component `.` or `..` (e.g. `foo/./bar` and
554 `foo/../bar` are invalid).
556 The root of the tree can be represented by an empty string as `<path>`.
558 It is recommended that `<path>` always be encoded using UTF-8.
562 Included in a `commit` command to remove a file or recursively
563 delete an entire directory from the branch. If the file or directory
564 removal makes its parent directory empty, the parent directory will
565 be automatically removed too. This cascades up the tree until the
566 first non-empty directory or the root is reached.
572 here `<path>` is the complete path of the file or subdirectory to
573 be removed from the branch.
574 See `filemodify` above for a detailed description of `<path>`.
578 Recursively copies an existing file or subdirectory to a different
579 location within the branch. The existing file or directory must
580 exist. If the destination exists it will be completely replaced
581 by the content copied from the source.
584 'C' SP <path> SP <path> LF
587 here the first `<path>` is the source location and the second
588 `<path>` is the destination. See `filemodify` above for a detailed
589 description of what `<path>` may look like. To use a source path
590 that contains SP the path must be quoted.
592 A `filecopy` command takes effect immediately. Once the source
593 location has been copied to the destination any future commands
594 applied to the source location will not impact the destination of
599 Renames an existing file or subdirectory to a different location
600 within the branch. The existing file or directory must exist. If
601 the destination exists it will be replaced by the source directory.
604 'R' SP <path> SP <path> LF
607 here the first `<path>` is the source location and the second
608 `<path>` is the destination. See `filemodify` above for a detailed
609 description of what `<path>` may look like. To use a source path
610 that contains SP the path must be quoted.
612 A `filerename` command takes effect immediately. Once the source
613 location has been renamed to the destination any future commands
614 applied to the source location will create new files there and not
615 impact the destination of the rename.
617 Note that a `filerename` is the same as a `filecopy` followed by a
618 `filedelete` of the source location. There is a slight performance
619 advantage to using `filerename`, but the advantage is so small
620 that it is never worth trying to convert a delete/add pair in
621 source material into a rename for fast-import. This `filerename`
622 command is provided just to simplify frontends that already have
623 rename information and don't want bother with decomposing it into a
624 `filecopy` followed by a `filedelete`.
628 Included in a `commit` command to remove all files (and also all
629 directories) from the branch. This command resets the internal
630 branch structure to have no files in it, allowing the frontend
631 to subsequently add all interesting files from scratch.
637 This command is extremely useful if the frontend does not know
638 (or does not care to know) what files are currently on the branch,
639 and therefore cannot generate the proper `filedelete` commands to
642 Issuing a `filedeleteall` followed by the needed `filemodify`
643 commands to set the correct content will produce the same results
644 as sending only the needed `filemodify` and `filedelete` commands.
645 The `filedeleteall` approach may however require fast-import to use slightly
646 more memory per active branch (less than 1 MiB for even most large
647 projects); so frontends that can easily obtain only the affected
648 paths for a commit are encouraged to do so.
652 Included in a `commit` command to add a new note (annotating a given
653 commit) or change the content of an existing note. This command has
654 two different means of specifying the content of the note.
656 External data format::
657 The data content for the note was already supplied by a prior
658 `blob` command. The frontend just needs to connect it to the
659 commit that is to be annotated.
662 'N' SP <dataref> SP <committish> LF
665 Here `<dataref>` can be either a mark reference (`:<idnum>`)
666 set by a prior `blob` command, or a full 40-byte SHA-1 of an
667 existing Git blob object.
670 The data content for the note has not been supplied yet.
671 The frontend wants to supply it as part of this modify
675 'N' SP 'inline' SP <committish> LF
679 See below for a detailed description of the `data` command.
681 In both formats `<committish>` is any of the commit specification
682 expressions also accepted by `from` (see above).
686 Arranges for fast-import to save a reference to the current object, allowing
687 the frontend to recall this object at a future point in time, without
688 knowing its SHA-1. Here the current object is the object creation
689 command the `mark` command appears within. This can be `commit`,
690 `tag`, and `blob`, but `commit` is the most common usage.
693 'mark' SP ':' <idnum> LF
696 where `<idnum>` is the number assigned by the frontend to this mark.
697 The value of `<idnum>` is expressed as an ASCII decimal integer.
698 The value 0 is reserved and cannot be used as
699 a mark. Only values greater than or equal to 1 may be used as marks.
701 New marks are created automatically. Existing marks can be moved
702 to another object simply by reusing the same `<idnum>` in another
707 Creates an annotated tag referring to a specific commit. To create
708 lightweight (non-annotated) tags see the `reset` command below.
712 'from' SP <committish> LF
713 'tagger' (SP <name>)? SP LT <email> GT SP <when> LF
717 where `<name>` is the name of the tag to create.
719 Tag names are automatically prefixed with `refs/tags/` when stored
720 in Git, so importing the CVS branch symbol `RELENG-1_0-FINAL` would
721 use just `RELENG-1_0-FINAL` for `<name>`, and fast-import will write the
722 corresponding ref as `refs/tags/RELENG-1_0-FINAL`.
724 The value of `<name>` must be a valid refname in Git and therefore
725 may contain forward slashes. As `LF` is not valid in a Git refname,
726 no quoting or escaping syntax is supported here.
728 The `from` command is the same as in the `commit` command; see
731 The `tagger` command uses the same format as `committer` within
732 `commit`; again see above for details.
734 The `data` command following `tagger` must supply the annotated tag
735 message (see below for `data` command syntax). To import an empty
736 tag message use a 0 length data. Tag messages are free-form and are
737 not interpreted by Git. Currently they must be encoded in UTF-8,
738 as fast-import does not permit other encodings to be specified.
740 Signing annotated tags during import from within fast-import is not
741 supported. Trying to include your own PGP/GPG signature is not
742 recommended, as the frontend does not (easily) have access to the
743 complete set of bytes which normally goes into such a signature.
744 If signing is required, create lightweight tags from within fast-import with
745 `reset`, then create the annotated versions of those tags offline
746 with the standard 'git tag' process.
750 Creates (or recreates) the named branch, optionally starting from
751 a specific revision. The reset command allows a frontend to issue
752 a new `from` command for an existing branch, or to create a new
753 branch from an existing commit without creating a new commit.
757 ('from' SP <committish> LF)?
761 For a detailed description of `<ref>` and `<committish>` see above
762 under `commit` and `from`.
764 The `LF` after the command is optional (it used to be required).
766 The `reset` command can also be used to create lightweight
767 (non-annotated) tags. For example:
774 would create the lightweight tag `refs/tags/938` referring to
775 whatever commit mark `:938` references.
779 Requests writing one file revision to the packfile. The revision
780 is not connected to any commit; this connection must be formed in
781 a subsequent `commit` command by referencing the blob through an
790 The mark command is optional here as some frontends have chosen
791 to generate the Git SHA-1 for the blob on their own, and feed that
792 directly to `commit`. This is typically more work than it's worth
793 however, as marks are inexpensive to store and easy to use.
797 Supplies raw data (for use as blob/file content, commit messages, or
798 annotated tag messages) to fast-import. Data can be supplied using an exact
799 byte count or delimited with a terminating line. Real frontends
800 intended for production-quality conversions should always use the
801 exact byte count format, as it is more robust and performs better.
802 The delimited format is intended primarily for testing fast-import.
804 Comment lines appearing within the `<raw>` part of `data` commands
805 are always taken to be part of the body of the data and are therefore
806 never ignored by fast-import. This makes it safe to import any
807 file/message content whose lines might start with `#`.
809 Exact byte count format::
810 The frontend must specify the number of bytes of data.
817 where `<count>` is the exact number of bytes appearing within
818 `<raw>`. The value of `<count>` is expressed as an ASCII decimal
819 integer. The `LF` on either side of `<raw>` is not
820 included in `<count>` and will not be included in the imported data.
822 The `LF` after `<raw>` is optional (it used to be required) but
823 recommended. Always including it makes debugging a fast-import
824 stream easier as the next command always starts in column 0
825 of the next line, even if `<raw>` did not end with an `LF`.
828 A delimiter string is used to mark the end of the data.
829 fast-import will compute the length by searching for the delimiter.
830 This format is primarily useful for testing and is not
831 recommended for real data.
834 'data' SP '<<' <delim> LF
840 where `<delim>` is the chosen delimiter string. The string `<delim>`
841 must not appear on a line by itself within `<raw>`, as otherwise
842 fast-import will think the data ends earlier than it really does. The `LF`
843 immediately trailing `<raw>` is part of `<raw>`. This is one of
844 the limitations of the delimited format, it is impossible to supply
845 a data chunk which does not have an LF as its last byte.
847 The `LF` after `<delim> LF` is optional (it used to be required).
851 Forces fast-import to close the current packfile, start a new one, and to
852 save out all current branch refs, tags and marks.
859 Note that fast-import automatically switches packfiles when the current
860 packfile reaches \--max-pack-size, or 4 GiB, whichever limit is
861 smaller. During an automatic packfile switch fast-import does not update
862 the branch refs, tags or marks.
864 As a `checkpoint` can require a significant amount of CPU time and
865 disk IO (to compute the overall pack SHA-1 checksum, generate the
866 corresponding index file, and update the refs) it can easily take
867 several minutes for a single `checkpoint` command to complete.
869 Frontends may choose to issue checkpoints during extremely large
870 and long running imports, or when they need to allow another Git
871 process access to a branch. However given that a 30 GiB Subversion
872 repository can be loaded into Git through fast-import in about 3 hours,
873 explicit checkpointing may not be necessary.
875 The `LF` after the command is optional (it used to be required).
879 Causes fast-import to print the entire `progress` line unmodified to
880 its standard output channel (file descriptor 1) when the command is
881 processed from the input stream. The command otherwise has no impact
882 on the current import, or on any of fast-import's internal state.
885 'progress' SP <any> LF
889 The `<any>` part of the command may contain any sequence of bytes
890 that does not contain `LF`. The `LF` after the command is optional.
891 Callers may wish to process the output through a tool such as sed to
892 remove the leading part of the line, for example:
895 frontend | git fast-import | sed 's/^progress //'
898 Placing a `progress` command immediately after a `checkpoint` will
899 inform the reader when the `checkpoint` has been completed and it
900 can safely access the refs that fast-import updated.
904 Causes fast-import to print a blob to a file descriptor previously
905 arranged with the `--cat-blob-fd` argument. The command otherwise
906 has no impact on the current import; its main purpose is to
907 retrieve blobs that may be in fast-import's memory but not
908 accessible from the target repository.
911 'cat-blob' SP <dataref> LF
914 The `<dataref>` can be either a mark reference (`:<idnum>`)
915 set previously or a full 40-byte SHA-1 of a Git blob, preexisting or
918 Output uses the same format as `git cat-file --batch`:
921 <sha1> SP 'blob' SP <size> LF
925 This command can be used anywhere in the stream that comments are
926 accepted. In particular, the `cat-blob` command can be used in the
927 middle of a commit but not in the middle of a `data` command.
931 Prints information about the object at a path to a file descriptor
932 previously arranged with the `--cat-blob-fd` argument. This allows
933 printing a blob from the active commit (with `cat-blob`) or copying a
934 blob or tree from a previous commit for use in the current one (with
937 The `ls` command can be used anywhere in the stream that comments are
938 accepted, including the middle of a commit.
940 Reading from the active commit::
941 This form can only be used in the middle of a `commit`.
942 The path names a directory entry within fast-import's
943 active commit. The path must be quoted in this case.
949 Reading from a named tree::
950 The `<dataref>` can be a mark reference (`:<idnum>`) or the
951 full 40-byte SHA-1 of a Git tag, commit, or tree object,
952 preexisting or waiting to be written.
953 The path is relative to the top level of the tree
954 named by `<dataref>`.
957 'ls' SP <dataref> SP <path> LF
960 See `filemodify` above for a detailed description of `<path>`.
962 Output uses the same format as `git ls-tree <tree> {litdd} <path>`:
965 <mode> SP ('blob' | 'tree' | 'commit') SP <dataref> HT <path> LF
968 The <dataref> represents the blob, tree, or commit object at <path>
969 and can be used in later 'cat-blob', 'filemodify', or 'ls' commands.
971 If there is no file or subtree at that path, 'git fast-import' will
980 Require that fast-import supports the specified feature, or abort if
984 'feature' SP <feature> ('=' <argument>)? LF
987 The <feature> part of the command may be any one of the following:
994 Act as though the corresponding command-line option with
995 a leading '--' was passed on the command line
996 (see OPTIONS, above).
999 Like --import-marks except in two respects: first, only one
1000 "feature import-marks" command is allowed per stream;
1001 second, an --import-marks= command-line option overrides
1002 any "feature import-marks" command in the stream.
1006 Require that the backend support the 'cat-blob' or 'ls' command.
1007 Versions of fast-import not supporting the specified command
1008 will exit with a message indicating so.
1009 This lets the import error out early with a clear message,
1010 rather than wasting time on the early part of an import
1011 before the unsupported command is detected.
1014 Require that the backend support the 'notemodify' (N)
1015 subcommand to the 'commit' command.
1016 Versions of fast-import not supporting notes will exit
1017 with a message indicating so.
1022 Processes the specified option so that git fast-import behaves in a
1023 way that suits the frontend's needs.
1024 Note that options specified by the frontend are overridden by any
1025 options the user may specify to git fast-import itself.
1028 'option' SP <option> LF
1031 The `<option>` part of the command may contain any of the options
1032 listed in the OPTIONS section that do not change import semantics,
1033 without the leading '--' and is treated in the same way.
1035 Option commands must be the first commands on the input (not counting
1036 feature commands), to give an option command after any non-option
1037 command is an error.
1039 The following commandline options change import semantics and may therefore
1040 not be passed as option:
1050 If fast-import is supplied invalid input it will terminate with a
1051 non-zero exit status and create a crash report in the top level of
1052 the Git repository it was importing into. Crash reports contain
1053 a snapshot of the internal fast-import state as well as the most
1054 recent commands that lead up to the crash.
1056 All recent commands (including stream comments, file changes and
1057 progress commands) are shown in the command history within the crash
1058 report, but raw file data and commit messages are excluded from the
1059 crash report. This exclusion saves space within the report file
1060 and reduces the amount of buffering that fast-import must perform
1063 After writing a crash report fast-import will close the current
1064 packfile and export the marks table. This allows the frontend
1065 developer to inspect the repository state and resume the import from
1066 the point where it crashed. The modified branches and tags are not
1067 updated during a crash, as the import did not complete successfully.
1068 Branch and tag information can be found in the crash report and
1069 must be applied manually if the update is needed.
1074 $ cat >in <<END_OF_INPUT
1075 # my very first test commit
1076 commit refs/heads/master
1077 committer Shawn O. Pearce <spearce> 19283 -0400
1078 # who is that guy anyway?
1082 M 644 inline .gitignore
1089 $ git fast-import <in
1090 fatal: Corrupt mode: M 777 inline bob
1091 fast-import: dumping crash report to .git/fast_import_crash_8434
1093 $ cat .git/fast_import_crash_8434
1094 fast-import crash report:
1095 fast-import process: 8434
1096 parent process : 1391
1097 at Sat Sep 1 00:58:12 2007
1099 fatal: Corrupt mode: M 777 inline bob
1101 Most Recent Commands Before Crash
1102 ---------------------------------
1103 # my very first test commit
1104 commit refs/heads/master
1105 committer Shawn O. Pearce <spearce> 19283 -0400
1106 # who is that guy anyway?
1108 M 644 inline .gitignore
1114 active_branches = 1 cur, 5 max
1117 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1118 1) 0 refs/heads/master
1123 status : active loaded dirty
1124 tip commit : 0000000000000000000000000000000000000000
1125 old tree : 0000000000000000000000000000000000000000
1126 cur tree : 0000000000000000000000000000000000000000
1137 The following tips and tricks have been collected from various
1138 users of fast-import, and are offered here as suggestions.
1140 Use One Mark Per Commit
1141 ~~~~~~~~~~~~~~~~~~~~~~~
1142 When doing a repository conversion, use a unique mark per commit
1143 (`mark :<n>`) and supply the \--export-marks option on the command
1144 line. fast-import will dump a file which lists every mark and the Git
1145 object SHA-1 that corresponds to it. If the frontend can tie
1146 the marks back to the source repository, it is easy to verify the
1147 accuracy and completeness of the import by comparing each Git
1148 commit to the corresponding source revision.
1150 Coming from a system such as Perforce or Subversion this should be
1151 quite simple, as the fast-import mark can also be the Perforce changeset
1152 number or the Subversion revision number.
1154 Freely Skip Around Branches
1155 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
1156 Don't bother trying to optimize the frontend to stick to one branch
1157 at a time during an import. Although doing so might be slightly
1158 faster for fast-import, it tends to increase the complexity of the frontend
1161 The branch LRU builtin to fast-import tends to behave very well, and the
1162 cost of activating an inactive branch is so low that bouncing around
1163 between branches has virtually no impact on import performance.
1167 When importing a renamed file or directory, simply delete the old
1168 name(s) and modify the new name(s) during the corresponding commit.
1169 Git performs rename detection after-the-fact, rather than explicitly
1172 Use Tag Fixup Branches
1173 ~~~~~~~~~~~~~~~~~~~~~~
1174 Some other SCM systems let the user create a tag from multiple
1175 files which are not from the same commit/changeset. Or to create
1176 tags which are a subset of the files available in the repository.
1178 Importing these tags as-is in Git is impossible without making at
1179 least one commit which ``fixes up'' the files to match the content
1180 of the tag. Use fast-import's `reset` command to reset a dummy branch
1181 outside of your normal branch space to the base commit for the tag,
1182 then commit one or more file fixup commits, and finally tag the
1185 For example since all normal branches are stored under `refs/heads/`
1186 name the tag fixup branch `TAG_FIXUP`. This way it is impossible for
1187 the fixup branch used by the importer to have namespace conflicts
1188 with real branches imported from the source (the name `TAG_FIXUP`
1189 is not `refs/heads/TAG_FIXUP`).
1191 When committing fixups, consider using `merge` to connect the
1192 commit(s) which are supplying file revisions to the fixup branch.
1193 Doing so will allow tools such as 'git blame' to track
1194 through the real commit history and properly annotate the source
1197 After fast-import terminates the frontend will need to do `rm .git/TAG_FIXUP`
1198 to remove the dummy branch.
1200 Import Now, Repack Later
1201 ~~~~~~~~~~~~~~~~~~~~~~~~
1202 As soon as fast-import completes the Git repository is completely valid
1203 and ready for use. Typically this takes only a very short time,
1204 even for considerably large projects (100,000+ commits).
1206 However repacking the repository is necessary to improve data
1207 locality and access performance. It can also take hours on extremely
1208 large projects (especially if -f and a large \--window parameter is
1209 used). Since repacking is safe to run alongside readers and writers,
1210 run the repack in the background and let it finish when it finishes.
1211 There is no reason to wait to explore your new Git project!
1213 If you choose to wait for the repack, don't try to run benchmarks
1214 or performance tests until repacking is completed. fast-import outputs
1215 suboptimal packfiles that are simply never seen in real use
1218 Repacking Historical Data
1219 ~~~~~~~~~~~~~~~~~~~~~~~~~
1220 If you are repacking very old imported data (e.g. older than the
1221 last year), consider expending some extra CPU time and supplying
1222 \--window=50 (or higher) when you run 'git repack'.
1223 This will take longer, but will also produce a smaller packfile.
1224 You only need to expend the effort once, and everyone using your
1225 project will benefit from the smaller repository.
1227 Include Some Progress Messages
1228 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1229 Every once in a while have your frontend emit a `progress` message
1230 to fast-import. The contents of the messages are entirely free-form,
1231 so one suggestion would be to output the current month and year
1232 each time the current commit date moves into the next month.
1233 Your users will feel better knowing how much of the data stream
1237 Packfile Optimization
1238 ---------------------
1239 When packing a blob fast-import always attempts to deltify against the last
1240 blob written. Unless specifically arranged for by the frontend,
1241 this will probably not be a prior version of the same file, so the
1242 generated delta will not be the smallest possible. The resulting
1243 packfile will be compressed, but will not be optimal.
1245 Frontends which have efficient access to all revisions of a
1246 single file (for example reading an RCS/CVS ,v file) can choose
1247 to supply all revisions of that file as a sequence of consecutive
1248 `blob` commands. This allows fast-import to deltify the different file
1249 revisions against each other, saving space in the final packfile.
1250 Marks can be used to later identify individual file revisions during
1251 a sequence of `commit` commands.
1253 The packfile(s) created by fast-import do not encourage good disk access
1254 patterns. This is caused by fast-import writing the data in the order
1255 it is received on standard input, while Git typically organizes
1256 data within packfiles to make the most recent (current tip) data
1257 appear before historical data. Git also clusters commits together,
1258 speeding up revision traversal through better cache locality.
1260 For this reason it is strongly recommended that users repack the
1261 repository with `git repack -a -d` after fast-import completes, allowing
1262 Git to reorganize the packfiles for faster data access. If blob
1263 deltas are suboptimal (see above) then also adding the `-f` option
1264 to force recomputation of all deltas can significantly reduce the
1265 final packfile size (30-50% smaller can be quite typical).
1270 There are a number of factors which affect how much memory fast-import
1271 requires to perform an import. Like critical sections of core
1272 Git, fast-import uses its own memory allocators to amortize any overheads
1273 associated with malloc. In practice fast-import tends to amortize any
1274 malloc overheads to 0, due to its use of large block allocations.
1278 fast-import maintains an in-memory structure for every object written in
1279 this execution. On a 32 bit system the structure is 32 bytes,
1280 on a 64 bit system the structure is 40 bytes (due to the larger
1281 pointer sizes). Objects in the table are not deallocated until
1282 fast-import terminates. Importing 2 million objects on a 32 bit system
1283 will require approximately 64 MiB of memory.
1285 The object table is actually a hashtable keyed on the object name
1286 (the unique SHA-1). This storage configuration allows fast-import to reuse
1287 an existing or already written object and avoid writing duplicates
1288 to the output packfile. Duplicate blobs are surprisingly common
1289 in an import, typically due to branch merges in the source.
1293 Marks are stored in a sparse array, using 1 pointer (4 bytes or 8
1294 bytes, depending on pointer size) per mark. Although the array
1295 is sparse, frontends are still strongly encouraged to use marks
1296 between 1 and n, where n is the total number of marks required for
1301 Branches are classified as active and inactive. The memory usage
1302 of the two classes is significantly different.
1304 Inactive branches are stored in a structure which uses 96 or 120
1305 bytes (32 bit or 64 bit systems, respectively), plus the length of
1306 the branch name (typically under 200 bytes), per branch. fast-import will
1307 easily handle as many as 10,000 inactive branches in under 2 MiB
1310 Active branches have the same overhead as inactive branches, but
1311 also contain copies of every tree that has been recently modified on
1312 that branch. If subtree `include` has not been modified since the
1313 branch became active, its contents will not be loaded into memory,
1314 but if subtree `src` has been modified by a commit since the branch
1315 became active, then its contents will be loaded in memory.
1317 As active branches store metadata about the files contained on that
1318 branch, their in-memory storage size can grow to a considerable size
1321 fast-import automatically moves active branches to inactive status based on
1322 a simple least-recently-used algorithm. The LRU chain is updated on
1323 each `commit` command. The maximum number of active branches can be
1324 increased or decreased on the command line with \--active-branches=.
1328 Trees (aka directories) use just 12 bytes of memory on top of the
1329 memory required for their entries (see ``per active file'' below).
1330 The cost of a tree is virtually 0, as its overhead amortizes out
1331 over the individual file entries.
1333 per active file entry
1334 ~~~~~~~~~~~~~~~~~~~~~
1335 Files (and pointers to subtrees) within active trees require 52 or 64
1336 bytes (32/64 bit platforms) per entry. To conserve space, file and
1337 tree names are pooled in a common string table, allowing the filename
1338 ``Makefile'' to use just 16 bytes (after including the string header
1339 overhead) no matter how many times it occurs within the project.
1341 The active branch LRU, when coupled with the filename string pool
1342 and lazy loading of subtrees, allows fast-import to efficiently import
1343 projects with 2,000+ branches and 45,114+ files in a very limited
1344 memory footprint (less than 2.7 MiB per active branch).
1348 Sending *SIGUSR1* to the 'git fast-import' process ends the current
1349 packfile early, simulating a `checkpoint` command. The impatient
1350 operator can use this facility to peek at the objects and refs from an
1351 import in progress, at the cost of some added running time and worse
1356 Part of the linkgit:git[1] suite
projects / git / blob