1GIT-FAST-IMPORT(1) Git Manual GIT-FAST-IMPORT(1)
2
6 git-fast-import - Backend for fast Git data importers
7
9 frontend | git fast-import [<options>]
10
13 This program is usually not what the end user wants to run directly.
14 Most end users want to use one of the existing frontend programs, which
15 parses a specific type of foreign source and feeds the contents stored
16 there to git fast-import.
17 fast-import reads a mixed command/data stream from standard input and
19 writes one or more packfiles directly into the current repository. When
20 EOF is received on standard input, fast import writes out updated
21 branch and tag refs, fully updating the current repository with the
22 newly imported data.
23 The fast-import backend itself can import into an empty repository (one
25 that has already been initialized by git init) or incrementally update
26 an existing populated repository. Whether or not incremental imports
27 are supported from a particular foreign source depends on the frontend
28 program in use.
29
31 --force
32 Force updating modified existing branches, even if doing so would
33 cause commits to be lost (as the new commit does not contain the
34 old commit).
35 --quiet
37 Disable the output shown by --stats, making fast-import usually be
38 silent when it is successful. However, if the import stream has
39 directives intended to show user output (e.g. progress
40 directives), the corresponding messages will still be shown.
41 --stats
43 Display some basic statistics about the objects fast-import has
44 created, the packfiles they were stored into, and the memory used
45 by fast-import during this run. Showing this output is currently
46 the default, but can be disabled with --quiet.
47 Options for Frontends
49 --cat-blob-fd=<fd>
50 Write responses to get-mark, cat-blob, and ls queries to the file
51 descriptor <fd> instead of stdout. Allows progress output intended
52 for the end-user to be separated from other output.
53 --date-format=<fmt>
55 Specify the type of dates the frontend will supply to fast-import
56 within author, committer and tagger commands. See “Date Formats”
57 below for details about which formats are supported, and their
58 syntax.
59 --done
61 Terminate with error if there is no done command at the end of the
62 stream. This option might be useful for detecting errors that cause
63 the frontend to terminate before it has started to write a stream.
64 Locations of Marks Files
66 --export-marks=<file>
67 Dumps the internal marks table to <file> when complete. Marks are
68 written one per line as :markid SHA-1. Frontends can use this file
69 to validate imports after they have been completed, or to save the
70 marks table across incremental runs. As <file> is only opened and
71 truncated at checkpoint (or completion) the same path can also be
72 safely given to --import-marks.
73 --import-marks=<file>
75 Before processing any input, load the marks specified in <file>.
76 The input file must exist, must be readable, and must use the same
77 format as produced by --export-marks. Multiple options may be
78 supplied to import more than one set of marks. If a mark is defined
79 to different values, the last file wins.
80 --import-marks-if-exists=<file>
82 Like --import-marks but instead of erroring out, silently skips the
83 file if it does not exist.
84 --[no-]relative-marks
86 After specifying --relative-marks the paths specified with
87 --import-marks= and --export-marks= are relative to an internal
88 directory in the current repository. In git-fast-import this means
89 that the paths are relative to the .git/info/fast-import directory.
90 However, other importers may use a different location.
91 Relative and non-relative marks may be combined by interweaving
93 --(no-)-relative-marks with the --(import|export)-marks= options.
94 Performance and Compression Tuning
96 --active-branches=<n>
97 Maximum number of branches to maintain active at once. See “Memory
98 Utilization” below for details. Default is 5.
99 --big-file-threshold=<n>
101 Maximum size of a blob that fast-import will attempt to create a
102 delta for, expressed in bytes. The default is 512m (512 MiB). Some
103 importers may wish to lower this on systems with constrained
104 memory.
105 --depth=<n>
107 Maximum delta depth, for blob and tree deltification. Default is
108 50.
109 --export-pack-edges=<file>
111 After creating a packfile, print a line of data to <file> listing
112 the filename of the packfile and the last commit on each branch
113 that was written to that packfile. This information may be useful
114 after importing projects whose total object set exceeds the 4 GiB
115 packfile limit, as these commits can be used as edge points during
116 calls to git pack-objects.
117 --max-pack-size=<n>
119 Maximum size of each output packfile. The default is unlimited.
120 fastimport.unpackLimit
122 See git-config(1)
123
125 The design of fast-import allows it to import large projects in a
126 minimum amount of memory usage and processing time. Assuming the
127 frontend is able to keep up with fast-import and feed it a constant
128 stream of data, import times for projects holding 10+ years of history
129 and containing 100,000+ individual commits are generally completed in
130 just 1-2 hours on quite modest (~$2,000 USD) hardware.
131 Most bottlenecks appear to be in foreign source data access (the source
133 just cannot extract revisions fast enough) or disk IO (fast-import
134 writes as fast as the disk will take the data). Imports will run faster
135 if the source data is stored on a different drive than the destination
136 Git repository (due to less IO contention).
137
139 A typical frontend for fast-import tends to weigh in at approximately
140 200 lines of Perl/Python/Ruby code. Most developers have been able to
141 create working importers in just a couple of hours, even though it is
142 their first exposure to fast-import, and sometimes even to Git. This is
143 an ideal situation, given that most conversion tools are throw-away
144 (use once, and never look back).
145
147 Like git push or git fetch, imports handled by fast-import are safe to
148 run alongside parallel git repack -a -d or git gc invocations, or any
149 other Git operation (including git prune, as loose objects are never
150 used by fast-import).
151 fast-import does not lock the branch or tag refs it is actively
153 importing. After the import, during its ref update phase, fast-import
154 tests each existing branch ref to verify the update will be a
155 fast-forward update (the commit stored in the ref is contained in the
156 new history of the commit to be written). If the update is not a
157 fast-forward update, fast-import will skip updating that ref and
158 instead prints a warning message. fast-import will always attempt to
159 update all branch refs, and does not stop on the first failure.
160 Branch updates can be forced with --force, but it’s recommended that
162 this only be used on an otherwise quiet repository. Using --force is
163 not necessary for an initial import into an empty repository.
164
166 fast-import tracks a set of branches in memory. Any branch can be
167 created or modified at any point during the import process by sending a
168 commit command on the input stream. This design allows a frontend
169 program to process an unlimited number of branches simultaneously,
170 generating commits in the order they are available from the source
171 data. It also simplifies the frontend programs considerably.
172 fast-import does not use or alter the current working directory, or any
174 file within it. (It does however update the current Git repository, as
175 referenced by GIT_DIR.) Therefore an import frontend may use the
176 working directory for its own purposes, such as extracting file
177 revisions from the foreign source. This ignorance of the working
178 directory also allows fast-import to run very quickly, as it does not
179 need to perform any costly file update operations when switching
180 between branches.
181
183 With the exception of raw file data (which Git does not interpret) the
184 fast-import input format is text (ASCII) based. This text based format
185 simplifies development and debugging of frontend programs, especially
186 when a higher level language such as Perl, Python or Ruby is being
187 used.
188 fast-import is very strict about its input. Where we say SP below we
190 mean exactly one space. Likewise LF means one (and only one) linefeed
191 and HT one (and only one) horizontal tab. Supplying additional
192 whitespace characters will cause unexpected results, such as branch
193 names or file names with leading or trailing spaces in their name, or
194 early termination of fast-import when it encounters unexpected input.
195 Stream Comments
197 To aid in debugging frontends fast-import ignores any line that begins
198 with # (ASCII pound/hash) up to and including the line ending LF. A
199 comment line may contain any sequence of bytes that does not contain an
200 LF and therefore may be used to include any detailed debugging
201 information that might be specific to the frontend and useful when
202 inspecting a fast-import data stream.
203 Date Formats
205 The following date formats are supported. A frontend should select the
206 format it will use for this import by passing the format name in the
207 --date-format=<fmt> command-line option.
208 raw
210 This is the Git native format and is <time> SP <offutc>. It is also
211 fast-import’s default format, if --date-format was not specified.
212 The time of the event is specified by <time> as the number of
214 seconds since the UNIX epoch (midnight, Jan 1, 1970, UTC) and is
215 written as an ASCII decimal integer.
216 The local offset is specified by <offutc> as a positive or negative
218 offset from UTC. For example EST (which is 5 hours behind UTC)
219 would be expressed in <tz> by “-0500” while UTC is “+0000”. The
220 local offset does not affect <time>; it is used only as an
221 advisement to help formatting routines display the timestamp.
222 If the local offset is not available in the source material, use
224 “+0000”, or the most common local offset. For example many
225 organizations have a CVS repository which has only ever been
226 accessed by users who are located in the same location and time
227 zone. In this case a reasonable offset from UTC could be assumed.
228 Unlike the rfc2822 format, this format is very strict. Any
230 variation in formatting will cause fast-import to reject the value.
231 rfc2822
233 This is the standard email format as described by RFC 2822.
234 An example value is “Tue Feb 6 11:22:18 2007 -0500”. The Git parser
236 is accurate, but a little on the lenient side. It is the same
237 parser used by git am when applying patches received from email.
238 Some malformed strings may be accepted as valid dates. In some of
240 these cases Git will still be able to obtain the correct date from
241 the malformed string. There are also some types of malformed
242 strings which Git will parse wrong, and yet consider valid.
243 Seriously malformed strings will be rejected.
244 Unlike the raw format above, the time zone/UTC offset information
246 contained in an RFC 2822 date string is used to adjust the date
247 value to UTC prior to storage. Therefore it is important that this
248 information be as accurate as possible.
249 If the source material uses RFC 2822 style dates, the frontend
251 should let fast-import handle the parsing and conversion (rather
252 than attempting to do it itself) as the Git parser has been well
253 tested in the wild.
254 Frontends should prefer the raw format if the source material
256 already uses UNIX-epoch format, can be coaxed to give dates in that
257 format, or its format is easily convertible to it, as there is no
258 ambiguity in parsing.
259 now
261 Always use the current time and time zone. The literal now must
262 always be supplied for <when>.
263 This is a toy format. The current time and time zone of this system
265 is always copied into the identity string at the time it is being
266 created by fast-import. There is no way to specify a different time
267 or time zone.
268 This particular format is supplied as it’s short to implement and
270 may be useful to a process that wants to create a new commit right
271 now, without needing to use a working directory or git
272 update-index.
273 If separate author and committer commands are used in a commit the
275 timestamps may not match, as the system clock will be polled twice
276 (once for each command). The only way to ensure that both author
277 and committer identity information has the same timestamp is to
278 omit author (thus copying from committer) or to use a date format
279 other than now.
280 Commands
282 fast-import accepts several commands to update the current repository
283 and control the current import process. More detailed discussion (with
284 examples) of each command follows later.
285 commit
287 Creates a new branch or updates an existing branch by creating a
288 new commit and updating the branch to point at the newly created
289 commit.
290 tag
292 Creates an annotated tag object from an existing commit or branch.
293 Lightweight tags are not supported by this command, as they are not
294 recommended for recording meaningful points in time.
295 reset
297 Reset an existing branch (or a new branch) to a specific revision.
298 This command must be used to change a branch to a specific revision
299 without making a commit on it.
300 blob
302 Convert raw file data into a blob, for future use in a commit
303 command. This command is optional and is not needed to perform an
304 import.
305 checkpoint
307 Forces fast-import to close the current packfile, generate its
308 unique SHA-1 checksum and index, and start a new packfile. This
309 command is optional and is not needed to perform an import.
310 progress
312 Causes fast-import to echo the entire line to its own standard
313 output. This command is optional and is not needed to perform an
314 import.
315 done
317 Marks the end of the stream. This command is optional unless the
318 done feature was requested using the --done command-line option or
319 feature done command.
320 get-mark
322 Causes fast-import to print the SHA-1 corresponding to a mark to
323 the file descriptor set with --cat-blob-fd, or stdout if
324 unspecified.
325 cat-blob
327 Causes fast-import to print a blob in cat-file --batch format to
328 the file descriptor set with --cat-blob-fd or stdout if
329 unspecified.
330 ls
332 Causes fast-import to print a line describing a directory entry in
333 ls-tree format to the file descriptor set with --cat-blob-fd or
334 stdout if unspecified.
335 feature
337 Enable the specified feature. This requires that fast-import
338 supports the specified feature, and aborts if it does not.
339 option
341 Specify any of the options listed under OPTIONS that do not change
342 stream semantic to suit the frontend’s needs. This command is
343 optional and is not needed to perform an import.
344 commit
346 Create or update a branch with a new commit, recording one logical
347 change to the project.
348 'commit' SP <ref> LF
350 mark?
351 original-oid?
352 ('author' (SP <name>)? SP LT <email> GT SP <when> LF)?
353 'committer' (SP <name>)? SP LT <email> GT SP <when> LF
354 data
355 ('from' SP <commit-ish> LF)?
356 ('merge' SP <commit-ish> LF)?
357 (filemodify | filedelete | filecopy | filerename | filedeleteall | notemodify)*
358 LF?
359 where <ref> is the name of the branch to make the commit on. Typically
361 branch names are prefixed with refs/heads/ in Git, so importing the CVS
362 branch symbol RELENG-1_0 would use refs/heads/RELENG-1_0 for the value
363 of <ref>. The value of <ref> must be a valid refname in Git. As LF is
364 not valid in a Git refname, no quoting or escaping syntax is supported
365 here.
366 A mark command may optionally appear, requesting fast-import to save a
368 reference to the newly created commit for future use by the frontend
369 (see below for format). It is very common for frontends to mark every
370 commit they create, thereby allowing future branch creation from any
371 imported commit.
372 The data command following committer must supply the commit message
374 (see below for data command syntax). To import an empty commit message
375 use a 0 length data. Commit messages are free-form and are not
376 interpreted by Git. Currently they must be encoded in UTF-8, as
377 fast-import does not permit other encodings to be specified.
378 Zero or more filemodify, filedelete, filecopy, filerename,
380 filedeleteall and notemodify commands may be included to update the
381 contents of the branch prior to creating the commit. These commands may
382 be supplied in any order. However it is recommended that a
383 filedeleteall command precede all filemodify, filecopy, filerename and
384 notemodify commands in the same commit, as filedeleteall wipes the
385 branch clean (see below).
386 The LF after the command is optional (it used to be required).
388 author
390 An author command may optionally appear, if the author information
391 might differ from the committer information. If author is omitted
392 then fast-import will automatically use the committer’s information
393 for the author portion of the commit. See below for a description
394 of the fields in author, as they are identical to committer.
395 committer
397 The committer command indicates who made this commit, and when they
398 made it.
399 Here <name> is the person’s display name (for example “Com M
401 Itter”) and <email> is the person’s email address
402 (“cm@example.com”). LT and GT are the literal less-than (\x3c) and
403 greater-than (\x3e) symbols. These are required to delimit the
404 email address from the other fields in the line. Note that <name>
405 and <email> are free-form and may contain any sequence of bytes,
406 except LT, GT and LF. <name> is typically UTF-8 encoded.
407 The time of the change is specified by <when> using the date format
409 that was selected by the --date-format=<fmt> command-line option.
410 See “Date Formats” above for the set of supported formats, and
411 their syntax.
412 from
414 The from command is used to specify the commit to initialize this
415 branch from. This revision will be the first ancestor of the new
416 commit. The state of the tree built at this commit will begin with
417 the state at the from commit, and be altered by the content
418 modifications in this commit.
419 Omitting the from command in the first commit of a new branch will
421 cause fast-import to create that commit with no ancestor. This
422 tends to be desired only for the initial commit of a project. If
423 the frontend creates all files from scratch when making a new
424 branch, a merge command may be used instead of from to start the
425 commit with an empty tree. Omitting the from command on existing
426 branches is usually desired, as the current commit on that branch
427 is automatically assumed to be the first ancestor of the new
428 commit.
429 As LF is not valid in a Git refname or SHA-1 expression, no quoting
431 or escaping syntax is supported within <commit-ish>.
432 Here <commit-ish> is any of the following:
434 · The name of an existing branch already in fast-import’s
436 internal branch table. If fast-import doesn’t know the name,
437 it’s treated as a SHA-1 expression.
438 · A mark reference, :<idnum>, where <idnum> is the mark number.
440 The reason fast-import uses : to denote a mark reference is
442 this character is not legal in a Git branch name. The leading :
443 makes it easy to distinguish between the mark 42 (:42) and the
444 branch 42 (42 or refs/heads/42), or an abbreviated SHA-1 which
445 happened to consist only of base-10 digits.
446 Marks must be declared (via mark) before they can be used.
448 · A complete 40 byte or abbreviated commit SHA-1 in hex.
450 · Any valid Git SHA-1 expression that resolves to a commit. See
452 “SPECIFYING REVISIONS” in gitrevisions(7) for details.
453 · The special null SHA-1 (40 zeros) specifies that the branch is
455 to be removed.
456 The special case of restarting an incremental import from the
458 current branch value should be written as:
459 from refs/heads/branch^0
461 The ^0 suffix is necessary as fast-import does not permit a branch
464 to start from itself, and the branch is created in memory before
465 the from command is even read from the input. Adding ^0 will force
466 fast-import to resolve the commit through Git’s revision parsing
467 library, rather than its internal branch table, thereby loading in
468 the existing value of the branch.
469 merge
471 Includes one additional ancestor commit. The additional ancestry
472 link does not change the way the tree state is built at this
473 commit. If the from command is omitted when creating a new branch,
474 the first merge commit will be the first ancestor of the current
475 commit, and the branch will start out with no files. An unlimited
476 number of merge commands per commit are permitted by fast-import,
477 thereby establishing an n-way merge.
478 Here <commit-ish> is any of the commit specification expressions
480 also accepted by from (see above).
481 filemodify
483 Included in a commit command to add a new file or change the
484 content of an existing file. This command has two different means
485 of specifying the content of the file.
486 External data format
488 The data content for the file was already supplied by a prior
489 blob command. The frontend just needs to connect it.
490 'M' SP <mode> SP <dataref> SP <path> LF
492 Here usually <dataref> must be either a mark reference
494 (:<idnum>) set by a prior blob command, or a full 40-byte SHA-1
495 of an existing Git blob object. If <mode> is 040000` then
496 <dataref> must be the full 40-byte SHA-1 of an existing Git
497 tree object or a mark reference set with --import-marks.
498 Inline data format
500 The data content for the file has not been supplied yet. The
501 frontend wants to supply it as part of this modify command.
502 'M' SP <mode> SP 'inline' SP <path> LF
504 data
505 See below for a detailed description of the data command.
507 In both formats <mode> is the type of file entry, specified in
509 octal. Git only supports the following modes:
510 · 100644 or 644: A normal (not-executable) file. The majority of
512 files in most projects use this mode. If in doubt, this is what
513 you want.
514 · 100755 or 755: A normal, but executable, file.
516 · 120000: A symlink, the content of the file will be the link
518 target.
519 · 160000: A gitlink, SHA-1 of the object refers to a commit in
521 another repository. Git links can only be specified by SHA or
522 through a commit mark. They are used to implement submodules.
523 · 040000: A subdirectory. Subdirectories can only be specified by
525 SHA or through a tree mark set with --import-marks.
526 In both formats <path> is the complete path of the file to be added
528 (if not already existing) or modified (if already existing).
529 A <path> string must use UNIX-style directory separators (forward
531 slash /), may contain any byte other than LF, and must not start
532 with double quote (").
533 A path can use C-style string quoting; this is accepted in all
535 cases and mandatory if the filename starts with double quote or
536 contains LF. In C-style quoting, the complete name should be
537 surrounded with double quotes, and any LF, backslash, or double
538 quote characters must be escaped by preceding them with a backslash
539 (e.g., "path/with\n, \\ and \" in it").
540 The value of <path> must be in canonical form. That is it must not:
542 · contain an empty directory component (e.g. foo//bar is
544 invalid),
545 · end with a directory separator (e.g. foo/ is invalid),
547 · start with a directory separator (e.g. /foo is invalid),
549 · contain the special component . or .. (e.g. foo/./bar and
551 foo/../bar are invalid).
552 The root of the tree can be represented by an empty string as
554 <path>.
555 It is recommended that <path> always be encoded using UTF-8.
557 filedelete
559 Included in a commit command to remove a file or recursively delete
560 an entire directory from the branch. If the file or directory
561 removal makes its parent directory empty, the parent directory will
562 be automatically removed too. This cascades up the tree until the
563 first non-empty directory or the root is reached.
564 'D' SP <path> LF
566 here <path> is the complete path of the file or subdirectory to be
568 removed from the branch. See filemodify above for a detailed
569 description of <path>.
570 filecopy
572 Recursively copies an existing file or subdirectory to a different
573 location within the branch. The existing file or directory must
574 exist. If the destination exists it will be completely replaced by
575 the content copied from the source.
576 'C' SP <path> SP <path> LF
578 here the first <path> is the source location and the second <path>
580 is the destination. See filemodify above for a detailed description
581 of what <path> may look like. To use a source path that contains SP
582 the path must be quoted.
583 A filecopy command takes effect immediately. Once the source
585 location has been copied to the destination any future commands
586 applied to the source location will not impact the destination of
587 the copy.
588 filerename
590 Renames an existing file or subdirectory to a different location
591 within the branch. The existing file or directory must exist. If
592 the destination exists it will be replaced by the source directory.
593 'R' SP <path> SP <path> LF
595 here the first <path> is the source location and the second <path>
597 is the destination. See filemodify above for a detailed description
598 of what <path> may look like. To use a source path that contains SP
599 the path must be quoted.
600 A filerename command takes effect immediately. Once the source
602 location has been renamed to the destination any future commands
603 applied to the source location will create new files there and not
604 impact the destination of the rename.
605 Note that a filerename is the same as a filecopy followed by a
607 filedelete of the source location. There is a slight performance
608 advantage to using filerename, but the advantage is so small that
609 it is never worth trying to convert a delete/add pair in source
610 material into a rename for fast-import. This filerename command is
611 provided just to simplify frontends that already have rename
612 information and don’t want bother with decomposing it into a
613 filecopy followed by a filedelete.
614 filedeleteall
616 Included in a commit command to remove all files (and also all
617 directories) from the branch. This command resets the internal
618 branch structure to have no files in it, allowing the frontend to
619 subsequently add all interesting files from scratch.
620 'deleteall' LF
622 This command is extremely useful if the frontend does not know (or
624 does not care to know) what files are currently on the branch, and
625 therefore cannot generate the proper filedelete commands to update
626 the content.
627 Issuing a filedeleteall followed by the needed filemodify commands
629 to set the correct content will produce the same results as sending
630 only the needed filemodify and filedelete commands. The
631 filedeleteall approach may however require fast-import to use
632 slightly more memory per active branch (less than 1 MiB for even
633 most large projects); so frontends that can easily obtain only the
634 affected paths for a commit are encouraged to do so.
635 notemodify
637 Included in a commit <notes_ref> command to add a new note
638 annotating a <commit-ish> or change this annotation contents.
639 Internally it is similar to filemodify 100644 on <commit-ish> path
640 (maybe split into subdirectories). It’s not advised to use any
641 other commands to write to the <notes_ref> tree except
642 filedeleteall to delete all existing notes in this tree. This
643 command has two different means of specifying the content of the
644 note.
645 External data format
647 The data content for the note was already supplied by a prior
648 blob command. The frontend just needs to connect it to the
649 commit that is to be annotated.
650 'N' SP <dataref> SP <commit-ish> LF
652 Here <dataref> can be either a mark reference (:<idnum>) set by
654 a prior blob command, or a full 40-byte SHA-1 of an existing
655 Git blob object.
656 Inline data format
658 The data content for the note has not been supplied yet. The
659 frontend wants to supply it as part of this modify command.
660 'N' SP 'inline' SP <commit-ish> LF
662 data
663 See below for a detailed description of the data command.
665 In both formats <commit-ish> is any of the commit specification
667 expressions also accepted by from (see above).
668 mark
670 Arranges for fast-import to save a reference to the current object,
671 allowing the frontend to recall this object at a future point in time,
672 without knowing its SHA-1. Here the current object is the object
673 creation command the mark command appears within. This can be commit,
674 tag, and blob, but commit is the most common usage.
675 'mark' SP ':' <idnum> LF
677 where <idnum> is the number assigned by the frontend to this mark. The
679 value of <idnum> is expressed as an ASCII decimal integer. The value 0
680 is reserved and cannot be used as a mark. Only values greater than or
681 equal to 1 may be used as marks.
682 New marks are created automatically. Existing marks can be moved to
684 another object simply by reusing the same <idnum> in another mark
685 command.
686 original-oid
688 Provides the name of the object in the original source control system.
689 fast-import will simply ignore this directive, but filter processes
690 which operate on and modify the stream before feeding to fast-import
691 may have uses for this information
692 'original-oid' SP <object-identifier> LF
694 where <object-identifer> is any string not containing LF.
696 tag
698 Creates an annotated tag referring to a specific commit. To create
699 lightweight (non-annotated) tags see the reset command below.
700 'tag' SP <name> LF
702 'from' SP <commit-ish> LF
703 original-oid?
704 'tagger' (SP <name>)? SP LT <email> GT SP <when> LF
705 data
706 where <name> is the name of the tag to create.
708 Tag names are automatically prefixed with refs/tags/ when stored in
710 Git, so importing the CVS branch symbol RELENG-1_0-FINAL would use just
711 RELENG-1_0-FINAL for <name>, and fast-import will write the
712 corresponding ref as refs/tags/RELENG-1_0-FINAL.
713 The value of <name> must be a valid refname in Git and therefore may
715 contain forward slashes. As LF is not valid in a Git refname, no
716 quoting or escaping syntax is supported here.
717 The from command is the same as in the commit command; see above for
719 details.
720 The tagger command uses the same format as committer within commit;
722 again see above for details.
723 The data command following tagger must supply the annotated tag message
725 (see below for data command syntax). To import an empty tag message use
726 a 0 length data. Tag messages are free-form and are not interpreted by
727 Git. Currently they must be encoded in UTF-8, as fast-import does not
728 permit other encodings to be specified.
729 Signing annotated tags during import from within fast-import is not
731 supported. Trying to include your own PGP/GPG signature is not
732 recommended, as the frontend does not (easily) have access to the
733 complete set of bytes which normally goes into such a signature. If
734 signing is required, create lightweight tags from within fast-import
735 with reset, then create the annotated versions of those tags offline
736 with the standard git tag process.
737 reset
739 Creates (or recreates) the named branch, optionally starting from a
740 specific revision. The reset command allows a frontend to issue a new
741 from command for an existing branch, or to create a new branch from an
742 existing commit without creating a new commit.
743 'reset' SP <ref> LF
745 ('from' SP <commit-ish> LF)?
746 LF?
747 For a detailed description of <ref> and <commit-ish> see above under
749 commit and from.
750 The LF after the command is optional (it used to be required).
752 The reset command can also be used to create lightweight
754 (non-annotated) tags. For example:
755 reset refs/tags/938
757 from :938
758 would create the lightweight tag refs/tags/938 referring to whatever
760 commit mark :938 references.
761 blob
763 Requests writing one file revision to the packfile. The revision is not
764 connected to any commit; this connection must be formed in a subsequent
765 commit command by referencing the blob through an assigned mark.
766 'blob' LF
768 mark?
769 original-oid?
770 data
771 The mark command is optional here as some frontends have chosen to
773 generate the Git SHA-1 for the blob on their own, and feed that
774 directly to commit. This is typically more work than it’s worth
775 however, as marks are inexpensive to store and easy to use.
776 data
778 Supplies raw data (for use as blob/file content, commit messages, or
779 annotated tag messages) to fast-import. Data can be supplied using an
780 exact byte count or delimited with a terminating line. Real frontends
781 intended for production-quality conversions should always use the exact
782 byte count format, as it is more robust and performs better. The
783 delimited format is intended primarily for testing fast-import.
784 Comment lines appearing within the <raw> part of data commands are
786 always taken to be part of the body of the data and are therefore never
787 ignored by fast-import. This makes it safe to import any file/message
788 content whose lines might start with #.
789 Exact byte count format
791 The frontend must specify the number of bytes of data.
792 'data' SP <count> LF
794 <raw> LF?
795 where <count> is the exact number of bytes appearing within <raw>.
797 The value of <count> is expressed as an ASCII decimal integer. The
798 LF on either side of <raw> is not included in <count> and will not
799 be included in the imported data.
800 The LF after <raw> is optional (it used to be required) but
802 recommended. Always including it makes debugging a fast-import
803 stream easier as the next command always starts in column 0 of the
804 next line, even if <raw> did not end with an LF.
805 Delimited format
807 A delimiter string is used to mark the end of the data. fast-import
808 will compute the length by searching for the delimiter. This format
809 is primarily useful for testing and is not recommended for real
810 data.
811 'data' SP '<<' <delim> LF
813 <raw> LF
814 <delim> LF
815 LF?
816 where <delim> is the chosen delimiter string. The string <delim>
818 must not appear on a line by itself within <raw>, as otherwise
819 fast-import will think the data ends earlier than it really does.
820 The LF immediately trailing <raw> is part of <raw>. This is one of
821 the limitations of the delimited format, it is impossible to supply
822 a data chunk which does not have an LF as its last byte.
823 The LF after <delim> LF is optional (it used to be required).
825 checkpoint
827 Forces fast-import to close the current packfile, start a new one, and
828 to save out all current branch refs, tags and marks.
829 'checkpoint' LF
831 LF?
832 Note that fast-import automatically switches packfiles when the current
834 packfile reaches --max-pack-size, or 4 GiB, whichever limit is smaller.
835 During an automatic packfile switch fast-import does not update the
836 branch refs, tags or marks.
837 As a checkpoint can require a significant amount of CPU time and disk
839 IO (to compute the overall pack SHA-1 checksum, generate the
840 corresponding index file, and update the refs) it can easily take
841 several minutes for a single checkpoint command to complete.
842 Frontends may choose to issue checkpoints during extremely large and
844 long running imports, or when they need to allow another Git process
845 access to a branch. However given that a 30 GiB Subversion repository
846 can be loaded into Git through fast-import in about 3 hours, explicit
847 checkpointing may not be necessary.
848 The LF after the command is optional (it used to be required).
850 progress
852 Causes fast-import to print the entire progress line unmodified to its
853 standard output channel (file descriptor 1) when the command is
854 processed from the input stream. The command otherwise has no impact on
855 the current import, or on any of fast-import’s internal state.
856 'progress' SP <any> LF
858 LF?
859 The <any> part of the command may contain any sequence of bytes that
861 does not contain LF. The LF after the command is optional. Callers may
862 wish to process the output through a tool such as sed to remove the
863 leading part of the line, for example:
864 frontend | git fast-import | sed 's/^progress //'
866 Placing a progress command immediately after a checkpoint will inform
868 the reader when the checkpoint has been completed and it can safely
869 access the refs that fast-import updated.
870 get-mark
872 Causes fast-import to print the SHA-1 corresponding to a mark to stdout
873 or to the file descriptor previously arranged with the --cat-blob-fd
874 argument. The command otherwise has no impact on the current import;
875 its purpose is to retrieve SHA-1s that later commits might want to
876 refer to in their commit messages.
877 'get-mark' SP ':' <idnum> LF
879 This command can be used anywhere in the stream that comments are
881 accepted. In particular, the get-mark command can be used in the middle
882 of a commit but not in the middle of a data command.
883 See “Responses To Commands” below for details about how to read this
885 output safely.
886 cat-blob
888 Causes fast-import to print a blob to a file descriptor previously
889 arranged with the --cat-blob-fd argument. The command otherwise has no
890 impact on the current import; its main purpose is to retrieve blobs
891 that may be in fast-import’s memory but not accessible from the target
892 repository.
893 'cat-blob' SP <dataref> LF
895 The <dataref> can be either a mark reference (:<idnum>) set previously
897 or a full 40-byte SHA-1 of a Git blob, preexisting or ready to be
898 written.
899 Output uses the same format as git cat-file --batch:
901 <sha1> SP 'blob' SP <size> LF
903 <contents> LF
904 This command can be used anywhere in the stream that comments are
906 accepted. In particular, the cat-blob command can be used in the middle
907 of a commit but not in the middle of a data command.
908 See “Responses To Commands” below for details about how to read this
910 output safely.
911 ls
913 Prints information about the object at a path to a file descriptor
914 previously arranged with the --cat-blob-fd argument. This allows
915 printing a blob from the active commit (with cat-blob) or copying a
916 blob or tree from a previous commit for use in the current one (with
917 filemodify).
918 The ls command can be used anywhere in the stream that comments are
920 accepted, including the middle of a commit.
921 Reading from the active commit
923 This form can only be used in the middle of a commit. The path
924 names a directory entry within fast-import’s active commit. The
925 path must be quoted in this case.
926 'ls' SP <path> LF
928 Reading from a named tree
930 The <dataref> can be a mark reference (:<idnum>) or the full
931 40-byte SHA-1 of a Git tag, commit, or tree object, preexisting or
932 waiting to be written. The path is relative to the top level of the
933 tree named by <dataref>.
934 'ls' SP <dataref> SP <path> LF
936 See filemodify above for a detailed description of <path>.
938 Output uses the same format as git ls-tree <tree> -- <path>:
940 <mode> SP ('blob' | 'tree' | 'commit') SP <dataref> HT <path> LF
942 The <dataref> represents the blob, tree, or commit object at <path> and
944 can be used in later get-mark, cat-blob, filemodify, or ls commands.
945 If there is no file or subtree at that path, git fast-import will
947 instead report
948 missing SP <path> LF
950 See “Responses To Commands” below for details about how to read this
952 output safely.
953 feature
955 Require that fast-import supports the specified feature, or abort if it
956 does not.
957 'feature' SP <feature> ('=' <argument>)? LF
959 The <feature> part of the command may be any one of the following:
961 date-format, export-marks, relative-marks, no-relative-marks, force
963 Act as though the corresponding command-line option with a leading
964 -- was passed on the command line (see OPTIONS, above).
965 import-marks, import-marks-if-exists
967 Like --import-marks except in two respects: first, only one
968 "feature import-marks" or "feature import-marks-if-exists" command
969 is allowed per stream; second, an --import-marks= or
970 --import-marks-if-exists command-line option overrides any of these
971 "feature" commands in the stream; third, "feature
972 import-marks-if-exists" like a corresponding command-line option
973 silently skips a nonexistent file.
974 get-mark, cat-blob, ls
976 Require that the backend support the get-mark, cat-blob, or ls
977 command respectively. Versions of fast-import not supporting the
978 specified command will exit with a message indicating so. This lets
979 the import error out early with a clear message, rather than
980 wasting time on the early part of an import before the unsupported
981 command is detected.
982 notes
984 Require that the backend support the notemodify (N) subcommand to
985 the commit command. Versions of fast-import not supporting notes
986 will exit with a message indicating so.
987 done
989 Error out if the stream ends without a done command. Without this
990 feature, errors causing the frontend to end abruptly at a
991 convenient point in the stream can go undetected. This may occur,
992 for example, if an import front end dies in mid-operation without
993 emitting SIGTERM or SIGKILL at its subordinate git fast-import
994 instance.
995 option
997 Processes the specified option so that git fast-import behaves in a way
998 that suits the frontend’s needs. Note that options specified by the
999 frontend are overridden by any options the user may specify to git
1000 fast-import itself.
1001 'option' SP <option> LF
1003 The <option> part of the command may contain any of the options listed
1005 in the OPTIONS section that do not change import semantics, without the
1006 leading -- and is treated in the same way.
1007 Option commands must be the first commands on the input (not counting
1009 feature commands), to give an option command after any non-option
1010 command is an error.
1011 The following command-line options change import semantics and may
1013 therefore not be passed as option:
1014 · date-format
1016 · import-marks
1018 · export-marks
1020 · cat-blob-fd
1022 · force
1024 done
1026 If the done feature is not in use, treated as if EOF was read. This can
1027 be used to tell fast-import to finish early.
1028 If the --done command-line option or feature done command is in use,
1030 the done command is mandatory and marks the end of the stream.
1031
1033 New objects written by fast-import are not available immediately. Most
1034 fast-import commands have no visible effect until the next checkpoint
1035 (or completion). The frontend can send commands to fill fast-import’s
1036 input pipe without worrying about how quickly they will take effect,
1037 which improves performance by simplifying scheduling.
1038 For some frontends, though, it is useful to be able to read back data
1040 from the current repository as it is being updated (for example when
1041 the source material describes objects in terms of patches to be applied
1042 to previously imported objects). This can be accomplished by connecting
1043 the frontend and fast-import via bidirectional pipes:
1044 mkfifo fast-import-output
1046 frontend <fast-import-output |
1047 git fast-import >fast-import-output
1048 A frontend set up this way can use progress, get-mark, ls, and cat-blob
1050 commands to read information from the import in progress.
1051 To avoid deadlock, such frontends must completely consume any pending
1053 output from progress, ls, get-mark, and cat-blob before performing
1054 writes to fast-import that might block.
1055
1057 If fast-import is supplied invalid input it will terminate with a
1058 non-zero exit status and create a crash report in the top level of the
1059 Git repository it was importing into. Crash reports contain a snapshot
1060 of the internal fast-import state as well as the most recent commands
1061 that lead up to the crash.
1062 All recent commands (including stream comments, file changes and
1064 progress commands) are shown in the command history within the crash
1065 report, but raw file data and commit messages are excluded from the
1066 crash report. This exclusion saves space within the report file and
1067 reduces the amount of buffering that fast-import must perform during
1068 execution.
1069 After writing a crash report fast-import will close the current
1071 packfile and export the marks table. This allows the frontend developer
1072 to inspect the repository state and resume the import from the point
1073 where it crashed. The modified branches and tags are not updated during
1074 a crash, as the import did not complete successfully. Branch and tag
1075 information can be found in the crash report and must be applied
1076 manually if the update is needed.
1077 An example crash:
1079 $ cat >in <<END_OF_INPUT
1081 # my very first test commit
1082 commit refs/heads/master
1083 committer Shawn O. Pearce <spearce> 19283 -0400
1084 # who is that guy anyway?
1085 data <<EOF
1086 this is my commit
1087 EOF
1088 M 644 inline .gitignore
1089 data <<EOF
1090 .gitignore
1091 EOF
1092 M 777 inline bob
1093 END_OF_INPUT
1094 $ git fast-import <in
1096 fatal: Corrupt mode: M 777 inline bob
1097 fast-import: dumping crash report to .git/fast_import_crash_8434
1098 $ cat .git/fast_import_crash_8434
1100 fast-import crash report:
1101 fast-import process: 8434
1102 parent process : 1391
1103 at Sat Sep 1 00:58:12 2007
1104 fatal: Corrupt mode: M 777 inline bob
1106 Most Recent Commands Before Crash
1108 ---------------------------------
1109 # my very first test commit
1110 commit refs/heads/master
1111 committer Shawn O. Pearce <spearce> 19283 -0400
1112 # who is that guy anyway?
1113 data <<EOF
1114 M 644 inline .gitignore
1115 data <<EOF
1116 * M 777 inline bob
1117 Active Branch LRU
1119 -----------------
1120 active_branches = 1 cur, 5 max
1121 pos clock name
1123 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1124 1) 0 refs/heads/master
1125 Inactive Branches
1127 -----------------
1128 refs/heads/master:
1129 status : active loaded dirty
1130 tip commit : 0000000000000000000000000000000000000000
1131 old tree : 0000000000000000000000000000000000000000
1132 cur tree : 0000000000000000000000000000000000000000
1133 commit clock: 0
1134 last pack :
1135 -------------------
1137 END OF CRASH REPORT
1138
1140 The following tips and tricks have been collected from various users of
1141 fast-import, and are offered here as suggestions.
1142 Use One Mark Per Commit
1144 When doing a repository conversion, use a unique mark per commit (mark
1145 :<n>) and supply the --export-marks option on the command line.
1146 fast-import will dump a file which lists every mark and the Git object
1147 SHA-1 that corresponds to it. If the frontend can tie the marks back to
1148 the source repository, it is easy to verify the accuracy and
1149 completeness of the import by comparing each Git commit to the
1150 corresponding source revision.
1151 Coming from a system such as Perforce or Subversion this should be
1153 quite simple, as the fast-import mark can also be the Perforce
1154 changeset number or the Subversion revision number.
1155 Freely Skip Around Branches
1157 Don’t bother trying to optimize the frontend to stick to one branch at
1158 a time during an import. Although doing so might be slightly faster for
1159 fast-import, it tends to increase the complexity of the frontend code
1160 considerably.
1161 The branch LRU builtin to fast-import tends to behave very well, and
1163 the cost of activating an inactive branch is so low that bouncing
1164 around between branches has virtually no impact on import performance.
1165 Handling Renames
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. Git
1169 performs rename detection after-the-fact, rather than explicitly during
1170 a commit.
1171 Use Tag Fixup Branches
1173 Some other SCM systems let the user create a tag from multiple files
1174 which are not from the same commit/changeset. Or to create tags which
1175 are a subset of the files available in the repository.
1176 Importing these tags as-is in Git is impossible without making at least
1178 one commit which “fixes up” the files to match the content of the tag.
1179 Use fast-import’s reset command to reset a dummy branch outside of your
1180 normal branch space to the base commit for the tag, then commit one or
1181 more file fixup commits, and finally tag the dummy branch.
1182 For example since all normal branches are stored under refs/heads/ name
1184 the tag fixup branch TAG_FIXUP. This way it is impossible for the fixup
1185 branch used by the importer to have namespace conflicts with real
1186 branches imported from the source (the name TAG_FIXUP is not
1187 refs/heads/TAG_FIXUP).
1188 When committing fixups, consider using merge to connect the commit(s)
1190 which are supplying file revisions to the fixup branch. Doing so will
1191 allow tools such as git blame to track through the real commit history
1192 and properly annotate the source files.
1193 After fast-import terminates the frontend will need to do rm
1195 .git/TAG_FIXUP to remove the dummy branch.
1196 Import Now, Repack Later
1198 As soon as fast-import completes the Git repository is completely valid
1199 and ready for use. Typically this takes only a very short time, even
1200 for considerably large projects (100,000+ commits).
1201 However repacking the repository is necessary to improve data locality
1203 and access performance. It can also take hours on extremely large
1204 projects (especially if -f and a large --window parameter is used).
1205 Since repacking is safe to run alongside readers and writers, run the
1206 repack in the background and let it finish when it finishes. There is
1207 no reason to wait to explore your new Git project!
1208 If you choose to wait for the repack, don’t try to run benchmarks or
1210 performance tests until repacking is completed. fast-import outputs
1211 suboptimal packfiles that are simply never seen in real use situations.
1212 Repacking Historical Data
1214 If you are repacking very old imported data (e.g. older than the last
1215 year), consider expending some extra CPU time and supplying --window=50
1216 (or higher) when you run git repack. This will take longer, but will
1217 also produce a smaller packfile. You only need to expend the effort
1218 once, and everyone using your project will benefit from the smaller
1219 repository.
1220 Include Some Progress Messages
1222 Every once in a while have your frontend emit a progress message to
1223 fast-import. The contents of the messages are entirely free-form, so
1224 one suggestion would be to output the current month and year each time
1225 the current commit date moves into the next month. Your users will feel
1226 better knowing how much of the data stream has been processed.
1227
1229 When packing a blob fast-import always attempts to deltify against the
1230 last blob written. Unless specifically arranged for by the frontend,
1231 this will probably not be a prior version of the same file, so the
1232 generated delta will not be the smallest possible. The resulting
1233 packfile will be compressed, but will not be optimal.
1234 Frontends which have efficient access to all revisions of a single file
1236 (for example reading an RCS/CVS ,v file) can choose to supply all
1237 revisions of that file as a sequence of consecutive blob commands. This
1238 allows fast-import to deltify the different file revisions against each
1239 other, saving space in the final packfile. Marks can be used to later
1240 identify individual file revisions during a sequence of commit
1241 commands.
1242 The packfile(s) created by fast-import do not encourage good disk
1244 access patterns. This is caused by fast-import writing the data in the
1245 order it is received on standard input, while Git typically organizes
1246 data within packfiles to make the most recent (current tip) data appear
1247 before historical data. Git also clusters commits together, speeding up
1248 revision traversal through better cache locality.
1249 For this reason it is strongly recommended that users repack the
1251 repository with git repack -a -d after fast-import completes, allowing
1252 Git to reorganize the packfiles for faster data access. If blob deltas
1253 are suboptimal (see above) then also adding the -f option to force
1254 recomputation of all deltas can significantly reduce the final packfile
1255 size (30-50% smaller can be quite typical).
1256
1258 There are a number of factors which affect how much memory fast-import
1259 requires to perform an import. Like critical sections of core Git,
1260 fast-import uses its own memory allocators to amortize any overheads
1261 associated with malloc. In practice fast-import tends to amortize any
1262 malloc overheads to 0, due to its use of large block allocations.
1263 per object
1265 fast-import maintains an in-memory structure for every object written
1266 in this execution. On a 32 bit system the structure is 32 bytes, on a
1267 64 bit system the structure is 40 bytes (due to the larger pointer
1268 sizes). Objects in the table are not deallocated until fast-import
1269 terminates. Importing 2 million objects on a 32 bit system will require
1270 approximately 64 MiB of memory.
1271 The object table is actually a hashtable keyed on the object name (the
1273 unique SHA-1). This storage configuration allows fast-import to reuse
1274 an existing or already written object and avoid writing duplicates to
1275 the output packfile. Duplicate blobs are surprisingly common in an
1276 import, typically due to branch merges in the source.
1277 per mark
1279 Marks are stored in a sparse array, using 1 pointer (4 bytes or 8
1280 bytes, depending on pointer size) per mark. Although the array is
1281 sparse, frontends are still strongly encouraged to use marks between 1
1282 and n, where n is the total number of marks required for this import.
1283 per branch
1285 Branches are classified as active and inactive. The memory usage of the
1286 two classes is significantly different.
1287 Inactive branches are stored in a structure which uses 96 or 120 bytes
1289 (32 bit or 64 bit systems, respectively), plus the length of the branch
1290 name (typically under 200 bytes), per branch. fast-import will easily
1291 handle as many as 10,000 inactive branches in under 2 MiB of memory.
1292 Active branches have the same overhead as inactive branches, but also
1294 contain copies of every tree that has been recently modified on that
1295 branch. If subtree include has not been modified since the branch
1296 became active, its contents will not be loaded into memory, but if
1297 subtree src has been modified by a commit since the branch became
1298 active, then its contents will be loaded in memory.
1299 As active branches store metadata about the files contained on that
1301 branch, their in-memory storage size can grow to a considerable size
1302 (see below).
1303 fast-import automatically moves active branches to inactive status
1305 based on a simple least-recently-used algorithm. The LRU chain is
1306 updated on each commit command. The maximum number of active branches
1307 can be increased or decreased on the command line with
1308 --active-branches=.
1309 per active tree
1311 Trees (aka directories) use just 12 bytes of memory on top of the
1312 memory required for their entries (see “per active file” below). The
1313 cost of a tree is virtually 0, as its overhead amortizes out over the
1314 individual file entries.
1315 per active file entry
1317 Files (and pointers to subtrees) within active trees require 52 or 64
1318 bytes (32/64 bit platforms) per entry. To conserve space, file and tree
1319 names are pooled in a common string table, allowing the filename
1320 “Makefile” to use just 16 bytes (after including the string header
1321 overhead) no matter how many times it occurs within the project.
1322 The active branch LRU, when coupled with the filename string pool and
1324 lazy loading of subtrees, allows fast-import to efficiently import
1325 projects with 2,000+ branches and 45,114+ files in a very limited
1326 memory footprint (less than 2.7 MiB per active branch).
1327
1329 Sending SIGUSR1 to the git fast-import process ends the current
1330 packfile early, simulating a checkpoint command. The impatient operator
1331 can use this facility to peek at the objects and refs from an import in
1332 progress, at the cost of some added running time and worse compression.
1333
1335 git-fast-export(1)
1336
1338 Part of the git(1) suite
1339Git 2.21.0 02/24/2019 GIT-FAST-IMPORT(1)