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 all non-fatal output, making fast-import silent when it is
38 successful. This option disables the output shown by --stats.
39 --stats
41 Display some basic statistics about the objects fast-import has
42 created, the packfiles they were stored into, and the memory used
43 by fast-import during this run. Showing this output is currently
44 the default, but can be disabled with --quiet.
45 Options for Frontends
47 --cat-blob-fd=<fd>
48 Write responses to cat-blob and ls queries to the file descriptor
49 <fd> instead of stdout. Allows progress output intended for the
50 end-user to be separated from other output.
51 --date-format=<fmt>
53 Specify the type of dates the frontend will supply to fast-import
54 within author, committer and tagger commands. See “Date Formats”
55 below for details about which formats are supported, and their
56 syntax.
57 --done
59 Terminate with error if there is no done command at the end of the
60 stream. This option might be useful for detecting errors that cause
61 the frontend to terminate before it has started to write a stream.
62 Locations of Marks Files
64 --export-marks=<file>
65 Dumps the internal marks table to <file> when complete. Marks are
66 written one per line as :markid SHA-1. Frontends can use this file
67 to validate imports after they have been completed, or to save the
68 marks table across incremental runs. As <file> is only opened and
69 truncated at checkpoint (or completion) the same path can also be
70 safely given to --import-marks.
71 --import-marks=<file>
73 Before processing any input, load the marks specified in <file>.
74 The input file must exist, must be readable, and must use the same
75 format as produced by --export-marks. Multiple options may be
76 supplied to import more than one set of marks. If a mark is defined
77 to different values, the last file wins.
78 --import-marks-if-exists=<file>
80 Like --import-marks but instead of erroring out, silently skips the
81 file if it does not exist.
82 --[no-]relative-marks
84 After specifying --relative-marks the paths specified with
85 --import-marks= and --export-marks= are relative to an internal
86 directory in the current repository. In git-fast-import this means
87 that the paths are relative to the .git/info/fast-import directory.
88 However, other importers may use a different location.
89 Relative and non-relative marks may be combined by interweaving
91 --(no-)-relative-marks with the --(import|export)-marks= options.
92 Performance and Compression Tuning
94 --active-branches=<n>
95 Maximum number of branches to maintain active at once. See “Memory
96 Utilization” below for details. Default is 5.
97 --big-file-threshold=<n>
99 Maximum size of a blob that fast-import will attempt to create a
100 delta for, expressed in bytes. The default is 512m (512 MiB). Some
101 importers may wish to lower this on systems with constrained
102 memory.
103 --depth=<n>
105 Maximum delta depth, for blob and tree deltification. Default is
106 10.
107 --export-pack-edges=<file>
109 After creating a packfile, print a line of data to <file> listing
110 the filename of the packfile and the last commit on each branch
111 that was written to that packfile. This information may be useful
112 after importing projects whose total object set exceeds the 4 GiB
113 packfile limit, as these commits can be used as edge points during
114 calls to git pack-objects.
115 --max-pack-size=<n>
117 Maximum size of each output packfile. The default is unlimited.
118
120 The design of fast-import allows it to import large projects in a
121 minimum amount of memory usage and processing time. Assuming the
122 frontend is able to keep up with fast-import and feed it a constant
123 stream of data, import times for projects holding 10+ years of history
124 and containing 100,000+ individual commits are generally completed in
125 just 1-2 hours on quite modest (~$2,000 USD) hardware.
126 Most bottlenecks appear to be in foreign source data access (the source
128 just cannot extract revisions fast enough) or disk IO (fast-import
129 writes as fast as the disk will take the data). Imports will run faster
130 if the source data is stored on a different drive than the destination
131 Git repository (due to less IO contention).
132
134 A typical frontend for fast-import tends to weigh in at approximately
135 200 lines of Perl/Python/Ruby code. Most developers have been able to
136 create working importers in just a couple of hours, even though it is
137 their first exposure to fast-import, and sometimes even to Git. This is
138 an ideal situation, given that most conversion tools are throw-away
139 (use once, and never look back).
140
142 Like git push or git fetch, imports handled by fast-import are safe to
143 run alongside parallel git repack -a -d or git gc invocations, or any
144 other Git operation (including git prune, as loose objects are never
145 used by fast-import).
146 fast-import does not lock the branch or tag refs it is actively
148 importing. After the import, during its ref update phase, fast-import
149 tests each existing branch ref to verify the update will be a
150 fast-forward update (the commit stored in the ref is contained in the
151 new history of the commit to be written). If the update is not a
152 fast-forward update, fast-import will skip updating that ref and
153 instead prints a warning message. fast-import will always attempt to
154 update all branch refs, and does not stop on the first failure.
155 Branch updates can be forced with --force, but it’s recommended that
157 this only be used on an otherwise quiet repository. Using --force is
158 not necessary for an initial import into an empty repository.
159
161 fast-import tracks a set of branches in memory. Any branch can be
162 created or modified at any point during the import process by sending a
163 commit command on the input stream. This design allows a frontend
164 program to process an unlimited number of branches simultaneously,
165 generating commits in the order they are available from the source
166 data. It also simplifies the frontend programs considerably.
167 fast-import does not use or alter the current working directory, or any
169 file within it. (It does however update the current Git repository, as
170 referenced by GIT_DIR.) Therefore an import frontend may use the
171 working directory for its own purposes, such as extracting file
172 revisions from the foreign source. This ignorance of the working
173 directory also allows fast-import to run very quickly, as it does not
174 need to perform any costly file update operations when switching
175 between branches.
176
178 With the exception of raw file data (which Git does not interpret) the
179 fast-import input format is text (ASCII) based. This text based format
180 simplifies development and debugging of frontend programs, especially
181 when a higher level language such as Perl, Python or Ruby is being
182 used.
183 fast-import is very strict about its input. Where we say SP below we
185 mean exactly one space. Likewise LF means one (and only one) linefeed
186 and HT one (and only one) horizontal tab. Supplying additional
187 whitespace characters will cause unexpected results, such as branch
188 names or file names with leading or trailing spaces in their name, or
189 early termination of fast-import when it encounters unexpected input.
190 Stream Comments
192 To aid in debugging frontends fast-import ignores any line that begins
193 with # (ASCII pound/hash) up to and including the line ending LF. A
194 comment line may contain any sequence of bytes that does not contain an
195 LF and therefore may be used to include any detailed debugging
196 information that might be specific to the frontend and useful when
197 inspecting a fast-import data stream.
198 Date Formats
200 The following date formats are supported. A frontend should select the
201 format it will use for this import by passing the format name in the
202 --date-format=<fmt> command line option.
203 raw
205 This is the Git native format and is <time> SP <offutc>. It is also
206 fast-import’s default format, if --date-format was not specified.
207 The time of the event is specified by <time> as the number of
209 seconds since the UNIX epoch (midnight, Jan 1, 1970, UTC) and is
210 written as an ASCII decimal integer.
211 The local offset is specified by <offutc> as a positive or negative
213 offset from UTC. For example EST (which is 5 hours behind UTC)
214 would be expressed in <tz> by “-0500” while UTC is “+0000”. The
215 local offset does not affect <time>; it is used only as an
216 advisement to help formatting routines display the timestamp.
217 If the local offset is not available in the source material, use
219 “+0000”, or the most common local offset. For example many
220 organizations have a CVS repository which has only ever been
221 accessed by users who are located in the same location and
222 timezone. In this case a reasonable offset from UTC could be
223 assumed.
224 Unlike the rfc2822 format, this format is very strict. Any
226 variation in formatting will cause fast-import to reject the value.
227 rfc2822
229 This is the standard email format as described by RFC 2822.
230 An example value is “Tue Feb 6 11:22:18 2007 -0500”. The Git parser
232 is accurate, but a little on the lenient side. It is the same
233 parser used by git am when applying patches received from email.
234 Some malformed strings may be accepted as valid dates. In some of
236 these cases Git will still be able to obtain the correct date from
237 the malformed string. There are also some types of malformed
238 strings which Git will parse wrong, and yet consider valid.
239 Seriously malformed strings will be rejected.
240 Unlike the raw format above, the timezone/UTC offset information
242 contained in an RFC 2822 date string is used to adjust the date
243 value to UTC prior to storage. Therefore it is important that this
244 information be as accurate as possible.
245 If the source material uses RFC 2822 style dates, the frontend
247 should let fast-import handle the parsing and conversion (rather
248 than attempting to do it itself) as the Git parser has been well
249 tested in the wild.
250 Frontends should prefer the raw format if the source material
252 already uses UNIX-epoch format, can be coaxed to give dates in that
253 format, or its format is easily convertible to it, as there is no
254 ambiguity in parsing.
255 now
257 Always use the current time and timezone. The literal now must
258 always be supplied for <when>.
259 This is a toy format. The current time and timezone of this system
261 is always copied into the identity string at the time it is being
262 created by fast-import. There is no way to specify a different time
263 or timezone.
264 This particular format is supplied as it’s short to implement and
266 may be useful to a process that wants to create a new commit right
267 now, without needing to use a working directory or git
268 update-index.
269 If separate author and committer commands are used in a commit the
271 timestamps may not match, as the system clock will be polled twice
272 (once for each command). The only way to ensure that both author
273 and committer identity information has the same timestamp is to
274 omit author (thus copying from committer) or to use a date format
275 other than now.
276 Commands
278 fast-import accepts several commands to update the current repository
279 and control the current import process. More detailed discussion (with
280 examples) of each command follows later.
281 commit
283 Creates a new branch or updates an existing branch by creating a
284 new commit and updating the branch to point at the newly created
285 commit.
286 tag
288 Creates an annotated tag object from an existing commit or branch.
289 Lightweight tags are not supported by this command, as they are not
290 recommended for recording meaningful points in time.
291 reset
293 Reset an existing branch (or a new branch) to a specific revision.
294 This command must be used to change a branch to a specific revision
295 without making a commit on it.
296 blob
298 Convert raw file data into a blob, for future use in a commit
299 command. This command is optional and is not needed to perform an
300 import.
301 checkpoint
303 Forces fast-import to close the current packfile, generate its
304 unique SHA-1 checksum and index, and start a new packfile. This
305 command is optional and is not needed to perform an import.
306 progress
308 Causes fast-import to echo the entire line to its own standard
309 output. This command is optional and is not needed to perform an
310 import.
311 done
313 Marks the end of the stream. This command is optional unless the
314 done feature was requested using the --done command line option or
315 feature done command.
316 cat-blob
318 Causes fast-import to print a blob in cat-file --batch format to
319 the file descriptor set with --cat-blob-fd or stdout if
320 unspecified.
321 ls
323 Causes fast-import to print a line describing a directory entry in
324 ls-tree format to the file descriptor set with --cat-blob-fd or
325 stdout if unspecified.
326 feature
328 Require that fast-import supports the specified feature, or abort
329 if it does not.
330 option
332 Specify any of the options listed under OPTIONS that do not change
333 stream semantic to suit the frontend’s needs. This command is
334 optional and is not needed to perform an import.
335 commit
337 Create or update a branch with a new commit, recording one logical
338 change to the project.
339 'commit' SP <ref> LF
341 mark?
342 ('author' (SP <name>)? SP LT <email> GT SP <when> LF)?
343 'committer' (SP <name>)? SP LT <email> GT SP <when> LF
344 data
345 ('from' SP <committish> LF)?
346 ('merge' SP <committish> LF)?
347 (filemodify | filedelete | filecopy | filerename | filedeleteall | notemodify)*
348 LF?
349 where <ref> is the name of the branch to make the commit on. Typically
351 branch names are prefixed with refs/heads/ in Git, so importing the CVS
352 branch symbol RELENG-1_0 would use refs/heads/RELENG-1_0 for the value
353 of <ref>. The value of <ref> must be a valid refname in Git. As LF is
354 not valid in a Git refname, no quoting or escaping syntax is supported
355 here.
356 A mark command may optionally appear, requesting fast-import to save a
358 reference to the newly created commit for future use by the frontend
359 (see below for format). It is very common for frontends to mark every
360 commit they create, thereby allowing future branch creation from any
361 imported commit.
362 The data command following committer must supply the commit message
364 (see below for data command syntax). To import an empty commit message
365 use a 0 length data. Commit messages are free-form and are not
366 interpreted by Git. Currently they must be encoded in UTF-8, as
367 fast-import does not permit other encodings to be specified.
368 Zero or more filemodify, filedelete, filecopy, filerename,
370 filedeleteall and notemodify commands may be included to update the
371 contents of the branch prior to creating the commit. These commands may
372 be supplied in any order. However it is recommended that a
373 filedeleteall command precede all filemodify, filecopy, filerename and
374 notemodify commands in the same commit, as filedeleteall wipes the
375 branch clean (see below).
376 The LF after the command is optional (it used to be required).
378 author
380 An author command may optionally appear, if the author information
381 might differ from the committer information. If author is omitted
382 then fast-import will automatically use the committer’s information
383 for the author portion of the commit. See below for a description
384 of the fields in author, as they are identical to committer.
385 committer
387 The committer command indicates who made this commit, and when they
388 made it.
389 Here <name> is the person’s display name (for example “Com M
391 Itter”) and <email> is the person’s email address
392 (“cm@example.com”). LT and GT are the literal less-than (\x3c) and
393 greater-than (\x3e) symbols. These are required to delimit the
394 email address from the other fields in the line. Note that <name>
395 and <email> are free-form and may contain any sequence of bytes,
396 except LT, GT and LF. <name> is typically UTF-8 encoded.
397 The time of the change is specified by <when> using the date format
399 that was selected by the --date-format=<fmt> command line option.
400 See “Date Formats” above for the set of supported formats, and
401 their syntax.
402 from
404 The from command is used to specify the commit to initialize this
405 branch from. This revision will be the first ancestor of the new
406 commit. The state of the tree built at this commit will begin with
407 the state at the from commit, and be altered by the content
408 modifications in this commit.
409 Omitting the from command in the first commit of a new branch will
411 cause fast-import to create that commit with no ancestor. This
412 tends to be desired only for the initial commit of a project. If
413 the frontend creates all files from scratch when making a new
414 branch, a merge command may be used instead of from to start the
415 commit with an empty tree. Omitting the from command on existing
416 branches is usually desired, as the current commit on that branch
417 is automatically assumed to be the first ancestor of the new
418 commit.
419 As LF is not valid in a Git refname or SHA-1 expression, no quoting
421 or escaping syntax is supported within <committish>.
422 Here <committish> is any of the following:
424 · The name of an existing branch already in fast-import’s
426 internal branch table. If fast-import doesn’t know the name,
427 it’s treated as a SHA-1 expression.
428 · A mark reference, :<idnum>, where <idnum> is the mark number.
430 The reason fast-import uses : to denote a mark reference is
432 this character is not legal in a Git branch name. The leading :
433 makes it easy to distinguish between the mark 42 (:42) and the
434 branch 42 (42 or refs/heads/42), or an abbreviated SHA-1 which
435 happened to consist only of base-10 digits.
436 Marks must be declared (via mark) before they can be used.
438 · A complete 40 byte or abbreviated commit SHA-1 in hex.
440 · Any valid Git SHA-1 expression that resolves to a commit. See
442 “SPECIFYING REVISIONS” in gitrevisions(7) for details.
443 The special case of restarting an incremental import from the
445 current branch value should be written as:
446 from refs/heads/branch^0
448 The ^0 suffix is necessary as fast-import does not permit a branch
451 to start from itself, and the branch is created in memory before
452 the from command is even read from the input. Adding ^0 will force
453 fast-import to resolve the commit through Git’s revision parsing
454 library, rather than its internal branch table, thereby loading in
455 the existing value of the branch.
456 merge
458 Includes one additional ancestor commit. The additional ancestry
459 link does not change the way the tree state is built at this
460 commit. If the from command is omitted when creating a new branch,
461 the first merge commit will be the first ancestor of the current
462 commit, and the branch will start out with no files. An unlimited
463 number of merge commands per commit are permitted by fast-import,
464 thereby establishing an n-way merge. However Git’s other tools
465 never create commits with more than 15 additional ancestors
466 (forming a 16-way merge). For this reason it is suggested that
467 frontends do not use more than 15 merge commands per commit; 16, if
468 starting a new, empty branch.
469 Here <committish> is any of the commit specification expressions
471 also accepted by from (see above).
472 filemodify
474 Included in a commit command to add a new file or change the
475 content of an existing file. This command has two different means
476 of specifying the content of the file.
477 External data format
479 The data content for the file was already supplied by a prior
480 blob command. The frontend just needs to connect it.
481 'M' SP <mode> SP <dataref> SP <path> LF
483 Here usually <dataref> must be either a mark reference
485 (:<idnum>) set by a prior blob command, or a full 40-byte SHA-1
486 of an existing Git blob object. If <mode> is 040000` then
487 <dataref> must be the full 40-byte SHA-1 of an existing Git
488 tree object or a mark reference set with --import-marks.
489 Inline data format
491 The data content for the file has not been supplied yet. The
492 frontend wants to supply it as part of this modify command.
493 'M' SP <mode> SP 'inline' SP <path> LF
495 data
496 See below for a detailed description of the data command.
498 In both formats <mode> is the type of file entry, specified in
500 octal. Git only supports the following modes:
501 · 100644 or 644: A normal (not-executable) file. The majority of
503 files in most projects use this mode. If in doubt, this is what
504 you want.
505 · 100755 or 755: A normal, but executable, file.
507 · 120000: A symlink, the content of the file will be the link
509 target.
510 · 160000: A gitlink, SHA-1 of the object refers to a commit in
512 another repository. Git links can only be specified by SHA or
513 through a commit mark. They are used to implement submodules.
514 · 040000: A subdirectory. Subdirectories can only be specified by
516 SHA or through a tree mark set with --import-marks.
517 In both formats <path> is the complete path of the file to be added
519 (if not already existing) or modified (if already existing).
520 A <path> string must use UNIX-style directory separators (forward
522 slash /), may contain any byte other than LF, and must not start
523 with double quote (").
524 A path can use C-style string quoting; this is accepted in all
526 cases and mandatory if the filename starts with double quote or
527 contains LF. In C-style quoting, the complete name should be
528 surrounded with double quotes, and any LF, backslash, or double
529 quote characters must be escaped by preceding them with a backslash
530 (e.g., "path/with\n, \\ and \" in it").
531 The value of <path> must be in canonical form. That is it must not:
533 · contain an empty directory component (e.g. foo//bar is
535 invalid),
536 · end with a directory separator (e.g. foo/ is invalid),
538 · start with a directory separator (e.g. /foo is invalid),
540 · contain the special component . or .. (e.g. foo/./bar and
542 foo/../bar are invalid).
543 The root of the tree can be represented by an empty string as
545 <path>.
546 It is recommended that <path> always be encoded using UTF-8.
548 filedelete
550 Included in a commit command to remove a file or recursively delete
551 an entire directory from the branch. If the file or directory
552 removal makes its parent directory empty, the parent directory will
553 be automatically removed too. This cascades up the tree until the
554 first non-empty directory or the root is reached.
555 'D' SP <path> LF
557 here <path> is the complete path of the file or subdirectory to be
559 removed from the branch. See filemodify above for a detailed
560 description of <path>.
561 filecopy
563 Recursively copies an existing file or subdirectory to a different
564 location within the branch. The existing file or directory must
565 exist. If the destination exists it will be completely replaced by
566 the content copied from the source.
567 'C' SP <path> SP <path> LF
569 here the first <path> is the source location and the second <path>
571 is the destination. See filemodify above for a detailed description
572 of what <path> may look like. To use a source path that contains SP
573 the path must be quoted.
574 A filecopy command takes effect immediately. Once the source
576 location has been copied to the destination any future commands
577 applied to the source location will not impact the destination of
578 the copy.
579 filerename
581 Renames an existing file or subdirectory to a different location
582 within the branch. The existing file or directory must exist. If
583 the destination exists it will be replaced by the source directory.
584 'R' SP <path> SP <path> LF
586 here the first <path> is the source location and the second <path>
588 is the destination. See filemodify above for a detailed description
589 of what <path> may look like. To use a source path that contains SP
590 the path must be quoted.
591 A filerename command takes effect immediately. Once the source
593 location has been renamed to the destination any future commands
594 applied to the source location will create new files there and not
595 impact the destination of the rename.
596 Note that a filerename is the same as a filecopy followed by a
598 filedelete of the source location. There is a slight performance
599 advantage to using filerename, but the advantage is so small that
600 it is never worth trying to convert a delete/add pair in source
601 material into a rename for fast-import. This filerename command is
602 provided just to simplify frontends that already have rename
603 information and don’t want bother with decomposing it into a
604 filecopy followed by a filedelete.
605 filedeleteall
607 Included in a commit command to remove all files (and also all
608 directories) from the branch. This command resets the internal
609 branch structure to have no files in it, allowing the frontend to
610 subsequently add all interesting files from scratch.
611 'deleteall' LF
613 This command is extremely useful if the frontend does not know (or
615 does not care to know) what files are currently on the branch, and
616 therefore cannot generate the proper filedelete commands to update
617 the content.
618 Issuing a filedeleteall followed by the needed filemodify commands
620 to set the correct content will produce the same results as sending
621 only the needed filemodify and filedelete commands. The
622 filedeleteall approach may however require fast-import to use
623 slightly more memory per active branch (less than 1 MiB for even
624 most large projects); so frontends that can easily obtain only the
625 affected paths for a commit are encouraged to do so.
626 notemodify
628 Included in a commit <notes_ref> command to add a new note
629 annotating a <committish> or change this annotation contents.
630 Internally it is similar to filemodify 100644 on <committish> path
631 (maybe split into subdirectories). It’s not advised to use any
632 other commands to write to the <notes_ref> tree except
633 filedeleteall to delete all existing notes in this tree. This
634 command has two different means of specifying the content of the
635 note.
636 External data format
638 The data content for the note was already supplied by a prior
639 blob command. The frontend just needs to connect it to the
640 commit that is to be annotated.
641 'N' SP <dataref> SP <committish> LF
643 Here <dataref> can be either a mark reference (:<idnum>) set by
645 a prior blob command, or a full 40-byte SHA-1 of an existing
646 Git blob object.
647 Inline data format
649 The data content for the note has not been supplied yet. The
650 frontend wants to supply it as part of this modify command.
651 'N' SP 'inline' SP <committish> LF
653 data
654 See below for a detailed description of the data command.
656 In both formats <committish> is any of the commit specification
658 expressions also accepted by from (see above).
659 mark
661 Arranges for fast-import to save a reference to the current object,
662 allowing the frontend to recall this object at a future point in time,
663 without knowing its SHA-1. Here the current object is the object
664 creation command the mark command appears within. This can be commit,
665 tag, and blob, but commit is the most common usage.
666 'mark' SP ':' <idnum> LF
668 where <idnum> is the number assigned by the frontend to this mark. The
670 value of <idnum> is expressed as an ASCII decimal integer. The value 0
671 is reserved and cannot be used as a mark. Only values greater than or
672 equal to 1 may be used as marks.
673 New marks are created automatically. Existing marks can be moved to
675 another object simply by reusing the same <idnum> in another mark
676 command.
677 tag
679 Creates an annotated tag referring to a specific commit. To create
680 lightweight (non-annotated) tags see the reset command below.
681 'tag' SP <name> LF
683 'from' SP <committish> LF
684 'tagger' (SP <name>)? SP LT <email> GT SP <when> LF
685 data
686 where <name> is the name of the tag to create.
688 Tag names are automatically prefixed with refs/tags/ when stored in
690 Git, so importing the CVS branch symbol RELENG-1_0-FINAL would use just
691 RELENG-1_0-FINAL for <name>, and fast-import will write the
692 corresponding ref as refs/tags/RELENG-1_0-FINAL.
693 The value of <name> must be a valid refname in Git and therefore may
695 contain forward slashes. As LF is not valid in a Git refname, no
696 quoting or escaping syntax is supported here.
697 The from command is the same as in the commit command; see above for
699 details.
700 The tagger command uses the same format as committer within commit;
702 again see above for details.
703 The data command following tagger must supply the annotated tag message
705 (see below for data command syntax). To import an empty tag message use
706 a 0 length data. Tag messages are free-form and are not interpreted by
707 Git. Currently they must be encoded in UTF-8, as fast-import does not
708 permit other encodings to be specified.
709 Signing annotated tags during import from within fast-import is not
711 supported. Trying to include your own PGP/GPG signature is not
712 recommended, as the frontend does not (easily) have access to the
713 complete set of bytes which normally goes into such a signature. If
714 signing is required, create lightweight tags from within fast-import
715 with reset, then create the annotated versions of those tags offline
716 with the standard git tag process.
717 reset
719 Creates (or recreates) the named branch, optionally starting from a
720 specific revision. The reset command allows a frontend to issue a new
721 from command for an existing branch, or to create a new branch from an
722 existing commit without creating a new commit.
723 'reset' SP <ref> LF
725 ('from' SP <committish> LF)?
726 LF?
727 For a detailed description of <ref> and <committish> see above under
729 commit and from.
730 The LF after the command is optional (it used to be required).
732 The reset command can also be used to create lightweight
734 (non-annotated) tags. For example:
735 reset refs/tags/938
737 from :938
738 would create the lightweight tag refs/tags/938 referring to whatever
740 commit mark :938 references.
741 blob
743 Requests writing one file revision to the packfile. The revision is not
744 connected to any commit; this connection must be formed in a subsequent
745 commit command by referencing the blob through an assigned mark.
746 'blob' LF
748 mark?
749 data
750 The mark command is optional here as some frontends have chosen to
752 generate the Git SHA-1 for the blob on their own, and feed that
753 directly to commit. This is typically more work than it’s worth
754 however, as marks are inexpensive to store and easy to use.
755 data
757 Supplies raw data (for use as blob/file content, commit messages, or
758 annotated tag messages) to fast-import. Data can be supplied using an
759 exact byte count or delimited with a terminating line. Real frontends
760 intended for production-quality conversions should always use the exact
761 byte count format, as it is more robust and performs better. The
762 delimited format is intended primarily for testing fast-import.
763 Comment lines appearing within the <raw> part of data commands are
765 always taken to be part of the body of the data and are therefore never
766 ignored by fast-import. This makes it safe to import any file/message
767 content whose lines might start with #.
768 Exact byte count format
770 The frontend must specify the number of bytes of data.
771 'data' SP <count> LF
773 <raw> LF?
774 where <count> is the exact number of bytes appearing within <raw>.
776 The value of <count> is expressed as an ASCII decimal integer. The
777 LF on either side of <raw> is not included in <count> and will not
778 be included in the imported data.
779 The LF after <raw> is optional (it used to be required) but
781 recommended. Always including it makes debugging a fast-import
782 stream easier as the next command always starts in column 0 of the
783 next line, even if <raw> did not end with an LF.
784 Delimited format
786 A delimiter string is used to mark the end of the data. fast-import
787 will compute the length by searching for the delimiter. This format
788 is primarily useful for testing and is not recommended for real
789 data.
790 'data' SP '<<' <delim> LF
792 <raw> LF
793 <delim> LF
794 LF?
795 where <delim> is the chosen delimiter string. The string <delim>
797 must not appear on a line by itself within <raw>, as otherwise
798 fast-import will think the data ends earlier than it really does.
799 The LF immediately trailing <raw> is part of <raw>. This is one of
800 the limitations of the delimited format, it is impossible to supply
801 a data chunk which does not have an LF as its last byte.
802 The LF after <delim> LF is optional (it used to be required).
804 checkpoint
806 Forces fast-import to close the current packfile, start a new one, and
807 to save out all current branch refs, tags and marks.
808 'checkpoint' LF
810 LF?
811 Note that fast-import automatically switches packfiles when the current
813 packfile reaches --max-pack-size, or 4 GiB, whichever limit is smaller.
814 During an automatic packfile switch fast-import does not update the
815 branch refs, tags or marks.
816 As a checkpoint can require a significant amount of CPU time and disk
818 IO (to compute the overall pack SHA-1 checksum, generate the
819 corresponding index file, and update the refs) it can easily take
820 several minutes for a single checkpoint command to complete.
821 Frontends may choose to issue checkpoints during extremely large and
823 long running imports, or when they need to allow another Git process
824 access to a branch. However given that a 30 GiB Subversion repository
825 can be loaded into Git through fast-import in about 3 hours, explicit
826 checkpointing may not be necessary.
827 The LF after the command is optional (it used to be required).
829 progress
831 Causes fast-import to print the entire progress line unmodified to its
832 standard output channel (file descriptor 1) when the command is
833 processed from the input stream. The command otherwise has no impact on
834 the current import, or on any of fast-import’s internal state.
835 'progress' SP <any> LF
837 LF?
838 The <any> part of the command may contain any sequence of bytes that
840 does not contain LF. The LF after the command is optional. Callers may
841 wish to process the output through a tool such as sed to remove the
842 leading part of the line, for example:
843 frontend | git fast-import | sed 's/^progress //'
845 Placing a progress command immediately after a checkpoint will inform
847 the reader when the checkpoint has been completed and it can safely
848 access the refs that fast-import updated.
849 cat-blob
851 Causes fast-import to print a blob to a file descriptor previously
852 arranged with the --cat-blob-fd argument. The command otherwise has no
853 impact on the current import; its main purpose is to retrieve blobs
854 that may be in fast-import’s memory but not accessible from the target
855 repository.
856 'cat-blob' SP <dataref> LF
858 The <dataref> can be either a mark reference (:<idnum>) set previously
860 or a full 40-byte SHA-1 of a Git blob, preexisting or ready to be
861 written.
862 Output uses the same format as git cat-file --batch:
864 <sha1> SP 'blob' SP <size> LF
866 <contents> LF
867 This command can be used anywhere in the stream that comments are
869 accepted. In particular, the cat-blob command can be used in the middle
870 of a commit but not in the middle of a data command.
871 See “Responses To Commands” below for details about how to read this
873 output safely.
874 ls
876 Prints information about the object at a path to a file descriptor
877 previously arranged with the --cat-blob-fd argument. This allows
878 printing a blob from the active commit (with cat-blob) or copying a
879 blob or tree from a previous commit for use in the current one (with
880 filemodify).
881 The ls command can be used anywhere in the stream that comments are
883 accepted, including the middle of a commit.
884 Reading from the active commit
886 This form can only be used in the middle of a commit. The path
887 names a directory entry within fast-import’s active commit. The
888 path must be quoted in this case.
889 'ls' SP <path> LF
891 Reading from a named tree
893 The <dataref> can be a mark reference (:<idnum>) or the full
894 40-byte SHA-1 of a Git tag, commit, or tree object, preexisting or
895 waiting to be written. The path is relative to the top level of the
896 tree named by <dataref>.
897 'ls' SP <dataref> SP <path> LF
899 See filemodify above for a detailed description of <path>.
901 Output uses the same format as git ls-tree <tree> -- <path>:
903 <mode> SP ('blob' | 'tree' | 'commit') SP <dataref> HT <path> LF
905 The <dataref> represents the blob, tree, or commit object at <path> and
907 can be used in later cat-blob, filemodify, or ls commands.
908 If there is no file or subtree at that path, git fast-import will
910 instead report
911 missing SP <path> LF
913 See “Responses To Commands” below for details about how to read this
915 output safely.
916 feature
918 Require that fast-import supports the specified feature, or abort if it
919 does not.
920 'feature' SP <feature> ('=' <argument>)? LF
922 The <feature> part of the command may be any one of the following:
924 date-format, export-marks, relative-marks, no-relative-marks, force
926 Act as though the corresponding command-line option with a leading
927 -- was passed on the command line (see OPTIONS, above).
928 import-marks, import-marks-if-exists
930 Like --import-marks except in two respects: first, only one
931 "feature import-marks" or "feature import-marks-if-exists" command
932 is allowed per stream; second, an --import-marks= or
933 --import-marks-if-exists command-line option overrides any of these
934 "feature" commands in the stream; third, "feature
935 import-marks-if-exists" like a corresponding command-line option
936 silently skips a nonexistent file.
937 cat-blob, ls
939 Require that the backend support the cat-blob or ls command.
940 Versions of fast-import not supporting the specified command will
941 exit with a message indicating so. This lets the import error out
942 early with a clear message, rather than wasting time on the early
943 part of an import before the unsupported command is detected.
944 notes
946 Require that the backend support the notemodify (N) subcommand to
947 the commit command. Versions of fast-import not supporting notes
948 will exit with a message indicating so.
949 done
951 Error out if the stream ends without a done command. Without this
952 feature, errors causing the frontend to end abruptly at a
953 convenient point in the stream can go undetected. This may occur,
954 for example, if an import front end dies in mid-operation without
955 emitting SIGTERM or SIGKILL at its subordinate git fast-import
956 instance.
957 option
959 Processes the specified option so that git fast-import behaves in a way
960 that suits the frontend’s needs. Note that options specified by the
961 frontend are overridden by any options the user may specify to git
962 fast-import itself.
963 'option' SP <option> LF
965 The <option> part of the command may contain any of the options listed
967 in the OPTIONS section that do not change import semantics, without the
968 leading -- and is treated in the same way.
969 Option commands must be the first commands on the input (not counting
971 feature commands), to give an option command after any non-option
972 command is an error.
973 The following commandline options change import semantics and may
975 therefore not be passed as option:
976 · date-format
978 · import-marks
980 · export-marks
982 · cat-blob-fd
984 · force
986 done
988 If the done feature is not in use, treated as if EOF was read. This can
989 be used to tell fast-import to finish early.
990 If the --done command line option or feature done command is in use,
992 the done command is mandatory and marks the end of the stream.
993
995 New objects written by fast-import are not available immediately. Most
996 fast-import commands have no visible effect until the next checkpoint
997 (or completion). The frontend can send commands to fill fast-import’s
998 input pipe without worrying about how quickly they will take effect,
999 which improves performance by simplifying scheduling.
1000 For some frontends, though, it is useful to be able to read back data
1002 from the current repository as it is being updated (for example when
1003 the source material describes objects in terms of patches to be applied
1004 to previously imported objects). This can be accomplished by connecting
1005 the frontend and fast-import via bidirectional pipes:
1006 mkfifo fast-import-output
1008 frontend <fast-import-output |
1009 git fast-import >fast-import-output
1010 A frontend set up this way can use progress, ls, and cat-blob commands
1012 to read information from the import in progress.
1013 To avoid deadlock, such frontends must completely consume any pending
1015 output from progress, ls, and cat-blob before performing writes to
1016 fast-import that might block.
1017
1019 If fast-import is supplied invalid input it will terminate with a
1020 non-zero exit status and create a crash report in the top level of the
1021 Git repository it was importing into. Crash reports contain a snapshot
1022 of the internal fast-import state as well as the most recent commands
1023 that lead up to the crash.
1024 All recent commands (including stream comments, file changes and
1026 progress commands) are shown in the command history within the crash
1027 report, but raw file data and commit messages are excluded from the
1028 crash report. This exclusion saves space within the report file and
1029 reduces the amount of buffering that fast-import must perform during
1030 execution.
1031 After writing a crash report fast-import will close the current
1033 packfile and export the marks table. This allows the frontend developer
1034 to inspect the repository state and resume the import from the point
1035 where it crashed. The modified branches and tags are not updated during
1036 a crash, as the import did not complete successfully. Branch and tag
1037 information can be found in the crash report and must be applied
1038 manually if the update is needed.
1039 An example crash:
1041 $ cat >in <<END_OF_INPUT
1043 # my very first test commit
1044 commit refs/heads/master
1045 committer Shawn O. Pearce <spearce> 19283 -0400
1046 # who is that guy anyway?
1047 data <<EOF
1048 this is my commit
1049 EOF
1050 M 644 inline .gitignore
1051 data <<EOF
1052 .gitignore
1053 EOF
1054 M 777 inline bob
1055 END_OF_INPUT
1056 $ git fast-import <in
1058 fatal: Corrupt mode: M 777 inline bob
1059 fast-import: dumping crash report to .git/fast_import_crash_8434
1060 $ cat .git/fast_import_crash_8434
1062 fast-import crash report:
1063 fast-import process: 8434
1064 parent process : 1391
1065 at Sat Sep 1 00:58:12 2007
1066 fatal: Corrupt mode: M 777 inline bob
1068 Most Recent Commands Before Crash
1070 ---------------------------------
1071 # my very first test commit
1072 commit refs/heads/master
1073 committer Shawn O. Pearce <spearce> 19283 -0400
1074 # who is that guy anyway?
1075 data <<EOF
1076 M 644 inline .gitignore
1077 data <<EOF
1078 * M 777 inline bob
1079 Active Branch LRU
1081 -----------------
1082 active_branches = 1 cur, 5 max
1083 pos clock name
1085 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1086 1) 0 refs/heads/master
1087 Inactive Branches
1089 -----------------
1090 refs/heads/master:
1091 status : active loaded dirty
1092 tip commit : 0000000000000000000000000000000000000000
1093 old tree : 0000000000000000000000000000000000000000
1094 cur tree : 0000000000000000000000000000000000000000
1095 commit clock: 0
1096 last pack :
1097 -------------------
1099 END OF CRASH REPORT
1100
1102 The following tips and tricks have been collected from various users of
1103 fast-import, and are offered here as suggestions.
1104 Use One Mark Per Commit
1106 When doing a repository conversion, use a unique mark per commit (mark
1107 :<n>) and supply the --export-marks option on the command line.
1108 fast-import will dump a file which lists every mark and the Git object
1109 SHA-1 that corresponds to it. If the frontend can tie the marks back to
1110 the source repository, it is easy to verify the accuracy and
1111 completeness of the import by comparing each Git commit to the
1112 corresponding source revision.
1113 Coming from a system such as Perforce or Subversion this should be
1115 quite simple, as the fast-import mark can also be the Perforce
1116 changeset number or the Subversion revision number.
1117 Freely Skip Around Branches
1119 Don’t bother trying to optimize the frontend to stick to one branch at
1120 a time during an import. Although doing so might be slightly faster for
1121 fast-import, it tends to increase the complexity of the frontend code
1122 considerably.
1123 The branch LRU builtin to fast-import tends to behave very well, and
1125 the cost of activating an inactive branch is so low that bouncing
1126 around between branches has virtually no impact on import performance.
1127 Handling Renames
1129 When importing a renamed file or directory, simply delete the old
1130 name(s) and modify the new name(s) during the corresponding commit. Git
1131 performs rename detection after-the-fact, rather than explicitly during
1132 a commit.
1133 Use Tag Fixup Branches
1135 Some other SCM systems let the user create a tag from multiple files
1136 which are not from the same commit/changeset. Or to create tags which
1137 are a subset of the files available in the repository.
1138 Importing these tags as-is in Git is impossible without making at least
1140 one commit which “fixes up” the files to match the content of the tag.
1141 Use fast-import’s reset command to reset a dummy branch outside of your
1142 normal branch space to the base commit for the tag, then commit one or
1143 more file fixup commits, and finally tag the dummy branch.
1144 For example since all normal branches are stored under refs/heads/ name
1146 the tag fixup branch TAG_FIXUP. This way it is impossible for the fixup
1147 branch used by the importer to have namespace conflicts with real
1148 branches imported from the source (the name TAG_FIXUP is not
1149 refs/heads/TAG_FIXUP).
1150 When committing fixups, consider using merge to connect the commit(s)
1152 which are supplying file revisions to the fixup branch. Doing so will
1153 allow tools such as git blame to track through the real commit history
1154 and properly annotate the source files.
1155 After fast-import terminates the frontend will need to do rm
1157 .git/TAG_FIXUP to remove the dummy branch.
1158 Import Now, Repack Later
1160 As soon as fast-import completes the Git repository is completely valid
1161 and ready for use. Typically this takes only a very short time, even
1162 for considerably large projects (100,000+ commits).
1163 However repacking the repository is necessary to improve data locality
1165 and access performance. It can also take hours on extremely large
1166 projects (especially if -f and a large --window parameter is used).
1167 Since repacking is safe to run alongside readers and writers, run the
1168 repack in the background and let it finish when it finishes. There is
1169 no reason to wait to explore your new Git project!
1170 If you choose to wait for the repack, don’t try to run benchmarks or
1172 performance tests until repacking is completed. fast-import outputs
1173 suboptimal packfiles that are simply never seen in real use situations.
1174 Repacking Historical Data
1176 If you are repacking very old imported data (e.g. older than the last
1177 year), consider expending some extra CPU time and supplying --window=50
1178 (or higher) when you run git repack. This will take longer, but will
1179 also produce a smaller packfile. You only need to expend the effort
1180 once, and everyone using your project will benefit from the smaller
1181 repository.
1182 Include Some Progress Messages
1184 Every once in a while have your frontend emit a progress message to
1185 fast-import. The contents of the messages are entirely free-form, so
1186 one suggestion would be to output the current month and year each time
1187 the current commit date moves into the next month. Your users will feel
1188 better knowing how much of the data stream has been processed.
1189
1191 When packing a blob fast-import always attempts to deltify against the
1192 last blob written. Unless specifically arranged for by the frontend,
1193 this will probably not be a prior version of the same file, so the
1194 generated delta will not be the smallest possible. The resulting
1195 packfile will be compressed, but will not be optimal.
1196 Frontends which have efficient access to all revisions of a single file
1198 (for example reading an RCS/CVS ,v file) can choose to supply all
1199 revisions of that file as a sequence of consecutive blob commands. This
1200 allows fast-import to deltify the different file revisions against each
1201 other, saving space in the final packfile. Marks can be used to later
1202 identify individual file revisions during a sequence of commit
1203 commands.
1204 The packfile(s) created by fast-import do not encourage good disk
1206 access patterns. This is caused by fast-import writing the data in the
1207 order it is received on standard input, while Git typically organizes
1208 data within packfiles to make the most recent (current tip) data appear
1209 before historical data. Git also clusters commits together, speeding up
1210 revision traversal through better cache locality.
1211 For this reason it is strongly recommended that users repack the
1213 repository with git repack -a -d after fast-import completes, allowing
1214 Git to reorganize the packfiles for faster data access. If blob deltas
1215 are suboptimal (see above) then also adding the -f option to force
1216 recomputation of all deltas can significantly reduce the final packfile
1217 size (30-50% smaller can be quite typical).
1218
1220 There are a number of factors which affect how much memory fast-import
1221 requires to perform an import. Like critical sections of core Git,
1222 fast-import uses its own memory allocators to amortize any overheads
1223 associated with malloc. In practice fast-import tends to amortize any
1224 malloc overheads to 0, due to its use of large block allocations.
1225 per object
1227 fast-import maintains an in-memory structure for every object written
1228 in this execution. On a 32 bit system the structure is 32 bytes, on a
1229 64 bit system the structure is 40 bytes (due to the larger pointer
1230 sizes). Objects in the table are not deallocated until fast-import
1231 terminates. Importing 2 million objects on a 32 bit system will require
1232 approximately 64 MiB of memory.
1233 The object table is actually a hashtable keyed on the object name (the
1235 unique SHA-1). This storage configuration allows fast-import to reuse
1236 an existing or already written object and avoid writing duplicates to
1237 the output packfile. Duplicate blobs are surprisingly common in an
1238 import, typically due to branch merges in the source.
1239 per mark
1241 Marks are stored in a sparse array, using 1 pointer (4 bytes or 8
1242 bytes, depending on pointer size) per mark. Although the array is
1243 sparse, frontends are still strongly encouraged to use marks between 1
1244 and n, where n is the total number of marks required for this import.
1245 per branch
1247 Branches are classified as active and inactive. The memory usage of the
1248 two classes is significantly different.
1249 Inactive branches are stored in a structure which uses 96 or 120 bytes
1251 (32 bit or 64 bit systems, respectively), plus the length of the branch
1252 name (typically under 200 bytes), per branch. fast-import will easily
1253 handle as many as 10,000 inactive branches in under 2 MiB of memory.
1254 Active branches have the same overhead as inactive branches, but also
1256 contain copies of every tree that has been recently modified on that
1257 branch. If subtree include has not been modified since the branch
1258 became active, its contents will not be loaded into memory, but if
1259 subtree src has been modified by a commit since the branch became
1260 active, then its contents will be loaded in memory.
1261 As active branches store metadata about the files contained on that
1263 branch, their in-memory storage size can grow to a considerable size
1264 (see below).
1265 fast-import automatically moves active branches to inactive status
1267 based on a simple least-recently-used algorithm. The LRU chain is
1268 updated on each commit command. The maximum number of active branches
1269 can be increased or decreased on the command line with
1270 --active-branches=.
1271 per active tree
1273 Trees (aka directories) use just 12 bytes of memory on top of the
1274 memory required for their entries (see “per active file” below). The
1275 cost of a tree is virtually 0, as its overhead amortizes out over the
1276 individual file entries.
1277 per active file entry
1279 Files (and pointers to subtrees) within active trees require 52 or 64
1280 bytes (32/64 bit platforms) per entry. To conserve space, file and tree
1281 names are pooled in a common string table, allowing the filename
1282 “Makefile” to use just 16 bytes (after including the string header
1283 overhead) no matter how many times it occurs within the project.
1284 The active branch LRU, when coupled with the filename string pool and
1286 lazy loading of subtrees, allows fast-import to efficiently import
1287 projects with 2,000+ branches and 45,114+ files in a very limited
1288 memory footprint (less than 2.7 MiB per active branch).
1289
1291 Sending SIGUSR1 to the git fast-import process ends the current
1292 packfile early, simulating a checkpoint command. The impatient operator
1293 can use this facility to peek at the objects and refs from an import in
1294 progress, at the cost of some added running time and worse compression.
1295
1297 Part of the git(1) suite
1298Git 1.8.3.1 11/19/2018 GIT-FAST-IMPORT(1)