tests: fix cleanup after tests in t1509-root-worktree
[git/git.git] / Documentation / git-fast-import.txt
1 git-fast-import(1)
2 ==================
3
4 NAME
5 ----
6 git-fast-import - Backend for fast Git data importers
7
8
9 SYNOPSIS
10 --------
11 [verse]
12 frontend | 'git fast-import' [options]
13
14 DESCRIPTION
15 -----------
16 This program is usually not what the end user wants to run directly.
17 Most end users want to use one of the existing frontend programs,
18 which parses a specific type of foreign source and feeds the contents
19 stored there to 'git fast-import'.
20
21 fast-import reads a mixed command/data stream from standard input and
22 writes one or more packfiles directly into the current repository.
23 When EOF is received on standard input, fast import writes out
24 updated branch and tag refs, fully updating the current repository
25 with the newly imported data.
26
27 The fast-import backend itself can import into an empty repository (one that
28 has already been initialized by 'git init') or incrementally
29 update an existing populated repository. Whether or not incremental
30 imports are supported from a particular foreign source depends on
31 the frontend program in use.
32
33
34 OPTIONS
35 -------
36
37 --force::
38 Force updating modified existing branches, even if doing
39 so would cause commits to be lost (as the new commit does
40 not contain the old commit).
41
42 --quiet::
43 Disable all non-fatal output, making fast-import silent when it
44 is successful. This option disables the output shown by
45 --stats.
46
47 --stats::
48 Display some basic statistics about the objects fast-import has
49 created, the packfiles they were stored into, and the
50 memory used by fast-import during this run. Showing this output
51 is currently the default, but can be disabled with --quiet.
52
53 Options for Frontends
54 ~~~~~~~~~~~~~~~~~~~~~
55
56 --cat-blob-fd=<fd>::
57 Write responses to `cat-blob` and `ls` queries to the
58 file descriptor <fd> instead of `stdout`. Allows `progress`
59 output intended for the end-user to be separated from other
60 output.
61
62 --date-format=<fmt>::
63 Specify the type of dates the frontend will supply to
64 fast-import within `author`, `committer` and `tagger` commands.
65 See ``Date Formats'' below for details about which formats
66 are supported, and their syntax.
67
68 --done::
69 Terminate with error if there is no `done` command at the end of
70 the stream. This option might be useful for detecting errors
71 that cause the frontend to terminate before it has started to
72 write a stream.
73
74 Locations of Marks Files
75 ~~~~~~~~~~~~~~~~~~~~~~~~
76
77 --export-marks=<file>::
78 Dumps the internal marks table to <file> when complete.
79 Marks are written one per line as `:markid SHA-1`.
80 Frontends can use this file to validate imports after they
81 have been completed, or to save the marks table across
82 incremental runs. As <file> is only opened and truncated
83 at checkpoint (or completion) the same path can also be
84 safely given to --import-marks.
85
86 --import-marks=<file>::
87 Before processing any input, load the marks specified in
88 <file>. The input file must exist, must be readable, and
89 must use the same format as produced by --export-marks.
90 Multiple options may be supplied to import more than one
91 set of marks. If a mark is defined to different values,
92 the last file wins.
93
94 --import-marks-if-exists=<file>::
95 Like --import-marks but instead of erroring out, silently
96 skips the file if it does not exist.
97
98 --[no-]relative-marks::
99 After specifying --relative-marks the paths specified
100 with --import-marks= and --export-marks= are relative
101 to an internal directory in the current repository.
102 In git-fast-import this means that the paths are relative
103 to the .git/info/fast-import directory. However, other
104 importers may use a different location.
105 +
106 Relative and non-relative marks may be combined by interweaving
107 --(no-)-relative-marks with the --(import|export)-marks= options.
108
109 Performance and Compression Tuning
110 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
111
112 --active-branches=<n>::
113 Maximum number of branches to maintain active at once.
114 See ``Memory Utilization'' below for details. Default is 5.
115
116 --big-file-threshold=<n>::
117 Maximum size of a blob that fast-import will attempt to
118 create a delta for, expressed in bytes. The default is 512m
119 (512 MiB). Some importers may wish to lower this on systems
120 with constrained memory.
121
122 --depth=<n>::
123 Maximum delta depth, for blob and tree deltification.
124 Default is 10.
125
126 --export-pack-edges=<file>::
127 After creating a packfile, print a line of data to
128 <file> listing the filename of the packfile and the last
129 commit on each branch that was written to that packfile.
130 This information may be useful after importing projects
131 whose total object set exceeds the 4 GiB packfile limit,
132 as these commits can be used as edge points during calls
133 to 'git pack-objects'.
134
135 --max-pack-size=<n>::
136 Maximum size of each output packfile.
137 The default is unlimited.
138
139
140 Performance
141 -----------
142 The design of fast-import allows it to import large projects in a minimum
143 amount of memory usage and processing time. Assuming the frontend
144 is able to keep up with fast-import and feed it a constant stream of data,
145 import times for projects holding 10+ years of history and containing
146 100,000+ individual commits are generally completed in just 1-2
147 hours on quite modest (~$2,000 USD) hardware.
148
149 Most bottlenecks appear to be in foreign source data access (the
150 source just cannot extract revisions fast enough) or disk IO (fast-import
151 writes as fast as the disk will take the data). Imports will run
152 faster if the source data is stored on a different drive than the
153 destination Git repository (due to less IO contention).
154
155
156 Development Cost
157 ----------------
158 A typical frontend for fast-import tends to weigh in at approximately 200
159 lines of Perl/Python/Ruby code. Most developers have been able to
160 create working importers in just a couple of hours, even though it
161 is their first exposure to fast-import, and sometimes even to Git. This is
162 an ideal situation, given that most conversion tools are throw-away
163 (use once, and never look back).
164
165
166 Parallel Operation
167 ------------------
168 Like 'git push' or 'git fetch', imports handled by fast-import are safe to
169 run alongside parallel `git repack -a -d` or `git gc` invocations,
170 or any other Git operation (including 'git prune', as loose objects
171 are never used by fast-import).
172
173 fast-import does not lock the branch or tag refs it is actively importing.
174 After the import, during its ref update phase, fast-import tests each
175 existing branch ref to verify the update will be a fast-forward
176 update (the commit stored in the ref is contained in the new
177 history of the commit to be written). If the update is not a
178 fast-forward update, fast-import will skip updating that ref and instead
179 prints a warning message. fast-import will always attempt to update all
180 branch refs, and does not stop on the first failure.
181
182 Branch updates can be forced with --force, but it's recommended that
183 this only be used on an otherwise quiet repository. Using --force
184 is not necessary for an initial import into an empty repository.
185
186
187 Technical Discussion
188 --------------------
189 fast-import tracks a set of branches in memory. Any branch can be created
190 or modified at any point during the import process by sending a
191 `commit` command on the input stream. This design allows a frontend
192 program to process an unlimited number of branches simultaneously,
193 generating commits in the order they are available from the source
194 data. It also simplifies the frontend programs considerably.
195
196 fast-import does not use or alter the current working directory, or any
197 file within it. (It does however update the current Git repository,
198 as referenced by `GIT_DIR`.) Therefore an import frontend may use
199 the working directory for its own purposes, such as extracting file
200 revisions from the foreign source. This ignorance of the working
201 directory also allows fast-import to run very quickly, as it does not
202 need to perform any costly file update operations when switching
203 between branches.
204
205 Input Format
206 ------------
207 With the exception of raw file data (which Git does not interpret)
208 the fast-import input format is text (ASCII) based. This text based
209 format simplifies development and debugging of frontend programs,
210 especially when a higher level language such as Perl, Python or
211 Ruby is being used.
212
213 fast-import is very strict about its input. Where we say SP below we mean
214 *exactly* one space. Likewise LF means one (and only one) linefeed
215 and HT one (and only one) horizontal tab.
216 Supplying additional whitespace characters will cause unexpected
217 results, such as branch names or file names with leading or trailing
218 spaces in their name, or early termination of fast-import when it encounters
219 unexpected input.
220
221 Stream Comments
222 ~~~~~~~~~~~~~~~
223 To aid in debugging frontends fast-import ignores any line that
224 begins with `#` (ASCII pound/hash) up to and including the line
225 ending `LF`. A comment line may contain any sequence of bytes
226 that does not contain an LF and therefore may be used to include
227 any detailed debugging information that might be specific to the
228 frontend and useful when inspecting a fast-import data stream.
229
230 Date Formats
231 ~~~~~~~~~~~~
232 The following date formats are supported. A frontend should select
233 the format it will use for this import by passing the format name
234 in the --date-format=<fmt> command-line option.
235
236 `raw`::
237 This is the Git native format and is `<time> SP <offutc>`.
238 It is also fast-import's default format, if --date-format was
239 not specified.
240 +
241 The time of the event is specified by `<time>` as the number of
242 seconds since the UNIX epoch (midnight, Jan 1, 1970, UTC) and is
243 written as an ASCII decimal integer.
244 +
245 The local offset is specified by `<offutc>` as a positive or negative
246 offset from UTC. For example EST (which is 5 hours behind UTC)
247 would be expressed in `<tz>` by ``-0500'' while UTC is ``+0000''.
248 The local offset does not affect `<time>`; it is used only as an
249 advisement to help formatting routines display the timestamp.
250 +
251 If the local offset is not available in the source material, use
252 ``+0000'', or the most common local offset. For example many
253 organizations have a CVS repository which has only ever been accessed
254 by users who are located in the same location and time zone. In this
255 case a reasonable offset from UTC could be assumed.
256 +
257 Unlike the `rfc2822` format, this format is very strict. Any
258 variation in formatting will cause fast-import to reject the value.
259
260 `rfc2822`::
261 This is the standard email format as described by RFC 2822.
262 +
263 An example value is ``Tue Feb 6 11:22:18 2007 -0500''. The Git
264 parser is accurate, but a little on the lenient side. It is the
265 same parser used by 'git am' when applying patches
266 received from email.
267 +
268 Some malformed strings may be accepted as valid dates. In some of
269 these cases Git will still be able to obtain the correct date from
270 the malformed string. There are also some types of malformed
271 strings which Git will parse wrong, and yet consider valid.
272 Seriously malformed strings will be rejected.
273 +
274 Unlike the `raw` format above, the time zone/UTC offset information
275 contained in an RFC 2822 date string is used to adjust the date
276 value to UTC prior to storage. Therefore it is important that
277 this information be as accurate as possible.
278 +
279 If the source material uses RFC 2822 style dates,
280 the frontend should let fast-import handle the parsing and conversion
281 (rather than attempting to do it itself) as the Git parser has
282 been well tested in the wild.
283 +
284 Frontends should prefer the `raw` format if the source material
285 already uses UNIX-epoch format, can be coaxed to give dates in that
286 format, or its format is easily convertible to it, as there is no
287 ambiguity in parsing.
288
289 `now`::
290 Always use the current time and time zone. The literal
291 `now` must always be supplied for `<when>`.
292 +
293 This is a toy format. The current time and time zone of this system
294 is always copied into the identity string at the time it is being
295 created by fast-import. There is no way to specify a different time or
296 time zone.
297 +
298 This particular format is supplied as it's short to implement and
299 may be useful to a process that wants to create a new commit
300 right now, without needing to use a working directory or
301 'git update-index'.
302 +
303 If separate `author` and `committer` commands are used in a `commit`
304 the timestamps may not match, as the system clock will be polled
305 twice (once for each command). The only way to ensure that both
306 author and committer identity information has the same timestamp
307 is to omit `author` (thus copying from `committer`) or to use a
308 date format other than `now`.
309
310 Commands
311 ~~~~~~~~
312 fast-import accepts several commands to update the current repository
313 and control the current import process. More detailed discussion
314 (with examples) of each command follows later.
315
316 `commit`::
317 Creates a new branch or updates an existing branch by
318 creating a new commit and updating the branch to point at
319 the newly created commit.
320
321 `tag`::
322 Creates an annotated tag object from an existing commit or
323 branch. Lightweight tags are not supported by this command,
324 as they are not recommended for recording meaningful points
325 in time.
326
327 `reset`::
328 Reset an existing branch (or a new branch) to a specific
329 revision. This command must be used to change a branch to
330 a specific revision without making a commit on it.
331
332 `blob`::
333 Convert raw file data into a blob, for future use in a
334 `commit` command. This command is optional and is not
335 needed to perform an import.
336
337 `checkpoint`::
338 Forces fast-import to close the current packfile, generate its
339 unique SHA-1 checksum and index, and start a new packfile.
340 This command is optional and is not needed to perform
341 an import.
342
343 `progress`::
344 Causes fast-import to echo the entire line to its own
345 standard output. This command is optional and is not needed
346 to perform an import.
347
348 `done`::
349 Marks the end of the stream. This command is optional
350 unless the `done` feature was requested using the
351 `--done` command-line option or `feature done` command.
352
353 `cat-blob`::
354 Causes fast-import to print a blob in 'cat-file --batch'
355 format to the file descriptor set with `--cat-blob-fd` or
356 `stdout` if unspecified.
357
358 `ls`::
359 Causes fast-import to print a line describing a directory
360 entry in 'ls-tree' format to the file descriptor set with
361 `--cat-blob-fd` or `stdout` if unspecified.
362
363 `feature`::
364 Enable the specified feature. This requires that fast-import
365 supports the specified feature, and aborts if it does not.
366
367 `option`::
368 Specify any of the options listed under OPTIONS that do not
369 change stream semantic to suit the frontend's needs. This
370 command is optional and is not needed to perform an import.
371
372 `commit`
373 ~~~~~~~~
374 Create or update a branch with a new commit, recording one logical
375 change to the project.
376
377 ....
378 'commit' SP <ref> LF
379 mark?
380 ('author' (SP <name>)? SP LT <email> GT SP <when> LF)?
381 'committer' (SP <name>)? SP LT <email> GT SP <when> LF
382 data
383 ('from' SP <commit-ish> LF)?
384 ('merge' SP <commit-ish> LF)?
385 (filemodify | filedelete | filecopy | filerename | filedeleteall | notemodify)*
386 LF?
387 ....
388
389 where `<ref>` is the name of the branch to make the commit on.
390 Typically branch names are prefixed with `refs/heads/` in
391 Git, so importing the CVS branch symbol `RELENG-1_0` would use
392 `refs/heads/RELENG-1_0` for the value of `<ref>`. The value of
393 `<ref>` must be a valid refname in Git. As `LF` is not valid in
394 a Git refname, no quoting or escaping syntax is supported here.
395
396 A `mark` command may optionally appear, requesting fast-import to save a
397 reference to the newly created commit for future use by the frontend
398 (see below for format). It is very common for frontends to mark
399 every commit they create, thereby allowing future branch creation
400 from any imported commit.
401
402 The `data` command following `committer` must supply the commit
403 message (see below for `data` command syntax). To import an empty
404 commit message use a 0 length data. Commit messages are free-form
405 and are not interpreted by Git. Currently they must be encoded in
406 UTF-8, as fast-import does not permit other encodings to be specified.
407
408 Zero or more `filemodify`, `filedelete`, `filecopy`, `filerename`,
409 `filedeleteall` and `notemodify` commands
410 may be included to update the contents of the branch prior to
411 creating the commit. These commands may be supplied in any order.
412 However it is recommended that a `filedeleteall` command precede
413 all `filemodify`, `filecopy`, `filerename` and `notemodify` commands in
414 the same commit, as `filedeleteall` wipes the branch clean (see below).
415
416 The `LF` after the command is optional (it used to be required).
417
418 `author`
419 ^^^^^^^^
420 An `author` command may optionally appear, if the author information
421 might differ from the committer information. If `author` is omitted
422 then fast-import will automatically use the committer's information for
423 the author portion of the commit. See below for a description of
424 the fields in `author`, as they are identical to `committer`.
425
426 `committer`
427 ^^^^^^^^^^^
428 The `committer` command indicates who made this commit, and when
429 they made it.
430
431 Here `<name>` is the person's display name (for example
432 ``Com M Itter'') and `<email>` is the person's email address
433 (``\cm@example.com''). `LT` and `GT` are the literal less-than (\x3c)
434 and greater-than (\x3e) symbols. These are required to delimit
435 the email address from the other fields in the line. Note that
436 `<name>` and `<email>` are free-form and may contain any sequence
437 of bytes, except `LT`, `GT` and `LF`. `<name>` is typically UTF-8 encoded.
438
439 The time of the change is specified by `<when>` using the date format
440 that was selected by the --date-format=<fmt> command-line option.
441 See ``Date Formats'' above for the set of supported formats, and
442 their syntax.
443
444 `from`
445 ^^^^^^
446 The `from` command is used to specify the commit to initialize
447 this branch from. This revision will be the first ancestor of the
448 new commit. The state of the tree built at this commit will begin
449 with the state at the `from` commit, and be altered by the content
450 modifications in this commit.
451
452 Omitting the `from` command in the first commit of a new branch
453 will cause fast-import to create that commit with no ancestor. This
454 tends to be desired only for the initial commit of a project.
455 If the frontend creates all files from scratch when making a new
456 branch, a `merge` command may be used instead of `from` to start
457 the commit with an empty tree.
458 Omitting the `from` command on existing branches is usually desired,
459 as the current commit on that branch is automatically assumed to
460 be the first ancestor of the new commit.
461
462 As `LF` is not valid in a Git refname or SHA-1 expression, no
463 quoting or escaping syntax is supported within `<commit-ish>`.
464
465 Here `<commit-ish>` is any of the following:
466
467 * The name of an existing branch already in fast-import's internal branch
468 table. If fast-import doesn't know the name, it's treated as a SHA-1
469 expression.
470
471 * A mark reference, `:<idnum>`, where `<idnum>` is the mark number.
472 +
473 The reason fast-import uses `:` to denote a mark reference is this character
474 is not legal in a Git branch name. The leading `:` makes it easy
475 to distinguish between the mark 42 (`:42`) and the branch 42 (`42`
476 or `refs/heads/42`), or an abbreviated SHA-1 which happened to
477 consist only of base-10 digits.
478 +
479 Marks must be declared (via `mark`) before they can be used.
480
481 * A complete 40 byte or abbreviated commit SHA-1 in hex.
482
483 * Any valid Git SHA-1 expression that resolves to a commit. See
484 ``SPECIFYING REVISIONS'' in linkgit:gitrevisions[7] for details.
485
486 * The special null SHA-1 (40 zeros) specifies that the branch is to be
487 removed.
488
489 The special case of restarting an incremental import from the
490 current branch value should be written as:
491 ----
492 from refs/heads/branch^0
493 ----
494 The `^0` suffix is necessary as fast-import does not permit a branch to
495 start from itself, and the branch is created in memory before the
496 `from` command is even read from the input. Adding `^0` will force
497 fast-import to resolve the commit through Git's revision parsing library,
498 rather than its internal branch table, thereby loading in the
499 existing value of the branch.
500
501 `merge`
502 ^^^^^^^
503 Includes one additional ancestor commit. The additional ancestry
504 link does not change the way the tree state is built at this commit.
505 If the `from` command is
506 omitted when creating a new branch, the first `merge` commit will be
507 the first ancestor of the current commit, and the branch will start
508 out with no files. An unlimited number of `merge` commands per
509 commit are permitted by fast-import, thereby establishing an n-way merge.
510
511 Here `<commit-ish>` is any of the commit specification expressions
512 also accepted by `from` (see above).
513
514 `filemodify`
515 ^^^^^^^^^^^^
516 Included in a `commit` command to add a new file or change the
517 content of an existing file. This command has two different means
518 of specifying the content of the file.
519
520 External data format::
521 The data content for the file was already supplied by a prior
522 `blob` command. The frontend just needs to connect it.
523 +
524 ....
525 'M' SP <mode> SP <dataref> SP <path> LF
526 ....
527 +
528 Here usually `<dataref>` must be either a mark reference (`:<idnum>`)
529 set by a prior `blob` command, or a full 40-byte SHA-1 of an
530 existing Git blob object. If `<mode>` is `040000`` then
531 `<dataref>` must be the full 40-byte SHA-1 of an existing
532 Git tree object or a mark reference set with `--import-marks`.
533
534 Inline data format::
535 The data content for the file has not been supplied yet.
536 The frontend wants to supply it as part of this modify
537 command.
538 +
539 ....
540 'M' SP <mode> SP 'inline' SP <path> LF
541 data
542 ....
543 +
544 See below for a detailed description of the `data` command.
545
546 In both formats `<mode>` is the type of file entry, specified
547 in octal. Git only supports the following modes:
548
549 * `100644` or `644`: A normal (not-executable) file. The majority
550 of files in most projects use this mode. If in doubt, this is
551 what you want.
552 * `100755` or `755`: A normal, but executable, file.
553 * `120000`: A symlink, the content of the file will be the link target.
554 * `160000`: A gitlink, SHA-1 of the object refers to a commit in
555 another repository. Git links can only be specified by SHA or through
556 a commit mark. They are used to implement submodules.
557 * `040000`: A subdirectory. Subdirectories can only be specified by
558 SHA or through a tree mark set with `--import-marks`.
559
560 In both formats `<path>` is the complete path of the file to be added
561 (if not already existing) or modified (if already existing).
562
563 A `<path>` string must use UNIX-style directory separators (forward
564 slash `/`), may contain any byte other than `LF`, and must not
565 start with double quote (`"`).
566
567 A path can use C-style string quoting; this is accepted in all cases
568 and mandatory if the filename starts with double quote or contains
569 `LF`. In C-style quoting, the complete name should be surrounded with
570 double quotes, and any `LF`, backslash, or double quote characters
571 must be escaped by preceding them with a backslash (e.g.,
572 `"path/with\n, \\ and \" in it"`).
573
574 The value of `<path>` must be in canonical form. That is it must not:
575
576 * contain an empty directory component (e.g. `foo//bar` is invalid),
577 * end with a directory separator (e.g. `foo/` is invalid),
578 * start with a directory separator (e.g. `/foo` is invalid),
579 * contain the special component `.` or `..` (e.g. `foo/./bar` and
580 `foo/../bar` are invalid).
581
582 The root of the tree can be represented by an empty string as `<path>`.
583
584 It is recommended that `<path>` always be encoded using UTF-8.
585
586 `filedelete`
587 ^^^^^^^^^^^^
588 Included in a `commit` command to remove a file or recursively
589 delete an entire directory from the branch. If the file or directory
590 removal makes its parent directory empty, the parent directory will
591 be automatically removed too. This cascades up the tree until the
592 first non-empty directory or the root is reached.
593
594 ....
595 'D' SP <path> LF
596 ....
597
598 here `<path>` is the complete path of the file or subdirectory to
599 be removed from the branch.
600 See `filemodify` above for a detailed description of `<path>`.
601
602 `filecopy`
603 ^^^^^^^^^^
604 Recursively copies an existing file or subdirectory to a different
605 location within the branch. The existing file or directory must
606 exist. If the destination exists it will be completely replaced
607 by the content copied from the source.
608
609 ....
610 'C' SP <path> SP <path> LF
611 ....
612
613 here the first `<path>` is the source location and the second
614 `<path>` is the destination. See `filemodify` above for a detailed
615 description of what `<path>` may look like. To use a source path
616 that contains SP the path must be quoted.
617
618 A `filecopy` command takes effect immediately. Once the source
619 location has been copied to the destination any future commands
620 applied to the source location will not impact the destination of
621 the copy.
622
623 `filerename`
624 ^^^^^^^^^^^^
625 Renames an existing file or subdirectory to a different location
626 within the branch. The existing file or directory must exist. If
627 the destination exists it will be replaced by the source directory.
628
629 ....
630 'R' SP <path> SP <path> LF
631 ....
632
633 here the first `<path>` is the source location and the second
634 `<path>` is the destination. See `filemodify` above for a detailed
635 description of what `<path>` may look like. To use a source path
636 that contains SP the path must be quoted.
637
638 A `filerename` command takes effect immediately. Once the source
639 location has been renamed to the destination any future commands
640 applied to the source location will create new files there and not
641 impact the destination of the rename.
642
643 Note that a `filerename` is the same as a `filecopy` followed by a
644 `filedelete` of the source location. There is a slight performance
645 advantage to using `filerename`, but the advantage is so small
646 that it is never worth trying to convert a delete/add pair in
647 source material into a rename for fast-import. This `filerename`
648 command is provided just to simplify frontends that already have
649 rename information and don't want bother with decomposing it into a
650 `filecopy` followed by a `filedelete`.
651
652 `filedeleteall`
653 ^^^^^^^^^^^^^^^
654 Included in a `commit` command to remove all files (and also all
655 directories) from the branch. This command resets the internal
656 branch structure to have no files in it, allowing the frontend
657 to subsequently add all interesting files from scratch.
658
659 ....
660 'deleteall' LF
661 ....
662
663 This command is extremely useful if the frontend does not know
664 (or does not care to know) what files are currently on the branch,
665 and therefore cannot generate the proper `filedelete` commands to
666 update the content.
667
668 Issuing a `filedeleteall` followed by the needed `filemodify`
669 commands to set the correct content will produce the same results
670 as sending only the needed `filemodify` and `filedelete` commands.
671 The `filedeleteall` approach may however require fast-import to use slightly
672 more memory per active branch (less than 1 MiB for even most large
673 projects); so frontends that can easily obtain only the affected
674 paths for a commit are encouraged to do so.
675
676 `notemodify`
677 ^^^^^^^^^^^^
678 Included in a `commit` `<notes_ref>` command to add a new note
679 annotating a `<commit-ish>` or change this annotation contents.
680 Internally it is similar to filemodify 100644 on `<commit-ish>`
681 path (maybe split into subdirectories). It's not advised to
682 use any other commands to write to the `<notes_ref>` tree except
683 `filedeleteall` to delete all existing notes in this tree.
684 This command has two different means of specifying the content
685 of the note.
686
687 External data format::
688 The data content for the note was already supplied by a prior
689 `blob` command. The frontend just needs to connect it to the
690 commit that is to be annotated.
691 +
692 ....
693 'N' SP <dataref> SP <commit-ish> LF
694 ....
695 +
696 Here `<dataref>` can be either a mark reference (`:<idnum>`)
697 set by a prior `blob` command, or a full 40-byte SHA-1 of an
698 existing Git blob object.
699
700 Inline data format::
701 The data content for the note has not been supplied yet.
702 The frontend wants to supply it as part of this modify
703 command.
704 +
705 ....
706 'N' SP 'inline' SP <commit-ish> LF
707 data
708 ....
709 +
710 See below for a detailed description of the `data` command.
711
712 In both formats `<commit-ish>` is any of the commit specification
713 expressions also accepted by `from` (see above).
714
715 `mark`
716 ~~~~~~
717 Arranges for fast-import to save a reference to the current object, allowing
718 the frontend to recall this object at a future point in time, without
719 knowing its SHA-1. Here the current object is the object creation
720 command the `mark` command appears within. This can be `commit`,
721 `tag`, and `blob`, but `commit` is the most common usage.
722
723 ....
724 'mark' SP ':' <idnum> LF
725 ....
726
727 where `<idnum>` is the number assigned by the frontend to this mark.
728 The value of `<idnum>` is expressed as an ASCII decimal integer.
729 The value 0 is reserved and cannot be used as
730 a mark. Only values greater than or equal to 1 may be used as marks.
731
732 New marks are created automatically. Existing marks can be moved
733 to another object simply by reusing the same `<idnum>` in another
734 `mark` command.
735
736 `tag`
737 ~~~~~
738 Creates an annotated tag referring to a specific commit. To create
739 lightweight (non-annotated) tags see the `reset` command below.
740
741 ....
742 'tag' SP <name> LF
743 'from' SP <commit-ish> LF
744 'tagger' (SP <name>)? SP LT <email> GT SP <when> LF
745 data
746 ....
747
748 where `<name>` is the name of the tag to create.
749
750 Tag names are automatically prefixed with `refs/tags/` when stored
751 in Git, so importing the CVS branch symbol `RELENG-1_0-FINAL` would
752 use just `RELENG-1_0-FINAL` for `<name>`, and fast-import will write the
753 corresponding ref as `refs/tags/RELENG-1_0-FINAL`.
754
755 The value of `<name>` must be a valid refname in Git and therefore
756 may contain forward slashes. As `LF` is not valid in a Git refname,
757 no quoting or escaping syntax is supported here.
758
759 The `from` command is the same as in the `commit` command; see
760 above for details.
761
762 The `tagger` command uses the same format as `committer` within
763 `commit`; again see above for details.
764
765 The `data` command following `tagger` must supply the annotated tag
766 message (see below for `data` command syntax). To import an empty
767 tag message use a 0 length data. Tag messages are free-form and are
768 not interpreted by Git. Currently they must be encoded in UTF-8,
769 as fast-import does not permit other encodings to be specified.
770
771 Signing annotated tags during import from within fast-import is not
772 supported. Trying to include your own PGP/GPG signature is not
773 recommended, as the frontend does not (easily) have access to the
774 complete set of bytes which normally goes into such a signature.
775 If signing is required, create lightweight tags from within fast-import with
776 `reset`, then create the annotated versions of those tags offline
777 with the standard 'git tag' process.
778
779 `reset`
780 ~~~~~~~
781 Creates (or recreates) the named branch, optionally starting from
782 a specific revision. The reset command allows a frontend to issue
783 a new `from` command for an existing branch, or to create a new
784 branch from an existing commit without creating a new commit.
785
786 ....
787 'reset' SP <ref> LF
788 ('from' SP <commit-ish> LF)?
789 LF?
790 ....
791
792 For a detailed description of `<ref>` and `<commit-ish>` see above
793 under `commit` and `from`.
794
795 The `LF` after the command is optional (it used to be required).
796
797 The `reset` command can also be used to create lightweight
798 (non-annotated) tags. For example:
799
800 ====
801 reset refs/tags/938
802 from :938
803 ====
804
805 would create the lightweight tag `refs/tags/938` referring to
806 whatever commit mark `:938` references.
807
808 `blob`
809 ~~~~~~
810 Requests writing one file revision to the packfile. The revision
811 is not connected to any commit; this connection must be formed in
812 a subsequent `commit` command by referencing the blob through an
813 assigned mark.
814
815 ....
816 'blob' LF
817 mark?
818 data
819 ....
820
821 The mark command is optional here as some frontends have chosen
822 to generate the Git SHA-1 for the blob on their own, and feed that
823 directly to `commit`. This is typically more work than it's worth
824 however, as marks are inexpensive to store and easy to use.
825
826 `data`
827 ~~~~~~
828 Supplies raw data (for use as blob/file content, commit messages, or
829 annotated tag messages) to fast-import. Data can be supplied using an exact
830 byte count or delimited with a terminating line. Real frontends
831 intended for production-quality conversions should always use the
832 exact byte count format, as it is more robust and performs better.
833 The delimited format is intended primarily for testing fast-import.
834
835 Comment lines appearing within the `<raw>` part of `data` commands
836 are always taken to be part of the body of the data and are therefore
837 never ignored by fast-import. This makes it safe to import any
838 file/message content whose lines might start with `#`.
839
840 Exact byte count format::
841 The frontend must specify the number of bytes of data.
842 +
843 ....
844 'data' SP <count> LF
845 <raw> LF?
846 ....
847 +
848 where `<count>` is the exact number of bytes appearing within
849 `<raw>`. The value of `<count>` is expressed as an ASCII decimal
850 integer. The `LF` on either side of `<raw>` is not
851 included in `<count>` and will not be included in the imported data.
852 +
853 The `LF` after `<raw>` is optional (it used to be required) but
854 recommended. Always including it makes debugging a fast-import
855 stream easier as the next command always starts in column 0
856 of the next line, even if `<raw>` did not end with an `LF`.
857
858 Delimited format::
859 A delimiter string is used to mark the end of the data.
860 fast-import will compute the length by searching for the delimiter.
861 This format is primarily useful for testing and is not
862 recommended for real data.
863 +
864 ....
865 'data' SP '<<' <delim> LF
866 <raw> LF
867 <delim> LF
868 LF?
869 ....
870 +
871 where `<delim>` is the chosen delimiter string. The string `<delim>`
872 must not appear on a line by itself within `<raw>`, as otherwise
873 fast-import will think the data ends earlier than it really does. The `LF`
874 immediately trailing `<raw>` is part of `<raw>`. This is one of
875 the limitations of the delimited format, it is impossible to supply
876 a data chunk which does not have an LF as its last byte.
877 +
878 The `LF` after `<delim> LF` is optional (it used to be required).
879
880 `checkpoint`
881 ~~~~~~~~~~~~
882 Forces fast-import to close the current packfile, start a new one, and to
883 save out all current branch refs, tags and marks.
884
885 ....
886 'checkpoint' LF
887 LF?
888 ....
889
890 Note that fast-import automatically switches packfiles when the current
891 packfile reaches --max-pack-size, or 4 GiB, whichever limit is
892 smaller. During an automatic packfile switch fast-import does not update
893 the branch refs, tags or marks.
894
895 As a `checkpoint` can require a significant amount of CPU time and
896 disk IO (to compute the overall pack SHA-1 checksum, generate the
897 corresponding index file, and update the refs) it can easily take
898 several minutes for a single `checkpoint` command to complete.
899
900 Frontends may choose to issue checkpoints during extremely large
901 and long running imports, or when they need to allow another Git
902 process access to a branch. However given that a 30 GiB Subversion
903 repository can be loaded into Git through fast-import in about 3 hours,
904 explicit checkpointing may not be necessary.
905
906 The `LF` after the command is optional (it used to be required).
907
908 `progress`
909 ~~~~~~~~~~
910 Causes fast-import to print the entire `progress` line unmodified to
911 its standard output channel (file descriptor 1) when the command is
912 processed from the input stream. The command otherwise has no impact
913 on the current import, or on any of fast-import's internal state.
914
915 ....
916 'progress' SP <any> LF
917 LF?
918 ....
919
920 The `<any>` part of the command may contain any sequence of bytes
921 that does not contain `LF`. The `LF` after the command is optional.
922 Callers may wish to process the output through a tool such as sed to
923 remove the leading part of the line, for example:
924
925 ====
926 frontend | git fast-import | sed 's/^progress //'
927 ====
928
929 Placing a `progress` command immediately after a `checkpoint` will
930 inform the reader when the `checkpoint` has been completed and it
931 can safely access the refs that fast-import updated.
932
933 `cat-blob`
934 ~~~~~~~~~~
935 Causes fast-import to print a blob to a file descriptor previously
936 arranged with the `--cat-blob-fd` argument. The command otherwise
937 has no impact on the current import; its main purpose is to
938 retrieve blobs that may be in fast-import's memory but not
939 accessible from the target repository.
940
941 ....
942 'cat-blob' SP <dataref> LF
943 ....
944
945 The `<dataref>` can be either a mark reference (`:<idnum>`)
946 set previously or a full 40-byte SHA-1 of a Git blob, preexisting or
947 ready to be written.
948
949 Output uses the same format as `git cat-file --batch`:
950
951 ====
952 <sha1> SP 'blob' SP <size> LF
953 <contents> LF
954 ====
955
956 This command can be used anywhere in the stream that comments are
957 accepted. In particular, the `cat-blob` command can be used in the
958 middle of a commit but not in the middle of a `data` command.
959
960 See ``Responses To Commands'' below for details about how to read
961 this output safely.
962
963 `ls`
964 ~~~~
965 Prints information about the object at a path to a file descriptor
966 previously arranged with the `--cat-blob-fd` argument. This allows
967 printing a blob from the active commit (with `cat-blob`) or copying a
968 blob or tree from a previous commit for use in the current one (with
969 `filemodify`).
970
971 The `ls` command can be used anywhere in the stream that comments are
972 accepted, including the middle of a commit.
973
974 Reading from the active commit::
975 This form can only be used in the middle of a `commit`.
976 The path names a directory entry within fast-import's
977 active commit. The path must be quoted in this case.
978 +
979 ....
980 'ls' SP <path> LF
981 ....
982
983 Reading from a named tree::
984 The `<dataref>` can be a mark reference (`:<idnum>`) or the
985 full 40-byte SHA-1 of a Git tag, commit, or tree object,
986 preexisting or waiting to be written.
987 The path is relative to the top level of the tree
988 named by `<dataref>`.
989 +
990 ....
991 'ls' SP <dataref> SP <path> LF
992 ....
993
994 See `filemodify` above for a detailed description of `<path>`.
995
996 Output uses the same format as `git ls-tree <tree> -- <path>`:
997
998 ====
999 <mode> SP ('blob' | 'tree' | 'commit') SP <dataref> HT <path> LF
1000 ====
1001
1002 The <dataref> represents the blob, tree, or commit object at <path>
1003 and can be used in later 'cat-blob', 'filemodify', or 'ls' commands.
1004
1005 If there is no file or subtree at that path, 'git fast-import' will
1006 instead report
1007
1008 ====
1009 missing SP <path> LF
1010 ====
1011
1012 See ``Responses To Commands'' below for details about how to read
1013 this output safely.
1014
1015 `feature`
1016 ~~~~~~~~~
1017 Require that fast-import supports the specified feature, or abort if
1018 it does not.
1019
1020 ....
1021 'feature' SP <feature> ('=' <argument>)? LF
1022 ....
1023
1024 The <feature> part of the command may be any one of the following:
1025
1026 date-format::
1027 export-marks::
1028 relative-marks::
1029 no-relative-marks::
1030 force::
1031 Act as though the corresponding command-line option with
1032 a leading '--' was passed on the command line
1033 (see OPTIONS, above).
1034
1035 import-marks::
1036 import-marks-if-exists::
1037 Like --import-marks except in two respects: first, only one
1038 "feature import-marks" or "feature import-marks-if-exists"
1039 command is allowed per stream; second, an --import-marks=
1040 or --import-marks-if-exists command-line option overrides
1041 any of these "feature" commands in the stream; third,
1042 "feature import-marks-if-exists" like a corresponding
1043 command-line option silently skips a nonexistent file.
1044
1045 cat-blob::
1046 ls::
1047 Require that the backend support the 'cat-blob' or 'ls' command.
1048 Versions of fast-import not supporting the specified command
1049 will exit with a message indicating so.
1050 This lets the import error out early with a clear message,
1051 rather than wasting time on the early part of an import
1052 before the unsupported command is detected.
1053
1054 notes::
1055 Require that the backend support the 'notemodify' (N)
1056 subcommand to the 'commit' command.
1057 Versions of fast-import not supporting notes will exit
1058 with a message indicating so.
1059
1060 done::
1061 Error out if the stream ends without a 'done' command.
1062 Without this feature, errors causing the frontend to end
1063 abruptly at a convenient point in the stream can go
1064 undetected. This may occur, for example, if an import
1065 front end dies in mid-operation without emitting SIGTERM
1066 or SIGKILL at its subordinate git fast-import instance.
1067
1068 `option`
1069 ~~~~~~~~
1070 Processes the specified option so that git fast-import behaves in a
1071 way that suits the frontend's needs.
1072 Note that options specified by the frontend are overridden by any
1073 options the user may specify to git fast-import itself.
1074
1075 ....
1076 'option' SP <option> LF
1077 ....
1078
1079 The `<option>` part of the command may contain any of the options
1080 listed in the OPTIONS section that do not change import semantics,
1081 without the leading '--' and is treated in the same way.
1082
1083 Option commands must be the first commands on the input (not counting
1084 feature commands), to give an option command after any non-option
1085 command is an error.
1086
1087 The following command-line options change import semantics and may therefore
1088 not be passed as option:
1089
1090 * date-format
1091 * import-marks
1092 * export-marks
1093 * cat-blob-fd
1094 * force
1095
1096 `done`
1097 ~~~~~~
1098 If the `done` feature is not in use, treated as if EOF was read.
1099 This can be used to tell fast-import to finish early.
1100
1101 If the `--done` command-line option or `feature done` command is
1102 in use, the `done` command is mandatory and marks the end of the
1103 stream.
1104
1105 Responses To Commands
1106 ---------------------
1107 New objects written by fast-import are not available immediately.
1108 Most fast-import commands have no visible effect until the next
1109 checkpoint (or completion). The frontend can send commands to
1110 fill fast-import's input pipe without worrying about how quickly
1111 they will take effect, which improves performance by simplifying
1112 scheduling.
1113
1114 For some frontends, though, it is useful to be able to read back
1115 data from the current repository as it is being updated (for
1116 example when the source material describes objects in terms of
1117 patches to be applied to previously imported objects). This can
1118 be accomplished by connecting the frontend and fast-import via
1119 bidirectional pipes:
1120
1121 ====
1122 mkfifo fast-import-output
1123 frontend <fast-import-output |
1124 git fast-import >fast-import-output
1125 ====
1126
1127 A frontend set up this way can use `progress`, `ls`, and `cat-blob`
1128 commands to read information from the import in progress.
1129
1130 To avoid deadlock, such frontends must completely consume any
1131 pending output from `progress`, `ls`, and `cat-blob` before
1132 performing writes to fast-import that might block.
1133
1134 Crash Reports
1135 -------------
1136 If fast-import is supplied invalid input it will terminate with a
1137 non-zero exit status and create a crash report in the top level of
1138 the Git repository it was importing into. Crash reports contain
1139 a snapshot of the internal fast-import state as well as the most
1140 recent commands that lead up to the crash.
1141
1142 All recent commands (including stream comments, file changes and
1143 progress commands) are shown in the command history within the crash
1144 report, but raw file data and commit messages are excluded from the
1145 crash report. This exclusion saves space within the report file
1146 and reduces the amount of buffering that fast-import must perform
1147 during execution.
1148
1149 After writing a crash report fast-import will close the current
1150 packfile and export the marks table. This allows the frontend
1151 developer to inspect the repository state and resume the import from
1152 the point where it crashed. The modified branches and tags are not
1153 updated during a crash, as the import did not complete successfully.
1154 Branch and tag information can be found in the crash report and
1155 must be applied manually if the update is needed.
1156
1157 An example crash:
1158
1159 ====
1160 $ cat >in <<END_OF_INPUT
1161 # my very first test commit
1162 commit refs/heads/master
1163 committer Shawn O. Pearce <spearce> 19283 -0400
1164 # who is that guy anyway?
1165 data <<EOF
1166 this is my commit
1167 EOF
1168 M 644 inline .gitignore
1169 data <<EOF
1170 .gitignore
1171 EOF
1172 M 777 inline bob
1173 END_OF_INPUT
1174
1175 $ git fast-import <in
1176 fatal: Corrupt mode: M 777 inline bob
1177 fast-import: dumping crash report to .git/fast_import_crash_8434
1178
1179 $ cat .git/fast_import_crash_8434
1180 fast-import crash report:
1181 fast-import process: 8434
1182 parent process : 1391
1183 at Sat Sep 1 00:58:12 2007
1184
1185 fatal: Corrupt mode: M 777 inline bob
1186
1187 Most Recent Commands Before Crash
1188 ---------------------------------
1189 # my very first test commit
1190 commit refs/heads/master
1191 committer Shawn O. Pearce <spearce> 19283 -0400
1192 # who is that guy anyway?
1193 data <<EOF
1194 M 644 inline .gitignore
1195 data <<EOF
1196 * M 777 inline bob
1197
1198 Active Branch LRU
1199 -----------------
1200 active_branches = 1 cur, 5 max
1201
1202 pos clock name
1203 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1204 1) 0 refs/heads/master
1205
1206 Inactive Branches
1207 -----------------
1208 refs/heads/master:
1209 status : active loaded dirty
1210 tip commit : 0000000000000000000000000000000000000000
1211 old tree : 0000000000000000000000000000000000000000
1212 cur tree : 0000000000000000000000000000000000000000
1213 commit clock: 0
1214 last pack :
1215
1216
1217 -------------------
1218 END OF CRASH REPORT
1219 ====
1220
1221 Tips and Tricks
1222 ---------------
1223 The following tips and tricks have been collected from various
1224 users of fast-import, and are offered here as suggestions.
1225
1226 Use One Mark Per Commit
1227 ~~~~~~~~~~~~~~~~~~~~~~~
1228 When doing a repository conversion, use a unique mark per commit
1229 (`mark :<n>`) and supply the --export-marks option on the command
1230 line. fast-import will dump a file which lists every mark and the Git
1231 object SHA-1 that corresponds to it. If the frontend can tie
1232 the marks back to the source repository, it is easy to verify the
1233 accuracy and completeness of the import by comparing each Git
1234 commit to the corresponding source revision.
1235
1236 Coming from a system such as Perforce or Subversion this should be
1237 quite simple, as the fast-import mark can also be the Perforce changeset
1238 number or the Subversion revision number.
1239
1240 Freely Skip Around Branches
1241 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
1242 Don't bother trying to optimize the frontend to stick to one branch
1243 at a time during an import. Although doing so might be slightly
1244 faster for fast-import, it tends to increase the complexity of the frontend
1245 code considerably.
1246
1247 The branch LRU builtin to fast-import tends to behave very well, and the
1248 cost of activating an inactive branch is so low that bouncing around
1249 between branches has virtually no impact on import performance.
1250
1251 Handling Renames
1252 ~~~~~~~~~~~~~~~~
1253 When importing a renamed file or directory, simply delete the old
1254 name(s) and modify the new name(s) during the corresponding commit.
1255 Git performs rename detection after-the-fact, rather than explicitly
1256 during a commit.
1257
1258 Use Tag Fixup Branches
1259 ~~~~~~~~~~~~~~~~~~~~~~
1260 Some other SCM systems let the user create a tag from multiple
1261 files which are not from the same commit/changeset. Or to create
1262 tags which are a subset of the files available in the repository.
1263
1264 Importing these tags as-is in Git is impossible without making at
1265 least one commit which ``fixes up'' the files to match the content
1266 of the tag. Use fast-import's `reset` command to reset a dummy branch
1267 outside of your normal branch space to the base commit for the tag,
1268 then commit one or more file fixup commits, and finally tag the
1269 dummy branch.
1270
1271 For example since all normal branches are stored under `refs/heads/`
1272 name the tag fixup branch `TAG_FIXUP`. This way it is impossible for
1273 the fixup branch used by the importer to have namespace conflicts
1274 with real branches imported from the source (the name `TAG_FIXUP`
1275 is not `refs/heads/TAG_FIXUP`).
1276
1277 When committing fixups, consider using `merge` to connect the
1278 commit(s) which are supplying file revisions to the fixup branch.
1279 Doing so will allow tools such as 'git blame' to track
1280 through the real commit history and properly annotate the source
1281 files.
1282
1283 After fast-import terminates the frontend will need to do `rm .git/TAG_FIXUP`
1284 to remove the dummy branch.
1285
1286 Import Now, Repack Later
1287 ~~~~~~~~~~~~~~~~~~~~~~~~
1288 As soon as fast-import completes the Git repository is completely valid
1289 and ready for use. Typically this takes only a very short time,
1290 even for considerably large projects (100,000+ commits).
1291
1292 However repacking the repository is necessary to improve data
1293 locality and access performance. It can also take hours on extremely
1294 large projects (especially if -f and a large --window parameter is
1295 used). Since repacking is safe to run alongside readers and writers,
1296 run the repack in the background and let it finish when it finishes.
1297 There is no reason to wait to explore your new Git project!
1298
1299 If you choose to wait for the repack, don't try to run benchmarks
1300 or performance tests until repacking is completed. fast-import outputs
1301 suboptimal packfiles that are simply never seen in real use
1302 situations.
1303
1304 Repacking Historical Data
1305 ~~~~~~~~~~~~~~~~~~~~~~~~~
1306 If you are repacking very old imported data (e.g. older than the
1307 last year), consider expending some extra CPU time and supplying
1308 --window=50 (or higher) when you run 'git repack'.
1309 This will take longer, but will also produce a smaller packfile.
1310 You only need to expend the effort once, and everyone using your
1311 project will benefit from the smaller repository.
1312
1313 Include Some Progress Messages
1314 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1315 Every once in a while have your frontend emit a `progress` message
1316 to fast-import. The contents of the messages are entirely free-form,
1317 so one suggestion would be to output the current month and year
1318 each time the current commit date moves into the next month.
1319 Your users will feel better knowing how much of the data stream
1320 has been processed.
1321
1322
1323 Packfile Optimization
1324 ---------------------
1325 When packing a blob fast-import always attempts to deltify against the last
1326 blob written. Unless specifically arranged for by the frontend,
1327 this will probably not be a prior version of the same file, so the
1328 generated delta will not be the smallest possible. The resulting
1329 packfile will be compressed, but will not be optimal.
1330
1331 Frontends which have efficient access to all revisions of a
1332 single file (for example reading an RCS/CVS ,v file) can choose
1333 to supply all revisions of that file as a sequence of consecutive
1334 `blob` commands. This allows fast-import to deltify the different file
1335 revisions against each other, saving space in the final packfile.
1336 Marks can be used to later identify individual file revisions during
1337 a sequence of `commit` commands.
1338
1339 The packfile(s) created by fast-import do not encourage good disk access
1340 patterns. This is caused by fast-import writing the data in the order
1341 it is received on standard input, while Git typically organizes
1342 data within packfiles to make the most recent (current tip) data
1343 appear before historical data. Git also clusters commits together,
1344 speeding up revision traversal through better cache locality.
1345
1346 For this reason it is strongly recommended that users repack the
1347 repository with `git repack -a -d` after fast-import completes, allowing
1348 Git to reorganize the packfiles for faster data access. If blob
1349 deltas are suboptimal (see above) then also adding the `-f` option
1350 to force recomputation of all deltas can significantly reduce the
1351 final packfile size (30-50% smaller can be quite typical).
1352
1353
1354 Memory Utilization
1355 ------------------
1356 There are a number of factors which affect how much memory fast-import
1357 requires to perform an import. Like critical sections of core
1358 Git, fast-import uses its own memory allocators to amortize any overheads
1359 associated with malloc. In practice fast-import tends to amortize any
1360 malloc overheads to 0, due to its use of large block allocations.
1361
1362 per object
1363 ~~~~~~~~~~
1364 fast-import maintains an in-memory structure for every object written in
1365 this execution. On a 32 bit system the structure is 32 bytes,
1366 on a 64 bit system the structure is 40 bytes (due to the larger
1367 pointer sizes). Objects in the table are not deallocated until
1368 fast-import terminates. Importing 2 million objects on a 32 bit system
1369 will require approximately 64 MiB of memory.
1370
1371 The object table is actually a hashtable keyed on the object name
1372 (the unique SHA-1). This storage configuration allows fast-import to reuse
1373 an existing or already written object and avoid writing duplicates
1374 to the output packfile. Duplicate blobs are surprisingly common
1375 in an import, typically due to branch merges in the source.
1376
1377 per mark
1378 ~~~~~~~~
1379 Marks are stored in a sparse array, using 1 pointer (4 bytes or 8
1380 bytes, depending on pointer size) per mark. Although the array
1381 is sparse, frontends are still strongly encouraged to use marks
1382 between 1 and n, where n is the total number of marks required for
1383 this import.
1384
1385 per branch
1386 ~~~~~~~~~~
1387 Branches are classified as active and inactive. The memory usage
1388 of the two classes is significantly different.
1389
1390 Inactive branches are stored in a structure which uses 96 or 120
1391 bytes (32 bit or 64 bit systems, respectively), plus the length of
1392 the branch name (typically under 200 bytes), per branch. fast-import will
1393 easily handle as many as 10,000 inactive branches in under 2 MiB
1394 of memory.
1395
1396 Active branches have the same overhead as inactive branches, but
1397 also contain copies of every tree that has been recently modified on
1398 that branch. If subtree `include` has not been modified since the
1399 branch became active, its contents will not be loaded into memory,
1400 but if subtree `src` has been modified by a commit since the branch
1401 became active, then its contents will be loaded in memory.
1402
1403 As active branches store metadata about the files contained on that
1404 branch, their in-memory storage size can grow to a considerable size
1405 (see below).
1406
1407 fast-import automatically moves active branches to inactive status based on
1408 a simple least-recently-used algorithm. The LRU chain is updated on
1409 each `commit` command. The maximum number of active branches can be
1410 increased or decreased on the command line with --active-branches=.
1411
1412 per active tree
1413 ~~~~~~~~~~~~~~~
1414 Trees (aka directories) use just 12 bytes of memory on top of the
1415 memory required for their entries (see ``per active file'' below).
1416 The cost of a tree is virtually 0, as its overhead amortizes out
1417 over the individual file entries.
1418
1419 per active file entry
1420 ~~~~~~~~~~~~~~~~~~~~~
1421 Files (and pointers to subtrees) within active trees require 52 or 64
1422 bytes (32/64 bit platforms) per entry. To conserve space, file and
1423 tree names are pooled in a common string table, allowing the filename
1424 ``Makefile'' to use just 16 bytes (after including the string header
1425 overhead) no matter how many times it occurs within the project.
1426
1427 The active branch LRU, when coupled with the filename string pool
1428 and lazy loading of subtrees, allows fast-import to efficiently import
1429 projects with 2,000+ branches and 45,114+ files in a very limited
1430 memory footprint (less than 2.7 MiB per active branch).
1431
1432 Signals
1433 -------
1434 Sending *SIGUSR1* to the 'git fast-import' process ends the current
1435 packfile early, simulating a `checkpoint` command. The impatient
1436 operator can use this facility to peek at the objects and refs from an
1437 import in progress, at the cost of some added running time and worse
1438 compression.
1439
1440 SEE ALSO
1441 --------
1442 linkgit:git-fast-export[1]
1443
1444 GIT
1445 ---
1446 Part of the linkgit:git[1] suite