]>
Commit | Line | Data |
---|---|---|
6e411d20 SP |
1 | git-fast-import(1) |
2 | ================== | |
3 | ||
4 | NAME | |
5 | ---- | |
7a33631f | 6 | git-fast-import - Backend for fast Git data importers |
6e411d20 SP |
7 | |
8 | ||
9 | SYNOPSIS | |
10 | -------- | |
b1889c36 | 11 | frontend | 'git fast-import' [options] |
6e411d20 SP |
12 | |
13 | DESCRIPTION | |
14 | ----------- | |
15 | This program is usually not what the end user wants to run directly. | |
16 | Most end users want to use one of the existing frontend programs, | |
17 | which parses a specific type of foreign source and feeds the contents | |
0b444cdb | 18 | stored there to 'git fast-import'. |
6e411d20 | 19 | |
882227f1 | 20 | fast-import reads a mixed command/data stream from standard input and |
6e411d20 SP |
21 | writes one or more packfiles directly into the current repository. |
22 | When EOF is received on standard input, fast import writes out | |
23 | updated branch and tag refs, fully updating the current repository | |
24 | with the newly imported data. | |
25 | ||
882227f1 | 26 | The fast-import backend itself can import into an empty repository (one that |
0b444cdb | 27 | has already been initialized by 'git init') or incrementally |
6e411d20 SP |
28 | update an existing populated repository. Whether or not incremental |
29 | imports are supported from a particular foreign source depends on | |
30 | the frontend program in use. | |
31 | ||
32 | ||
33 | OPTIONS | |
34 | ------- | |
63e0c8b3 SP |
35 | --date-format=<fmt>:: |
36 | Specify the type of dates the frontend will supply to | |
882227f1 | 37 | fast-import within `author`, `committer` and `tagger` commands. |
63e0c8b3 SP |
38 | See ``Date Formats'' below for details about which formats |
39 | are supported, and their syntax. | |
40 | ||
7073e69e SP |
41 | --force:: |
42 | Force updating modified existing branches, even if doing | |
43 | so would cause commits to be lost (as the new commit does | |
44 | not contain the old commit). | |
45 | ||
6e411d20 | 46 | --max-pack-size=<n>:: |
4d0cc224 JH |
47 | Maximum size of each output packfile. |
48 | The default is 4 GiB as that is the maximum allowed | |
6e411d20 SP |
49 | packfile size (due to file format limitations). Some |
50 | importers may wish to lower this, such as to ensure the | |
51 | resulting packfiles fit on CDs. | |
52 | ||
5eef828b SP |
53 | --big-file-threshold=<n>:: |
54 | Maximum size of a blob that fast-import will attempt to | |
55 | create a delta for, expressed in bytes. The default is 512m | |
56 | (512 MiB). Some importers may wish to lower this on systems | |
57 | with constrained memory. | |
58 | ||
6e411d20 SP |
59 | --depth=<n>:: |
60 | Maximum delta depth, for blob and tree deltification. | |
61 | Default is 10. | |
62 | ||
63 | --active-branches=<n>:: | |
64 | Maximum number of branches to maintain active at once. | |
65 | See ``Memory Utilization'' below for details. Default is 5. | |
66 | ||
67 | --export-marks=<file>:: | |
68 | Dumps the internal marks table to <file> when complete. | |
69 | Marks are written one per line as `:markid SHA-1`. | |
70 | Frontends can use this file to validate imports after they | |
e8438420 SP |
71 | have been completed, or to save the marks table across |
72 | incremental runs. As <file> is only opened and truncated | |
73 | at checkpoint (or completion) the same path can also be | |
74 | safely given to \--import-marks. | |
75 | ||
76 | --import-marks=<file>:: | |
77 | Before processing any input, load the marks specified in | |
78 | <file>. The input file must exist, must be readable, and | |
79 | must use the same format as produced by \--export-marks. | |
80 | Multiple options may be supplied to import more than one | |
81 | set of marks. If a mark is defined to different values, | |
82 | the last file wins. | |
6e411d20 | 83 | |
bc3c79ae SR |
84 | --relative-marks:: |
85 | After specifying --relative-marks= the paths specified | |
86 | with --import-marks= and --export-marks= are relative | |
87 | to an internal directory in the current repository. | |
88 | In git-fast-import this means that the paths are relative | |
89 | to the .git/info/fast-import directory. However, other | |
90 | importers may use a different location. | |
91 | ||
92 | --no-relative-marks:: | |
93 | Negates a previous --relative-marks. Allows for combining | |
94 | relative and non-relative marks by interweaving | |
95 | --(no-)-relative-marks= with the --(import|export)-marks= | |
96 | options. | |
97 | ||
bdf1c06d SP |
98 | --export-pack-edges=<file>:: |
99 | After creating a packfile, print a line of data to | |
100 | <file> listing the filename of the packfile and the last | |
101 | commit on each branch that was written to that packfile. | |
102 | This information may be useful after importing projects | |
103 | whose total object set exceeds the 4 GiB packfile limit, | |
104 | as these commits can be used as edge points during calls | |
0b444cdb | 105 | to 'git pack-objects'. |
bdf1c06d | 106 | |
c499d768 | 107 | --quiet:: |
882227f1 | 108 | Disable all non-fatal output, making fast-import silent when it |
7f9d77f2 | 109 | is successful. This option disables the output shown by |
c499d768 SP |
110 | \--stats. |
111 | ||
112 | --stats:: | |
882227f1 | 113 | Display some basic statistics about the objects fast-import has |
c499d768 | 114 | created, the packfiles they were stored into, and the |
882227f1 | 115 | memory used by fast-import during this run. Showing this output |
c499d768 SP |
116 | is currently the default, but can be disabled with \--quiet. |
117 | ||
118 | ||
6e411d20 SP |
119 | Performance |
120 | ----------- | |
882227f1 | 121 | The design of fast-import allows it to import large projects in a minimum |
6e411d20 | 122 | amount of memory usage and processing time. Assuming the frontend |
882227f1 | 123 | is able to keep up with fast-import and feed it a constant stream of data, |
6e411d20 SP |
124 | import times for projects holding 10+ years of history and containing |
125 | 100,000+ individual commits are generally completed in just 1-2 | |
126 | hours on quite modest (~$2,000 USD) hardware. | |
127 | ||
128 | Most bottlenecks appear to be in foreign source data access (the | |
882227f1 | 129 | source just cannot extract revisions fast enough) or disk IO (fast-import |
6e411d20 SP |
130 | writes as fast as the disk will take the data). Imports will run |
131 | faster if the source data is stored on a different drive than the | |
132 | destination Git repository (due to less IO contention). | |
133 | ||
134 | ||
135 | Development Cost | |
136 | ---------------- | |
882227f1 | 137 | A typical frontend for fast-import tends to weigh in at approximately 200 |
6e411d20 SP |
138 | lines of Perl/Python/Ruby code. Most developers have been able to |
139 | create working importers in just a couple of hours, even though it | |
882227f1 | 140 | is their first exposure to fast-import, and sometimes even to Git. This is |
6e411d20 SP |
141 | an ideal situation, given that most conversion tools are throw-away |
142 | (use once, and never look back). | |
143 | ||
144 | ||
145 | Parallel Operation | |
146 | ------------------ | |
0b444cdb | 147 | Like 'git push' or 'git fetch', imports handled by fast-import are safe to |
6e411d20 | 148 | run alongside parallel `git repack -a -d` or `git gc` invocations, |
0b444cdb | 149 | or any other Git operation (including 'git prune', as loose objects |
882227f1 | 150 | are never used by fast-import). |
6e411d20 | 151 | |
882227f1 SP |
152 | fast-import does not lock the branch or tag refs it is actively importing. |
153 | After the import, during its ref update phase, fast-import tests each | |
7073e69e SP |
154 | existing branch ref to verify the update will be a fast-forward |
155 | update (the commit stored in the ref is contained in the new | |
156 | history of the commit to be written). If the update is not a | |
882227f1 SP |
157 | fast-forward update, fast-import will skip updating that ref and instead |
158 | prints a warning message. fast-import will always attempt to update all | |
7073e69e SP |
159 | branch refs, and does not stop on the first failure. |
160 | ||
6a5d0b0a | 161 | Branch updates can be forced with \--force, but it's recommended that |
c499d768 | 162 | this only be used on an otherwise quiet repository. Using \--force |
7073e69e | 163 | is not necessary for an initial import into an empty repository. |
6e411d20 SP |
164 | |
165 | ||
166 | Technical Discussion | |
167 | -------------------- | |
882227f1 | 168 | fast-import tracks a set of branches in memory. Any branch can be created |
6e411d20 SP |
169 | or modified at any point during the import process by sending a |
170 | `commit` command on the input stream. This design allows a frontend | |
171 | program to process an unlimited number of branches simultaneously, | |
172 | generating commits in the order they are available from the source | |
173 | data. It also simplifies the frontend programs considerably. | |
174 | ||
882227f1 | 175 | fast-import does not use or alter the current working directory, or any |
6e411d20 SP |
176 | file within it. (It does however update the current Git repository, |
177 | as referenced by `GIT_DIR`.) Therefore an import frontend may use | |
178 | the working directory for its own purposes, such as extracting file | |
179 | revisions from the foreign source. This ignorance of the working | |
882227f1 | 180 | directory also allows fast-import to run very quickly, as it does not |
6e411d20 SP |
181 | need to perform any costly file update operations when switching |
182 | between branches. | |
183 | ||
184 | Input Format | |
185 | ------------ | |
186 | With the exception of raw file data (which Git does not interpret) | |
882227f1 | 187 | the fast-import input format is text (ASCII) based. This text based |
6e411d20 SP |
188 | format simplifies development and debugging of frontend programs, |
189 | especially when a higher level language such as Perl, Python or | |
190 | Ruby is being used. | |
191 | ||
882227f1 | 192 | fast-import is very strict about its input. Where we say SP below we mean |
6e411d20 SP |
193 | *exactly* one space. Likewise LF means one (and only one) linefeed. |
194 | Supplying additional whitespace characters will cause unexpected | |
195 | results, such as branch names or file names with leading or trailing | |
882227f1 | 196 | spaces in their name, or early termination of fast-import when it encounters |
6e411d20 SP |
197 | unexpected input. |
198 | ||
401d53fa SP |
199 | Stream Comments |
200 | ~~~~~~~~~~~~~~~ | |
201 | To aid in debugging frontends fast-import ignores any line that | |
202 | begins with `#` (ASCII pound/hash) up to and including the line | |
203 | ending `LF`. A comment line may contain any sequence of bytes | |
204 | that does not contain an LF and therefore may be used to include | |
205 | any detailed debugging information that might be specific to the | |
206 | frontend and useful when inspecting a fast-import data stream. | |
207 | ||
63e0c8b3 SP |
208 | Date Formats |
209 | ~~~~~~~~~~~~ | |
210 | The following date formats are supported. A frontend should select | |
211 | the format it will use for this import by passing the format name | |
c499d768 | 212 | in the \--date-format=<fmt> command line option. |
63e0c8b3 SP |
213 | |
214 | `raw`:: | |
9b92c82f | 215 | This is the Git native format and is `<time> SP <offutc>`. |
882227f1 | 216 | It is also fast-import's default format, if \--date-format was |
63e0c8b3 SP |
217 | not specified. |
218 | + | |
219 | The time of the event is specified by `<time>` as the number of | |
220 | seconds since the UNIX epoch (midnight, Jan 1, 1970, UTC) and is | |
221 | written as an ASCII decimal integer. | |
222 | + | |
9b92c82f SP |
223 | The local offset is specified by `<offutc>` as a positive or negative |
224 | offset from UTC. For example EST (which is 5 hours behind UTC) | |
225 | would be expressed in `<tz>` by ``-0500'' while UTC is ``+0000''. | |
226 | The local offset does not affect `<time>`; it is used only as an | |
227 | advisement to help formatting routines display the timestamp. | |
63e0c8b3 | 228 | + |
9b92c82f SP |
229 | If the local offset is not available in the source material, use |
230 | ``+0000'', or the most common local offset. For example many | |
63e0c8b3 SP |
231 | organizations have a CVS repository which has only ever been accessed |
232 | by users who are located in the same location and timezone. In this | |
f842fdb0 | 233 | case a reasonable offset from UTC could be assumed. |
63e0c8b3 SP |
234 | + |
235 | Unlike the `rfc2822` format, this format is very strict. Any | |
882227f1 | 236 | variation in formatting will cause fast-import to reject the value. |
63e0c8b3 SP |
237 | |
238 | `rfc2822`:: | |
239 | This is the standard email format as described by RFC 2822. | |
240 | + | |
241 | An example value is ``Tue Feb 6 11:22:18 2007 -0500''. The Git | |
f842fdb0 | 242 | parser is accurate, but a little on the lenient side. It is the |
0b444cdb | 243 | same parser used by 'git am' when applying patches |
63e0c8b3 SP |
244 | received from email. |
245 | + | |
246 | Some malformed strings may be accepted as valid dates. In some of | |
247 | these cases Git will still be able to obtain the correct date from | |
248 | the malformed string. There are also some types of malformed | |
249 | strings which Git will parse wrong, and yet consider valid. | |
250 | Seriously malformed strings will be rejected. | |
251 | + | |
9b92c82f SP |
252 | Unlike the `raw` format above, the timezone/UTC offset information |
253 | contained in an RFC 2822 date string is used to adjust the date | |
254 | value to UTC prior to storage. Therefore it is important that | |
255 | this information be as accurate as possible. | |
256 | + | |
f842fdb0 | 257 | If the source material uses RFC 2822 style dates, |
882227f1 | 258 | the frontend should let fast-import handle the parsing and conversion |
63e0c8b3 SP |
259 | (rather than attempting to do it itself) as the Git parser has |
260 | been well tested in the wild. | |
261 | + | |
262 | Frontends should prefer the `raw` format if the source material | |
f842fdb0 | 263 | already uses UNIX-epoch format, can be coaxed to give dates in that |
02783075 | 264 | format, or its format is easily convertible to it, as there is no |
f842fdb0 | 265 | ambiguity in parsing. |
63e0c8b3 SP |
266 | |
267 | `now`:: | |
268 | Always use the current time and timezone. The literal | |
269 | `now` must always be supplied for `<when>`. | |
270 | + | |
271 | This is a toy format. The current time and timezone of this system | |
272 | is always copied into the identity string at the time it is being | |
882227f1 | 273 | created by fast-import. There is no way to specify a different time or |
63e0c8b3 SP |
274 | timezone. |
275 | + | |
6a5d0b0a | 276 | This particular format is supplied as it's short to implement and |
63e0c8b3 SP |
277 | may be useful to a process that wants to create a new commit |
278 | right now, without needing to use a working directory or | |
0b444cdb | 279 | 'git update-index'. |
63e0c8b3 SP |
280 | + |
281 | If separate `author` and `committer` commands are used in a `commit` | |
282 | the timestamps may not match, as the system clock will be polled | |
283 | twice (once for each command). The only way to ensure that both | |
284 | author and committer identity information has the same timestamp | |
285 | is to omit `author` (thus copying from `committer`) or to use a | |
286 | date format other than `now`. | |
287 | ||
6e411d20 SP |
288 | Commands |
289 | ~~~~~~~~ | |
882227f1 | 290 | fast-import accepts several commands to update the current repository |
6e411d20 SP |
291 | and control the current import process. More detailed discussion |
292 | (with examples) of each command follows later. | |
293 | ||
294 | `commit`:: | |
295 | Creates a new branch or updates an existing branch by | |
296 | creating a new commit and updating the branch to point at | |
297 | the newly created commit. | |
298 | ||
299 | `tag`:: | |
300 | Creates an annotated tag object from an existing commit or | |
301 | branch. Lightweight tags are not supported by this command, | |
302 | as they are not recommended for recording meaningful points | |
303 | in time. | |
304 | ||
305 | `reset`:: | |
306 | Reset an existing branch (or a new branch) to a specific | |
307 | revision. This command must be used to change a branch to | |
308 | a specific revision without making a commit on it. | |
309 | ||
310 | `blob`:: | |
311 | Convert raw file data into a blob, for future use in a | |
312 | `commit` command. This command is optional and is not | |
313 | needed to perform an import. | |
314 | ||
315 | `checkpoint`:: | |
882227f1 | 316 | Forces fast-import to close the current packfile, generate its |
6e411d20 SP |
317 | unique SHA-1 checksum and index, and start a new packfile. |
318 | This command is optional and is not needed to perform | |
319 | an import. | |
320 | ||
ac053c02 SP |
321 | `progress`:: |
322 | Causes fast-import to echo the entire line to its own | |
323 | standard output. This command is optional and is not needed | |
324 | to perform an import. | |
325 | ||
f963bd5d SR |
326 | `feature`:: |
327 | Require that fast-import supports the specified feature, or | |
328 | abort if it does not. | |
329 | ||
9c8398f0 SR |
330 | `option`:: |
331 | Specify any of the options listed under OPTIONS that do not | |
332 | change stream semantic to suit the frontend's needs. This | |
333 | command is optional and is not needed to perform an import. | |
334 | ||
6e411d20 SP |
335 | `commit` |
336 | ~~~~~~~~ | |
337 | Create or update a branch with a new commit, recording one logical | |
338 | change to the project. | |
339 | ||
340 | .... | |
341 | 'commit' SP <ref> LF | |
342 | mark? | |
74fbd118 SP |
343 | ('author' (SP <name>)? SP LT <email> GT SP <when> LF)? |
344 | 'committer' (SP <name>)? SP LT <email> GT SP <when> LF | |
6e411d20 SP |
345 | data |
346 | ('from' SP <committish> LF)? | |
347 | ('merge' SP <committish> LF)? | |
a8dd2e7d | 348 | (filemodify | filedelete | filecopy | filerename | filedeleteall | notemodify)* |
1fdb649c | 349 | LF? |
6e411d20 SP |
350 | .... |
351 | ||
352 | where `<ref>` is the name of the branch to make the commit on. | |
353 | Typically branch names are prefixed with `refs/heads/` in | |
354 | Git, so importing the CVS branch symbol `RELENG-1_0` would use | |
355 | `refs/heads/RELENG-1_0` for the value of `<ref>`. The value of | |
356 | `<ref>` must be a valid refname in Git. As `LF` is not valid in | |
357 | a Git refname, no quoting or escaping syntax is supported here. | |
358 | ||
882227f1 | 359 | A `mark` command may optionally appear, requesting fast-import to save a |
6e411d20 SP |
360 | reference to the newly created commit for future use by the frontend |
361 | (see below for format). It is very common for frontends to mark | |
362 | every commit they create, thereby allowing future branch creation | |
363 | from any imported commit. | |
364 | ||
365 | The `data` command following `committer` must supply the commit | |
366 | message (see below for `data` command syntax). To import an empty | |
367 | commit message use a 0 length data. Commit messages are free-form | |
368 | and are not interpreted by Git. Currently they must be encoded in | |
882227f1 | 369 | UTF-8, as fast-import does not permit other encodings to be specified. |
6e411d20 | 370 | |
a8dd2e7d JH |
371 | Zero or more `filemodify`, `filedelete`, `filecopy`, `filerename`, |
372 | `filedeleteall` and `notemodify` commands | |
825769a8 SP |
373 | may be included to update the contents of the branch prior to |
374 | creating the commit. These commands may be supplied in any order. | |
02783075 | 375 | However it is recommended that a `filedeleteall` command precede |
a8dd2e7d JH |
376 | all `filemodify`, `filecopy`, `filerename` and `notemodify` commands in |
377 | the same commit, as `filedeleteall` wipes the branch clean (see below). | |
6e411d20 | 378 | |
1fdb649c SP |
379 | The `LF` after the command is optional (it used to be required). |
380 | ||
6e411d20 SP |
381 | `author` |
382 | ^^^^^^^^ | |
383 | An `author` command may optionally appear, if the author information | |
384 | might differ from the committer information. If `author` is omitted | |
882227f1 | 385 | then fast-import will automatically use the committer's information for |
6e411d20 SP |
386 | the author portion of the commit. See below for a description of |
387 | the fields in `author`, as they are identical to `committer`. | |
388 | ||
389 | `committer` | |
390 | ^^^^^^^^^^^ | |
391 | The `committer` command indicates who made this commit, and when | |
392 | they made it. | |
393 | ||
394 | Here `<name>` is the person's display name (for example | |
395 | ``Com M Itter'') and `<email>` is the person's email address | |
396 | (``cm@example.com''). `LT` and `GT` are the literal less-than (\x3c) | |
397 | and greater-than (\x3e) symbols. These are required to delimit | |
398 | the email address from the other fields in the line. Note that | |
399 | `<name>` is free-form and may contain any sequence of bytes, except | |
400 | `LT` and `LF`. It is typically UTF-8 encoded. | |
401 | ||
63e0c8b3 | 402 | The time of the change is specified by `<when>` using the date format |
c499d768 | 403 | that was selected by the \--date-format=<fmt> command line option. |
63e0c8b3 SP |
404 | See ``Date Formats'' above for the set of supported formats, and |
405 | their syntax. | |
6e411d20 SP |
406 | |
407 | `from` | |
408 | ^^^^^^ | |
ea5e370a SP |
409 | The `from` command is used to specify the commit to initialize |
410 | this branch from. This revision will be the first ancestor of the | |
411 | new commit. | |
412 | ||
413 | Omitting the `from` command in the first commit of a new branch | |
414 | will cause fast-import to create that commit with no ancestor. This | |
415 | tends to be desired only for the initial commit of a project. | |
9b33fa08 EB |
416 | If the frontend creates all files from scratch when making a new |
417 | branch, a `merge` command may be used instead of `from` to start | |
418 | the commit with an empty tree. | |
ea5e370a SP |
419 | Omitting the `from` command on existing branches is usually desired, |
420 | as the current commit on that branch is automatically assumed to | |
421 | be the first ancestor of the new commit. | |
6e411d20 SP |
422 | |
423 | As `LF` is not valid in a Git refname or SHA-1 expression, no | |
424 | quoting or escaping syntax is supported within `<committish>`. | |
425 | ||
426 | Here `<committish>` is any of the following: | |
427 | ||
882227f1 | 428 | * The name of an existing branch already in fast-import's internal branch |
6a5d0b0a | 429 | table. If fast-import doesn't know the name, it's treated as a SHA-1 |
6e411d20 SP |
430 | expression. |
431 | ||
432 | * A mark reference, `:<idnum>`, where `<idnum>` is the mark number. | |
433 | + | |
882227f1 | 434 | The reason fast-import uses `:` to denote a mark reference is this character |
6e411d20 | 435 | is not legal in a Git branch name. The leading `:` makes it easy |
02783075 | 436 | to distinguish between the mark 42 (`:42`) and the branch 42 (`42` |
6e411d20 SP |
437 | or `refs/heads/42`), or an abbreviated SHA-1 which happened to |
438 | consist only of base-10 digits. | |
439 | + | |
440 | Marks must be declared (via `mark`) before they can be used. | |
441 | ||
442 | * A complete 40 byte or abbreviated commit SHA-1 in hex. | |
443 | ||
444 | * Any valid Git SHA-1 expression that resolves to a commit. See | |
5162e697 | 445 | ``SPECIFYING REVISIONS'' in linkgit:git-rev-parse[1] for details. |
6e411d20 SP |
446 | |
447 | The special case of restarting an incremental import from the | |
448 | current branch value should be written as: | |
449 | ---- | |
450 | from refs/heads/branch^0 | |
451 | ---- | |
882227f1 | 452 | The `{caret}0` suffix is necessary as fast-import does not permit a branch to |
6e411d20 | 453 | start from itself, and the branch is created in memory before the |
209f1298 | 454 | `from` command is even read from the input. Adding `{caret}0` will force |
882227f1 | 455 | fast-import to resolve the commit through Git's revision parsing library, |
6e411d20 SP |
456 | rather than its internal branch table, thereby loading in the |
457 | existing value of the branch. | |
458 | ||
459 | `merge` | |
460 | ^^^^^^^ | |
9b33fa08 EB |
461 | Includes one additional ancestor commit. If the `from` command is |
462 | omitted when creating a new branch, the first `merge` commit will be | |
463 | the first ancestor of the current commit, and the branch will start | |
464 | out with no files. An unlimited number of `merge` commands per | |
882227f1 | 465 | commit are permitted by fast-import, thereby establishing an n-way merge. |
6e411d20 SP |
466 | However Git's other tools never create commits with more than 15 |
467 | additional ancestors (forming a 16-way merge). For this reason | |
468 | it is suggested that frontends do not use more than 15 `merge` | |
9b33fa08 | 469 | commands per commit; 16, if starting a new, empty branch. |
6e411d20 SP |
470 | |
471 | Here `<committish>` is any of the commit specification expressions | |
472 | also accepted by `from` (see above). | |
473 | ||
474 | `filemodify` | |
ef94edb5 | 475 | ^^^^^^^^^^^^ |
6e411d20 SP |
476 | Included in a `commit` command to add a new file or change the |
477 | content of an existing file. This command has two different means | |
478 | of specifying the content of the file. | |
479 | ||
480 | External data format:: | |
481 | The data content for the file was already supplied by a prior | |
482 | `blob` command. The frontend just needs to connect it. | |
483 | + | |
484 | .... | |
485 | 'M' SP <mode> SP <dataref> SP <path> LF | |
486 | .... | |
487 | + | |
488 | Here `<dataref>` can be either a mark reference (`:<idnum>`) | |
489 | set by a prior `blob` command, or a full 40-byte SHA-1 of an | |
490 | existing Git blob object. | |
491 | ||
492 | Inline data format:: | |
493 | The data content for the file has not been supplied yet. | |
494 | The frontend wants to supply it as part of this modify | |
495 | command. | |
496 | + | |
497 | .... | |
498 | 'M' SP <mode> SP 'inline' SP <path> LF | |
499 | data | |
500 | .... | |
501 | + | |
502 | See below for a detailed description of the `data` command. | |
503 | ||
504 | In both formats `<mode>` is the type of file entry, specified | |
505 | in octal. Git only supports the following modes: | |
506 | ||
507 | * `100644` or `644`: A normal (not-executable) file. The majority | |
508 | of files in most projects use this mode. If in doubt, this is | |
509 | what you want. | |
510 | * `100755` or `755`: A normal, but executable, file. | |
9981b6d9 | 511 | * `120000`: A symlink, the content of the file will be the link target. |
03db4525 AG |
512 | * `160000`: A gitlink, SHA-1 of the object refers to a commit in |
513 | another repository. Git links can only be specified by SHA or through | |
514 | a commit mark. They are used to implement submodules. | |
6e411d20 SP |
515 | |
516 | In both formats `<path>` is the complete path of the file to be added | |
517 | (if not already existing) or modified (if already existing). | |
518 | ||
c4431d38 | 519 | A `<path>` string must use UNIX-style directory separators (forward |
6e411d20 SP |
520 | slash `/`), may contain any byte other than `LF`, and must not |
521 | start with double quote (`"`). | |
522 | ||
523 | If an `LF` or double quote must be encoded into `<path>` shell-style | |
524 | quoting should be used, e.g. `"path/with\n and \" in it"`. | |
525 | ||
02783075 | 526 | The value of `<path>` must be in canonical form. That is it must not: |
6e411d20 SP |
527 | |
528 | * contain an empty directory component (e.g. `foo//bar` is invalid), | |
c4431d38 JK |
529 | * end with a directory separator (e.g. `foo/` is invalid), |
530 | * start with a directory separator (e.g. `/foo` is invalid), | |
6e411d20 SP |
531 | * contain the special component `.` or `..` (e.g. `foo/./bar` and |
532 | `foo/../bar` are invalid). | |
533 | ||
534 | It is recommended that `<path>` always be encoded using UTF-8. | |
535 | ||
6e411d20 | 536 | `filedelete` |
ef94edb5 | 537 | ^^^^^^^^^^^^ |
512e44b2 SP |
538 | Included in a `commit` command to remove a file or recursively |
539 | delete an entire directory from the branch. If the file or directory | |
540 | removal makes its parent directory empty, the parent directory will | |
6e411d20 SP |
541 | be automatically removed too. This cascades up the tree until the |
542 | first non-empty directory or the root is reached. | |
543 | ||
544 | .... | |
545 | 'D' SP <path> LF | |
546 | .... | |
547 | ||
512e44b2 SP |
548 | here `<path>` is the complete path of the file or subdirectory to |
549 | be removed from the branch. | |
6e411d20 SP |
550 | See `filemodify` above for a detailed description of `<path>`. |
551 | ||
b6f3481b SP |
552 | `filecopy` |
553 | ^^^^^^^^^^^^ | |
554 | Recursively copies an existing file or subdirectory to a different | |
555 | location within the branch. The existing file or directory must | |
556 | exist. If the destination exists it will be completely replaced | |
557 | by the content copied from the source. | |
558 | ||
559 | .... | |
560 | 'C' SP <path> SP <path> LF | |
561 | .... | |
562 | ||
563 | here the first `<path>` is the source location and the second | |
564 | `<path>` is the destination. See `filemodify` above for a detailed | |
565 | description of what `<path>` may look like. To use a source path | |
566 | that contains SP the path must be quoted. | |
567 | ||
568 | A `filecopy` command takes effect immediately. Once the source | |
569 | location has been copied to the destination any future commands | |
570 | applied to the source location will not impact the destination of | |
571 | the copy. | |
572 | ||
f39a946a SP |
573 | `filerename` |
574 | ^^^^^^^^^^^^ | |
575 | Renames an existing file or subdirectory to a different location | |
576 | within the branch. The existing file or directory must exist. If | |
577 | the destination exists it will be replaced by the source directory. | |
578 | ||
579 | .... | |
580 | 'R' SP <path> SP <path> LF | |
581 | .... | |
582 | ||
583 | here the first `<path>` is the source location and the second | |
584 | `<path>` is the destination. See `filemodify` above for a detailed | |
585 | description of what `<path>` may look like. To use a source path | |
586 | that contains SP the path must be quoted. | |
587 | ||
588 | A `filerename` command takes effect immediately. Once the source | |
589 | location has been renamed to the destination any future commands | |
590 | applied to the source location will create new files there and not | |
591 | impact the destination of the rename. | |
592 | ||
b6f3481b SP |
593 | Note that a `filerename` is the same as a `filecopy` followed by a |
594 | `filedelete` of the source location. There is a slight performance | |
595 | advantage to using `filerename`, but the advantage is so small | |
596 | that it is never worth trying to convert a delete/add pair in | |
597 | source material into a rename for fast-import. This `filerename` | |
598 | command is provided just to simplify frontends that already have | |
599 | rename information and don't want bother with decomposing it into a | |
600 | `filecopy` followed by a `filedelete`. | |
601 | ||
825769a8 SP |
602 | `filedeleteall` |
603 | ^^^^^^^^^^^^^^^ | |
604 | Included in a `commit` command to remove all files (and also all | |
605 | directories) from the branch. This command resets the internal | |
606 | branch structure to have no files in it, allowing the frontend | |
607 | to subsequently add all interesting files from scratch. | |
608 | ||
609 | .... | |
610 | 'deleteall' LF | |
611 | .... | |
612 | ||
613 | This command is extremely useful if the frontend does not know | |
614 | (or does not care to know) what files are currently on the branch, | |
615 | and therefore cannot generate the proper `filedelete` commands to | |
616 | update the content. | |
617 | ||
618 | Issuing a `filedeleteall` followed by the needed `filemodify` | |
619 | commands to set the correct content will produce the same results | |
620 | as sending only the needed `filemodify` and `filedelete` commands. | |
882227f1 | 621 | The `filedeleteall` approach may however require fast-import to use slightly |
825769a8 SP |
622 | more memory per active branch (less than 1 MiB for even most large |
623 | projects); so frontends that can easily obtain only the affected | |
624 | paths for a commit are encouraged to do so. | |
625 | ||
a8dd2e7d JH |
626 | `notemodify` |
627 | ^^^^^^^^^^^^ | |
628 | Included in a `commit` command to add a new note (annotating a given | |
629 | commit) or change the content of an existing note. This command has | |
630 | two different means of specifying the content of the note. | |
631 | ||
632 | External data format:: | |
633 | The data content for the note was already supplied by a prior | |
634 | `blob` command. The frontend just needs to connect it to the | |
635 | commit that is to be annotated. | |
636 | + | |
637 | .... | |
638 | 'N' SP <dataref> SP <committish> LF | |
639 | .... | |
640 | + | |
641 | Here `<dataref>` can be either a mark reference (`:<idnum>`) | |
642 | set by a prior `blob` command, or a full 40-byte SHA-1 of an | |
643 | existing Git blob object. | |
644 | ||
645 | Inline data format:: | |
646 | The data content for the note has not been supplied yet. | |
647 | The frontend wants to supply it as part of this modify | |
648 | command. | |
649 | + | |
650 | .... | |
651 | 'N' SP 'inline' SP <committish> LF | |
652 | data | |
653 | .... | |
654 | + | |
655 | See below for a detailed description of the `data` command. | |
656 | ||
657 | In both formats `<committish>` is any of the commit specification | |
658 | expressions also accepted by `from` (see above). | |
659 | ||
6e411d20 SP |
660 | `mark` |
661 | ~~~~~~ | |
882227f1 | 662 | Arranges for fast-import to save a reference to the current object, allowing |
6e411d20 SP |
663 | the frontend to recall this object at a future point in time, without |
664 | knowing its SHA-1. Here the current object is the object creation | |
665 | command the `mark` command appears within. This can be `commit`, | |
666 | `tag`, and `blob`, but `commit` is the most common usage. | |
667 | ||
668 | .... | |
669 | 'mark' SP ':' <idnum> LF | |
670 | .... | |
671 | ||
672 | where `<idnum>` is the number assigned by the frontend to this mark. | |
ef94edb5 SP |
673 | The value of `<idnum>` is expressed as an ASCII decimal integer. |
674 | The value 0 is reserved and cannot be used as | |
6e411d20 SP |
675 | a mark. Only values greater than or equal to 1 may be used as marks. |
676 | ||
677 | New marks are created automatically. Existing marks can be moved | |
678 | to another object simply by reusing the same `<idnum>` in another | |
679 | `mark` command. | |
680 | ||
681 | `tag` | |
682 | ~~~~~ | |
683 | Creates an annotated tag referring to a specific commit. To create | |
684 | lightweight (non-annotated) tags see the `reset` command below. | |
685 | ||
686 | .... | |
687 | 'tag' SP <name> LF | |
688 | 'from' SP <committish> LF | |
74fbd118 | 689 | 'tagger' (SP <name>)? SP LT <email> GT SP <when> LF |
6e411d20 | 690 | data |
6e411d20 SP |
691 | .... |
692 | ||
693 | where `<name>` is the name of the tag to create. | |
694 | ||
695 | Tag names are automatically prefixed with `refs/tags/` when stored | |
696 | in Git, so importing the CVS branch symbol `RELENG-1_0-FINAL` would | |
882227f1 | 697 | use just `RELENG-1_0-FINAL` for `<name>`, and fast-import will write the |
6e411d20 SP |
698 | corresponding ref as `refs/tags/RELENG-1_0-FINAL`. |
699 | ||
700 | The value of `<name>` must be a valid refname in Git and therefore | |
701 | may contain forward slashes. As `LF` is not valid in a Git refname, | |
702 | no quoting or escaping syntax is supported here. | |
703 | ||
704 | The `from` command is the same as in the `commit` command; see | |
705 | above for details. | |
706 | ||
707 | The `tagger` command uses the same format as `committer` within | |
708 | `commit`; again see above for details. | |
709 | ||
710 | The `data` command following `tagger` must supply the annotated tag | |
711 | message (see below for `data` command syntax). To import an empty | |
712 | tag message use a 0 length data. Tag messages are free-form and are | |
713 | not interpreted by Git. Currently they must be encoded in UTF-8, | |
882227f1 | 714 | as fast-import does not permit other encodings to be specified. |
6e411d20 | 715 | |
882227f1 | 716 | Signing annotated tags during import from within fast-import is not |
6e411d20 SP |
717 | supported. Trying to include your own PGP/GPG signature is not |
718 | recommended, as the frontend does not (easily) have access to the | |
719 | complete set of bytes which normally goes into such a signature. | |
882227f1 | 720 | If signing is required, create lightweight tags from within fast-import with |
6e411d20 | 721 | `reset`, then create the annotated versions of those tags offline |
0b444cdb | 722 | with the standard 'git tag' process. |
6e411d20 SP |
723 | |
724 | `reset` | |
725 | ~~~~~~~ | |
726 | Creates (or recreates) the named branch, optionally starting from | |
727 | a specific revision. The reset command allows a frontend to issue | |
728 | a new `from` command for an existing branch, or to create a new | |
729 | branch from an existing commit without creating a new commit. | |
730 | ||
731 | .... | |
732 | 'reset' SP <ref> LF | |
733 | ('from' SP <committish> LF)? | |
1fdb649c | 734 | LF? |
6e411d20 SP |
735 | .... |
736 | ||
737 | For a detailed description of `<ref>` and `<committish>` see above | |
738 | under `commit` and `from`. | |
739 | ||
1fdb649c SP |
740 | The `LF` after the command is optional (it used to be required). |
741 | ||
6e411d20 SP |
742 | The `reset` command can also be used to create lightweight |
743 | (non-annotated) tags. For example: | |
744 | ||
745 | ==== | |
746 | reset refs/tags/938 | |
747 | from :938 | |
748 | ==== | |
749 | ||
750 | would create the lightweight tag `refs/tags/938` referring to | |
751 | whatever commit mark `:938` references. | |
752 | ||
753 | `blob` | |
754 | ~~~~~~ | |
755 | Requests writing one file revision to the packfile. The revision | |
756 | is not connected to any commit; this connection must be formed in | |
757 | a subsequent `commit` command by referencing the blob through an | |
758 | assigned mark. | |
759 | ||
760 | .... | |
761 | 'blob' LF | |
762 | mark? | |
763 | data | |
764 | .... | |
765 | ||
766 | The mark command is optional here as some frontends have chosen | |
767 | to generate the Git SHA-1 for the blob on their own, and feed that | |
6a5d0b0a | 768 | directly to `commit`. This is typically more work than it's worth |
6e411d20 SP |
769 | however, as marks are inexpensive to store and easy to use. |
770 | ||
771 | `data` | |
772 | ~~~~~~ | |
773 | Supplies raw data (for use as blob/file content, commit messages, or | |
882227f1 | 774 | annotated tag messages) to fast-import. Data can be supplied using an exact |
6e411d20 SP |
775 | byte count or delimited with a terminating line. Real frontends |
776 | intended for production-quality conversions should always use the | |
777 | exact byte count format, as it is more robust and performs better. | |
882227f1 | 778 | The delimited format is intended primarily for testing fast-import. |
6e411d20 | 779 | |
401d53fa SP |
780 | Comment lines appearing within the `<raw>` part of `data` commands |
781 | are always taken to be part of the body of the data and are therefore | |
782 | never ignored by fast-import. This makes it safe to import any | |
783 | file/message content whose lines might start with `#`. | |
784 | ||
ef94edb5 SP |
785 | Exact byte count format:: |
786 | The frontend must specify the number of bytes of data. | |
787 | + | |
6e411d20 SP |
788 | .... |
789 | 'data' SP <count> LF | |
2c570cde | 790 | <raw> LF? |
6e411d20 | 791 | .... |
ef94edb5 | 792 | + |
6e411d20 | 793 | where `<count>` is the exact number of bytes appearing within |
ef94edb5 SP |
794 | `<raw>`. The value of `<count>` is expressed as an ASCII decimal |
795 | integer. The `LF` on either side of `<raw>` is not | |
6e411d20 | 796 | included in `<count>` and will not be included in the imported data. |
2c570cde SP |
797 | + |
798 | The `LF` after `<raw>` is optional (it used to be required) but | |
799 | recommended. Always including it makes debugging a fast-import | |
800 | stream easier as the next command always starts in column 0 | |
801 | of the next line, even if `<raw>` did not end with an `LF`. | |
6e411d20 | 802 | |
ef94edb5 SP |
803 | Delimited format:: |
804 | A delimiter string is used to mark the end of the data. | |
882227f1 | 805 | fast-import will compute the length by searching for the delimiter. |
02783075 | 806 | This format is primarily useful for testing and is not |
ef94edb5 SP |
807 | recommended for real data. |
808 | + | |
6e411d20 SP |
809 | .... |
810 | 'data' SP '<<' <delim> LF | |
811 | <raw> LF | |
812 | <delim> LF | |
2c570cde | 813 | LF? |
6e411d20 | 814 | .... |
ef94edb5 | 815 | + |
6e411d20 SP |
816 | where `<delim>` is the chosen delimiter string. The string `<delim>` |
817 | must not appear on a line by itself within `<raw>`, as otherwise | |
882227f1 | 818 | fast-import will think the data ends earlier than it really does. The `LF` |
6e411d20 SP |
819 | immediately trailing `<raw>` is part of `<raw>`. This is one of |
820 | the limitations of the delimited format, it is impossible to supply | |
821 | a data chunk which does not have an LF as its last byte. | |
2c570cde SP |
822 | + |
823 | The `LF` after `<delim> LF` is optional (it used to be required). | |
6e411d20 SP |
824 | |
825 | `checkpoint` | |
826 | ~~~~~~~~~~~~ | |
882227f1 | 827 | Forces fast-import to close the current packfile, start a new one, and to |
820b9310 | 828 | save out all current branch refs, tags and marks. |
6e411d20 SP |
829 | |
830 | .... | |
831 | 'checkpoint' LF | |
1fdb649c | 832 | LF? |
6e411d20 SP |
833 | .... |
834 | ||
882227f1 | 835 | Note that fast-import automatically switches packfiles when the current |
820b9310 | 836 | packfile reaches \--max-pack-size, or 4 GiB, whichever limit is |
882227f1 | 837 | smaller. During an automatic packfile switch fast-import does not update |
820b9310 SP |
838 | the branch refs, tags or marks. |
839 | ||
840 | As a `checkpoint` can require a significant amount of CPU time and | |
841 | disk IO (to compute the overall pack SHA-1 checksum, generate the | |
842 | corresponding index file, and update the refs) it can easily take | |
843 | several minutes for a single `checkpoint` command to complete. | |
844 | ||
845 | Frontends may choose to issue checkpoints during extremely large | |
846 | and long running imports, or when they need to allow another Git | |
847 | process access to a branch. However given that a 30 GiB Subversion | |
882227f1 | 848 | repository can be loaded into Git through fast-import in about 3 hours, |
820b9310 SP |
849 | explicit checkpointing may not be necessary. |
850 | ||
1fdb649c | 851 | The `LF` after the command is optional (it used to be required). |
820b9310 | 852 | |
ac053c02 SP |
853 | `progress` |
854 | ~~~~~~~~~~ | |
855 | Causes fast-import to print the entire `progress` line unmodified to | |
856 | its standard output channel (file descriptor 1) when the command is | |
857 | processed from the input stream. The command otherwise has no impact | |
858 | on the current import, or on any of fast-import's internal state. | |
859 | ||
860 | .... | |
861 | 'progress' SP <any> LF | |
862 | LF? | |
863 | .... | |
864 | ||
865 | The `<any>` part of the command may contain any sequence of bytes | |
866 | that does not contain `LF`. The `LF` after the command is optional. | |
867 | Callers may wish to process the output through a tool such as sed to | |
868 | remove the leading part of the line, for example: | |
869 | ||
870 | ==== | |
b1889c36 | 871 | frontend | git fast-import | sed 's/^progress //' |
ac053c02 SP |
872 | ==== |
873 | ||
874 | Placing a `progress` command immediately after a `checkpoint` will | |
875 | inform the reader when the `checkpoint` has been completed and it | |
876 | can safely access the refs that fast-import updated. | |
877 | ||
f963bd5d SR |
878 | `feature` |
879 | ~~~~~~~~~ | |
880 | Require that fast-import supports the specified feature, or abort if | |
881 | it does not. | |
882 | ||
883 | .... | |
884 | 'feature' SP <feature> LF | |
885 | .... | |
886 | ||
887 | The <feature> part of the command may be any string matching | |
888 | ^[a-zA-Z][a-zA-Z-]*$ and should be understood by fast-import. | |
889 | ||
081751c8 SR |
890 | Feature work identical as their option counterparts with the |
891 | exception of the import-marks feature, see below. | |
f963bd5d SR |
892 | |
893 | The following features are currently supported: | |
894 | ||
895 | * date-format | |
896 | * import-marks | |
897 | * export-marks | |
bc3c79ae SR |
898 | * relative-marks |
899 | * no-relative-marks | |
f963bd5d SR |
900 | * force |
901 | ||
081751c8 SR |
902 | The import-marks behaves differently from when it is specified as |
903 | commandline option in that only one "feature import-marks" is allowed | |
904 | per stream. Also, any --import-marks= specified on the commandline | |
905 | will override those from the stream (if any). | |
906 | ||
9c8398f0 SR |
907 | `option` |
908 | ~~~~~~~~ | |
909 | Processes the specified option so that git fast-import behaves in a | |
910 | way that suits the frontend's needs. | |
911 | Note that options specified by the frontend are overridden by any | |
912 | options the user may specify to git fast-import itself. | |
913 | ||
914 | .... | |
915 | 'option' SP <option> LF | |
916 | .... | |
917 | ||
918 | The `<option>` part of the command may contain any of the options | |
919 | listed in the OPTIONS section that do not change import semantics, | |
920 | without the leading '--' and is treated in the same way. | |
921 | ||
922 | Option commands must be the first commands on the input (not counting | |
923 | feature commands), to give an option command after any non-option | |
924 | command is an error. | |
925 | ||
926 | The following commandline options change import semantics and may therefore | |
927 | not be passed as option: | |
928 | ||
929 | * date-format | |
930 | * import-marks | |
931 | * export-marks | |
932 | * force | |
933 | ||
e7e5170f SP |
934 | Crash Reports |
935 | ------------- | |
936 | If fast-import is supplied invalid input it will terminate with a | |
937 | non-zero exit status and create a crash report in the top level of | |
938 | the Git repository it was importing into. Crash reports contain | |
939 | a snapshot of the internal fast-import state as well as the most | |
940 | recent commands that lead up to the crash. | |
941 | ||
942 | All recent commands (including stream comments, file changes and | |
943 | progress commands) are shown in the command history within the crash | |
944 | report, but raw file data and commit messages are excluded from the | |
945 | crash report. This exclusion saves space within the report file | |
946 | and reduces the amount of buffering that fast-import must perform | |
947 | during execution. | |
948 | ||
949 | After writing a crash report fast-import will close the current | |
950 | packfile and export the marks table. This allows the frontend | |
951 | developer to inspect the repository state and resume the import from | |
952 | the point where it crashed. The modified branches and tags are not | |
953 | updated during a crash, as the import did not complete successfully. | |
954 | Branch and tag information can be found in the crash report and | |
955 | must be applied manually if the update is needed. | |
956 | ||
957 | An example crash: | |
958 | ||
959 | ==== | |
960 | $ cat >in <<END_OF_INPUT | |
961 | # my very first test commit | |
962 | commit refs/heads/master | |
963 | committer Shawn O. Pearce <spearce> 19283 -0400 | |
964 | # who is that guy anyway? | |
965 | data <<EOF | |
966 | this is my commit | |
967 | EOF | |
968 | M 644 inline .gitignore | |
969 | data <<EOF | |
970 | .gitignore | |
971 | EOF | |
972 | M 777 inline bob | |
973 | END_OF_INPUT | |
974 | ||
b1889c36 | 975 | $ git fast-import <in |
e7e5170f SP |
976 | fatal: Corrupt mode: M 777 inline bob |
977 | fast-import: dumping crash report to .git/fast_import_crash_8434 | |
978 | ||
979 | $ cat .git/fast_import_crash_8434 | |
980 | fast-import crash report: | |
981 | fast-import process: 8434 | |
982 | parent process : 1391 | |
983 | at Sat Sep 1 00:58:12 2007 | |
984 | ||
985 | fatal: Corrupt mode: M 777 inline bob | |
986 | ||
987 | Most Recent Commands Before Crash | |
988 | --------------------------------- | |
989 | # my very first test commit | |
990 | commit refs/heads/master | |
991 | committer Shawn O. Pearce <spearce> 19283 -0400 | |
992 | # who is that guy anyway? | |
993 | data <<EOF | |
994 | M 644 inline .gitignore | |
995 | data <<EOF | |
996 | * M 777 inline bob | |
997 | ||
998 | Active Branch LRU | |
999 | ----------------- | |
1000 | active_branches = 1 cur, 5 max | |
1001 | ||
1002 | pos clock name | |
1003 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
1004 | 1) 0 refs/heads/master | |
1005 | ||
1006 | Inactive Branches | |
1007 | ----------------- | |
1008 | refs/heads/master: | |
1009 | status : active loaded dirty | |
1010 | tip commit : 0000000000000000000000000000000000000000 | |
1011 | old tree : 0000000000000000000000000000000000000000 | |
1012 | cur tree : 0000000000000000000000000000000000000000 | |
1013 | commit clock: 0 | |
1014 | last pack : | |
1015 | ||
1016 | ||
1017 | ------------------- | |
1018 | END OF CRASH REPORT | |
1019 | ==== | |
1020 | ||
bdd9f424 SP |
1021 | Tips and Tricks |
1022 | --------------- | |
1023 | The following tips and tricks have been collected from various | |
882227f1 | 1024 | users of fast-import, and are offered here as suggestions. |
bdd9f424 SP |
1025 | |
1026 | Use One Mark Per Commit | |
1027 | ~~~~~~~~~~~~~~~~~~~~~~~ | |
1028 | When doing a repository conversion, use a unique mark per commit | |
1029 | (`mark :<n>`) and supply the \--export-marks option on the command | |
882227f1 | 1030 | line. fast-import will dump a file which lists every mark and the Git |
bdd9f424 SP |
1031 | object SHA-1 that corresponds to it. If the frontend can tie |
1032 | the marks back to the source repository, it is easy to verify the | |
1033 | accuracy and completeness of the import by comparing each Git | |
1034 | commit to the corresponding source revision. | |
1035 | ||
1036 | Coming from a system such as Perforce or Subversion this should be | |
882227f1 | 1037 | quite simple, as the fast-import mark can also be the Perforce changeset |
bdd9f424 SP |
1038 | number or the Subversion revision number. |
1039 | ||
1040 | Freely Skip Around Branches | |
1041 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
1042 | Don't bother trying to optimize the frontend to stick to one branch | |
1043 | at a time during an import. Although doing so might be slightly | |
882227f1 | 1044 | faster for fast-import, it tends to increase the complexity of the frontend |
bdd9f424 SP |
1045 | code considerably. |
1046 | ||
882227f1 | 1047 | The branch LRU builtin to fast-import tends to behave very well, and the |
bdd9f424 SP |
1048 | cost of activating an inactive branch is so low that bouncing around |
1049 | between branches has virtually no impact on import performance. | |
1050 | ||
c7346156 SP |
1051 | Handling Renames |
1052 | ~~~~~~~~~~~~~~~~ | |
1053 | When importing a renamed file or directory, simply delete the old | |
1054 | name(s) and modify the new name(s) during the corresponding commit. | |
1055 | Git performs rename detection after-the-fact, rather than explicitly | |
1056 | during a commit. | |
1057 | ||
bdd9f424 SP |
1058 | Use Tag Fixup Branches |
1059 | ~~~~~~~~~~~~~~~~~~~~~~ | |
1060 | Some other SCM systems let the user create a tag from multiple | |
1061 | files which are not from the same commit/changeset. Or to create | |
1062 | tags which are a subset of the files available in the repository. | |
1063 | ||
1064 | Importing these tags as-is in Git is impossible without making at | |
1065 | least one commit which ``fixes up'' the files to match the content | |
882227f1 | 1066 | of the tag. Use fast-import's `reset` command to reset a dummy branch |
bdd9f424 SP |
1067 | outside of your normal branch space to the base commit for the tag, |
1068 | then commit one or more file fixup commits, and finally tag the | |
1069 | dummy branch. | |
1070 | ||
1071 | For example since all normal branches are stored under `refs/heads/` | |
1072 | name the tag fixup branch `TAG_FIXUP`. This way it is impossible for | |
1073 | the fixup branch used by the importer to have namespace conflicts | |
1074 | with real branches imported from the source (the name `TAG_FIXUP` | |
1075 | is not `refs/heads/TAG_FIXUP`). | |
1076 | ||
1077 | When committing fixups, consider using `merge` to connect the | |
1078 | commit(s) which are supplying file revisions to the fixup branch. | |
0b444cdb | 1079 | Doing so will allow tools such as 'git blame' to track |
bdd9f424 SP |
1080 | through the real commit history and properly annotate the source |
1081 | files. | |
1082 | ||
882227f1 | 1083 | After fast-import terminates the frontend will need to do `rm .git/TAG_FIXUP` |
bdd9f424 SP |
1084 | to remove the dummy branch. |
1085 | ||
1086 | Import Now, Repack Later | |
1087 | ~~~~~~~~~~~~~~~~~~~~~~~~ | |
882227f1 | 1088 | As soon as fast-import completes the Git repository is completely valid |
02783075 | 1089 | and ready for use. Typically this takes only a very short time, |
bdd9f424 SP |
1090 | even for considerably large projects (100,000+ commits). |
1091 | ||
1092 | However repacking the repository is necessary to improve data | |
1093 | locality and access performance. It can also take hours on extremely | |
1094 | large projects (especially if -f and a large \--window parameter is | |
1095 | used). Since repacking is safe to run alongside readers and writers, | |
1096 | run the repack in the background and let it finish when it finishes. | |
1097 | There is no reason to wait to explore your new Git project! | |
1098 | ||
1099 | If you choose to wait for the repack, don't try to run benchmarks | |
882227f1 | 1100 | or performance tests until repacking is completed. fast-import outputs |
bdd9f424 SP |
1101 | suboptimal packfiles that are simply never seen in real use |
1102 | situations. | |
1103 | ||
1104 | Repacking Historical Data | |
1105 | ~~~~~~~~~~~~~~~~~~~~~~~~~ | |
1106 | If you are repacking very old imported data (e.g. older than the | |
1107 | last year), consider expending some extra CPU time and supplying | |
0b444cdb | 1108 | \--window=50 (or higher) when you run 'git repack'. |
bdd9f424 SP |
1109 | This will take longer, but will also produce a smaller packfile. |
1110 | You only need to expend the effort once, and everyone using your | |
1111 | project will benefit from the smaller repository. | |
1112 | ||
ac053c02 SP |
1113 | Include Some Progress Messages |
1114 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
1115 | Every once in a while have your frontend emit a `progress` message | |
1116 | to fast-import. The contents of the messages are entirely free-form, | |
1117 | so one suggestion would be to output the current month and year | |
1118 | each time the current commit date moves into the next month. | |
1119 | Your users will feel better knowing how much of the data stream | |
1120 | has been processed. | |
1121 | ||
bdd9f424 | 1122 | |
6e411d20 SP |
1123 | Packfile Optimization |
1124 | --------------------- | |
882227f1 | 1125 | When packing a blob fast-import always attempts to deltify against the last |
6e411d20 SP |
1126 | blob written. Unless specifically arranged for by the frontend, |
1127 | this will probably not be a prior version of the same file, so the | |
1128 | generated delta will not be the smallest possible. The resulting | |
1129 | packfile will be compressed, but will not be optimal. | |
1130 | ||
1131 | Frontends which have efficient access to all revisions of a | |
1132 | single file (for example reading an RCS/CVS ,v file) can choose | |
1133 | to supply all revisions of that file as a sequence of consecutive | |
882227f1 | 1134 | `blob` commands. This allows fast-import to deltify the different file |
6e411d20 SP |
1135 | revisions against each other, saving space in the final packfile. |
1136 | Marks can be used to later identify individual file revisions during | |
1137 | a sequence of `commit` commands. | |
1138 | ||
882227f1 SP |
1139 | The packfile(s) created by fast-import do not encourage good disk access |
1140 | patterns. This is caused by fast-import writing the data in the order | |
6e411d20 SP |
1141 | it is received on standard input, while Git typically organizes |
1142 | data within packfiles to make the most recent (current tip) data | |
1143 | appear before historical data. Git also clusters commits together, | |
1144 | speeding up revision traversal through better cache locality. | |
1145 | ||
1146 | For this reason it is strongly recommended that users repack the | |
882227f1 | 1147 | repository with `git repack -a -d` after fast-import completes, allowing |
6e411d20 SP |
1148 | Git to reorganize the packfiles for faster data access. If blob |
1149 | deltas are suboptimal (see above) then also adding the `-f` option | |
1150 | to force recomputation of all deltas can significantly reduce the | |
1151 | final packfile size (30-50% smaller can be quite typical). | |
1152 | ||
bdd9f424 | 1153 | |
6e411d20 SP |
1154 | Memory Utilization |
1155 | ------------------ | |
882227f1 | 1156 | There are a number of factors which affect how much memory fast-import |
6e411d20 | 1157 | requires to perform an import. Like critical sections of core |
02783075 BH |
1158 | Git, fast-import uses its own memory allocators to amortize any overheads |
1159 | associated with malloc. In practice fast-import tends to amortize any | |
6e411d20 SP |
1160 | malloc overheads to 0, due to its use of large block allocations. |
1161 | ||
1162 | per object | |
1163 | ~~~~~~~~~~ | |
882227f1 | 1164 | fast-import maintains an in-memory structure for every object written in |
6e411d20 SP |
1165 | this execution. On a 32 bit system the structure is 32 bytes, |
1166 | on a 64 bit system the structure is 40 bytes (due to the larger | |
1167 | pointer sizes). Objects in the table are not deallocated until | |
882227f1 | 1168 | fast-import terminates. Importing 2 million objects on a 32 bit system |
6e411d20 SP |
1169 | will require approximately 64 MiB of memory. |
1170 | ||
1171 | The object table is actually a hashtable keyed on the object name | |
882227f1 | 1172 | (the unique SHA-1). This storage configuration allows fast-import to reuse |
6e411d20 SP |
1173 | an existing or already written object and avoid writing duplicates |
1174 | to the output packfile. Duplicate blobs are surprisingly common | |
1175 | in an import, typically due to branch merges in the source. | |
1176 | ||
1177 | per mark | |
1178 | ~~~~~~~~ | |
1179 | Marks are stored in a sparse array, using 1 pointer (4 bytes or 8 | |
1180 | bytes, depending on pointer size) per mark. Although the array | |
1181 | is sparse, frontends are still strongly encouraged to use marks | |
1182 | between 1 and n, where n is the total number of marks required for | |
1183 | this import. | |
1184 | ||
1185 | per branch | |
1186 | ~~~~~~~~~~ | |
1187 | Branches are classified as active and inactive. The memory usage | |
1188 | of the two classes is significantly different. | |
1189 | ||
1190 | Inactive branches are stored in a structure which uses 96 or 120 | |
1191 | bytes (32 bit or 64 bit systems, respectively), plus the length of | |
882227f1 | 1192 | the branch name (typically under 200 bytes), per branch. fast-import will |
6e411d20 SP |
1193 | easily handle as many as 10,000 inactive branches in under 2 MiB |
1194 | of memory. | |
1195 | ||
1196 | Active branches have the same overhead as inactive branches, but | |
1197 | also contain copies of every tree that has been recently modified on | |
1198 | that branch. If subtree `include` has not been modified since the | |
1199 | branch became active, its contents will not be loaded into memory, | |
1200 | but if subtree `src` has been modified by a commit since the branch | |
1201 | became active, then its contents will be loaded in memory. | |
1202 | ||
1203 | As active branches store metadata about the files contained on that | |
1204 | branch, their in-memory storage size can grow to a considerable size | |
1205 | (see below). | |
1206 | ||
882227f1 | 1207 | fast-import automatically moves active branches to inactive status based on |
6e411d20 SP |
1208 | a simple least-recently-used algorithm. The LRU chain is updated on |
1209 | each `commit` command. The maximum number of active branches can be | |
c499d768 | 1210 | increased or decreased on the command line with \--active-branches=. |
6e411d20 SP |
1211 | |
1212 | per active tree | |
1213 | ~~~~~~~~~~~~~~~ | |
1214 | Trees (aka directories) use just 12 bytes of memory on top of the | |
1215 | memory required for their entries (see ``per active file'' below). | |
02783075 | 1216 | The cost of a tree is virtually 0, as its overhead amortizes out |
6e411d20 SP |
1217 | over the individual file entries. |
1218 | ||
1219 | per active file entry | |
1220 | ~~~~~~~~~~~~~~~~~~~~~ | |
1221 | Files (and pointers to subtrees) within active trees require 52 or 64 | |
1222 | bytes (32/64 bit platforms) per entry. To conserve space, file and | |
1223 | tree names are pooled in a common string table, allowing the filename | |
1224 | ``Makefile'' to use just 16 bytes (after including the string header | |
1225 | overhead) no matter how many times it occurs within the project. | |
1226 | ||
1227 | The active branch LRU, when coupled with the filename string pool | |
882227f1 | 1228 | and lazy loading of subtrees, allows fast-import to efficiently import |
6e411d20 SP |
1229 | projects with 2,000+ branches and 45,114+ files in a very limited |
1230 | memory footprint (less than 2.7 MiB per active branch). | |
1231 | ||
1232 | ||
1233 | Author | |
1234 | ------ | |
1235 | Written by Shawn O. Pearce <spearce@spearce.org>. | |
1236 | ||
1237 | Documentation | |
1238 | -------------- | |
1239 | Documentation by Shawn O. Pearce <spearce@spearce.org>. | |
1240 | ||
1241 | GIT | |
1242 | --- | |
9e1f0a85 | 1243 | Part of the linkgit:git[1] suite |