]>
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 | 47 | Maximum size of each output packfile. |
89e0a3a1 | 48 | The default is unlimited. |
6e411d20 | 49 | |
5eef828b SP |
50 | --big-file-threshold=<n>:: |
51 | Maximum size of a blob that fast-import will attempt to | |
52 | create a delta for, expressed in bytes. The default is 512m | |
53 | (512 MiB). Some importers may wish to lower this on systems | |
54 | with constrained memory. | |
55 | ||
6e411d20 SP |
56 | --depth=<n>:: |
57 | Maximum delta depth, for blob and tree deltification. | |
58 | Default is 10. | |
59 | ||
60 | --active-branches=<n>:: | |
61 | Maximum number of branches to maintain active at once. | |
62 | See ``Memory Utilization'' below for details. Default is 5. | |
63 | ||
64 | --export-marks=<file>:: | |
65 | Dumps the internal marks table to <file> when complete. | |
66 | Marks are written one per line as `:markid SHA-1`. | |
67 | Frontends can use this file to validate imports after they | |
e8438420 SP |
68 | have been completed, or to save the marks table across |
69 | incremental runs. As <file> is only opened and truncated | |
70 | at checkpoint (or completion) the same path can also be | |
71 | safely given to \--import-marks. | |
72 | ||
73 | --import-marks=<file>:: | |
74 | Before processing any input, load the marks specified in | |
75 | <file>. The input file must exist, must be readable, and | |
76 | must use the same format as produced by \--export-marks. | |
77 | Multiple options may be supplied to import more than one | |
78 | set of marks. If a mark is defined to different values, | |
79 | the last file wins. | |
6e411d20 | 80 | |
bc3c79ae SR |
81 | --relative-marks:: |
82 | After specifying --relative-marks= the paths specified | |
83 | with --import-marks= and --export-marks= are relative | |
84 | to an internal directory in the current repository. | |
85 | In git-fast-import this means that the paths are relative | |
86 | to the .git/info/fast-import directory. However, other | |
87 | importers may use a different location. | |
88 | ||
89 | --no-relative-marks:: | |
90 | Negates a previous --relative-marks. Allows for combining | |
91 | relative and non-relative marks by interweaving | |
92 | --(no-)-relative-marks= with the --(import|export)-marks= | |
93 | options. | |
94 | ||
bdf1c06d SP |
95 | --export-pack-edges=<file>:: |
96 | After creating a packfile, print a line of data to | |
97 | <file> listing the filename of the packfile and the last | |
98 | commit on each branch that was written to that packfile. | |
99 | This information may be useful after importing projects | |
100 | whose total object set exceeds the 4 GiB packfile limit, | |
101 | as these commits can be used as edge points during calls | |
0b444cdb | 102 | to 'git pack-objects'. |
bdf1c06d | 103 | |
c499d768 | 104 | --quiet:: |
882227f1 | 105 | Disable all non-fatal output, making fast-import silent when it |
7f9d77f2 | 106 | is successful. This option disables the output shown by |
c499d768 SP |
107 | \--stats. |
108 | ||
109 | --stats:: | |
882227f1 | 110 | Display some basic statistics about the objects fast-import has |
c499d768 | 111 | created, the packfiles they were stored into, and the |
882227f1 | 112 | memory used by fast-import during this run. Showing this output |
c499d768 SP |
113 | is currently the default, but can be disabled with \--quiet. |
114 | ||
115 | ||
6e411d20 SP |
116 | Performance |
117 | ----------- | |
882227f1 | 118 | The design of fast-import allows it to import large projects in a minimum |
6e411d20 | 119 | amount of memory usage and processing time. Assuming the frontend |
882227f1 | 120 | is able to keep up with fast-import and feed it a constant stream of data, |
6e411d20 SP |
121 | import times for projects holding 10+ years of history and containing |
122 | 100,000+ individual commits are generally completed in just 1-2 | |
123 | hours on quite modest (~$2,000 USD) hardware. | |
124 | ||
125 | Most bottlenecks appear to be in foreign source data access (the | |
882227f1 | 126 | source just cannot extract revisions fast enough) or disk IO (fast-import |
6e411d20 SP |
127 | writes as fast as the disk will take the data). Imports will run |
128 | faster if the source data is stored on a different drive than the | |
129 | destination Git repository (due to less IO contention). | |
130 | ||
131 | ||
132 | Development Cost | |
133 | ---------------- | |
882227f1 | 134 | A typical frontend for fast-import tends to weigh in at approximately 200 |
6e411d20 SP |
135 | lines of Perl/Python/Ruby code. Most developers have been able to |
136 | create working importers in just a couple of hours, even though it | |
882227f1 | 137 | is their first exposure to fast-import, and sometimes even to Git. This is |
6e411d20 SP |
138 | an ideal situation, given that most conversion tools are throw-away |
139 | (use once, and never look back). | |
140 | ||
141 | ||
142 | Parallel Operation | |
143 | ------------------ | |
0b444cdb | 144 | Like 'git push' or 'git fetch', imports handled by fast-import are safe to |
6e411d20 | 145 | run alongside parallel `git repack -a -d` or `git gc` invocations, |
0b444cdb | 146 | or any other Git operation (including 'git prune', as loose objects |
882227f1 | 147 | are never used by fast-import). |
6e411d20 | 148 | |
882227f1 SP |
149 | fast-import does not lock the branch or tag refs it is actively importing. |
150 | After the import, during its ref update phase, fast-import tests each | |
7073e69e SP |
151 | existing branch ref to verify the update will be a fast-forward |
152 | update (the commit stored in the ref is contained in the new | |
153 | history of the commit to be written). If the update is not a | |
882227f1 SP |
154 | fast-forward update, fast-import will skip updating that ref and instead |
155 | prints a warning message. fast-import will always attempt to update all | |
7073e69e SP |
156 | branch refs, and does not stop on the first failure. |
157 | ||
6a5d0b0a | 158 | Branch updates can be forced with \--force, but it's recommended that |
c499d768 | 159 | this only be used on an otherwise quiet repository. Using \--force |
7073e69e | 160 | is not necessary for an initial import into an empty repository. |
6e411d20 SP |
161 | |
162 | ||
163 | Technical Discussion | |
164 | -------------------- | |
882227f1 | 165 | fast-import tracks a set of branches in memory. Any branch can be created |
6e411d20 SP |
166 | or modified at any point during the import process by sending a |
167 | `commit` command on the input stream. This design allows a frontend | |
168 | program to process an unlimited number of branches simultaneously, | |
169 | generating commits in the order they are available from the source | |
170 | data. It also simplifies the frontend programs considerably. | |
171 | ||
882227f1 | 172 | fast-import does not use or alter the current working directory, or any |
6e411d20 SP |
173 | file within it. (It does however update the current Git repository, |
174 | as referenced by `GIT_DIR`.) Therefore an import frontend may use | |
175 | the working directory for its own purposes, such as extracting file | |
176 | revisions from the foreign source. This ignorance of the working | |
882227f1 | 177 | directory also allows fast-import to run very quickly, as it does not |
6e411d20 SP |
178 | need to perform any costly file update operations when switching |
179 | between branches. | |
180 | ||
181 | Input Format | |
182 | ------------ | |
183 | With the exception of raw file data (which Git does not interpret) | |
882227f1 | 184 | the fast-import input format is text (ASCII) based. This text based |
6e411d20 SP |
185 | format simplifies development and debugging of frontend programs, |
186 | especially when a higher level language such as Perl, Python or | |
187 | Ruby is being used. | |
188 | ||
882227f1 | 189 | fast-import is very strict about its input. Where we say SP below we mean |
6e411d20 SP |
190 | *exactly* one space. Likewise LF means one (and only one) linefeed. |
191 | Supplying additional whitespace characters will cause unexpected | |
192 | results, such as branch names or file names with leading or trailing | |
882227f1 | 193 | spaces in their name, or early termination of fast-import when it encounters |
6e411d20 SP |
194 | unexpected input. |
195 | ||
401d53fa SP |
196 | Stream Comments |
197 | ~~~~~~~~~~~~~~~ | |
198 | To aid in debugging frontends fast-import ignores any line that | |
199 | begins with `#` (ASCII pound/hash) up to and including the line | |
200 | ending `LF`. A comment line may contain any sequence of bytes | |
201 | that does not contain an LF and therefore may be used to include | |
202 | any detailed debugging information that might be specific to the | |
203 | frontend and useful when inspecting a fast-import data stream. | |
204 | ||
63e0c8b3 SP |
205 | Date Formats |
206 | ~~~~~~~~~~~~ | |
207 | The following date formats are supported. A frontend should select | |
208 | the format it will use for this import by passing the format name | |
c499d768 | 209 | in the \--date-format=<fmt> command line option. |
63e0c8b3 SP |
210 | |
211 | `raw`:: | |
9b92c82f | 212 | This is the Git native format and is `<time> SP <offutc>`. |
882227f1 | 213 | It is also fast-import's default format, if \--date-format was |
63e0c8b3 SP |
214 | not specified. |
215 | + | |
216 | The time of the event is specified by `<time>` as the number of | |
217 | seconds since the UNIX epoch (midnight, Jan 1, 1970, UTC) and is | |
218 | written as an ASCII decimal integer. | |
219 | + | |
9b92c82f SP |
220 | The local offset is specified by `<offutc>` as a positive or negative |
221 | offset from UTC. For example EST (which is 5 hours behind UTC) | |
222 | would be expressed in `<tz>` by ``-0500'' while UTC is ``+0000''. | |
223 | The local offset does not affect `<time>`; it is used only as an | |
224 | advisement to help formatting routines display the timestamp. | |
63e0c8b3 | 225 | + |
9b92c82f SP |
226 | If the local offset is not available in the source material, use |
227 | ``+0000'', or the most common local offset. For example many | |
63e0c8b3 SP |
228 | organizations have a CVS repository which has only ever been accessed |
229 | by users who are located in the same location and timezone. In this | |
f842fdb0 | 230 | case a reasonable offset from UTC could be assumed. |
63e0c8b3 SP |
231 | + |
232 | Unlike the `rfc2822` format, this format is very strict. Any | |
882227f1 | 233 | variation in formatting will cause fast-import to reject the value. |
63e0c8b3 SP |
234 | |
235 | `rfc2822`:: | |
236 | This is the standard email format as described by RFC 2822. | |
237 | + | |
238 | An example value is ``Tue Feb 6 11:22:18 2007 -0500''. The Git | |
f842fdb0 | 239 | parser is accurate, but a little on the lenient side. It is the |
0b444cdb | 240 | same parser used by 'git am' when applying patches |
63e0c8b3 SP |
241 | received from email. |
242 | + | |
243 | Some malformed strings may be accepted as valid dates. In some of | |
244 | these cases Git will still be able to obtain the correct date from | |
245 | the malformed string. There are also some types of malformed | |
246 | strings which Git will parse wrong, and yet consider valid. | |
247 | Seriously malformed strings will be rejected. | |
248 | + | |
9b92c82f SP |
249 | Unlike the `raw` format above, the timezone/UTC offset information |
250 | contained in an RFC 2822 date string is used to adjust the date | |
251 | value to UTC prior to storage. Therefore it is important that | |
252 | this information be as accurate as possible. | |
253 | + | |
f842fdb0 | 254 | If the source material uses RFC 2822 style dates, |
882227f1 | 255 | the frontend should let fast-import handle the parsing and conversion |
63e0c8b3 SP |
256 | (rather than attempting to do it itself) as the Git parser has |
257 | been well tested in the wild. | |
258 | + | |
259 | Frontends should prefer the `raw` format if the source material | |
f842fdb0 | 260 | already uses UNIX-epoch format, can be coaxed to give dates in that |
02783075 | 261 | format, or its format is easily convertible to it, as there is no |
f842fdb0 | 262 | ambiguity in parsing. |
63e0c8b3 SP |
263 | |
264 | `now`:: | |
265 | Always use the current time and timezone. The literal | |
266 | `now` must always be supplied for `<when>`. | |
267 | + | |
268 | This is a toy format. The current time and timezone of this system | |
269 | is always copied into the identity string at the time it is being | |
882227f1 | 270 | created by fast-import. There is no way to specify a different time or |
63e0c8b3 SP |
271 | timezone. |
272 | + | |
6a5d0b0a | 273 | This particular format is supplied as it's short to implement and |
63e0c8b3 SP |
274 | may be useful to a process that wants to create a new commit |
275 | right now, without needing to use a working directory or | |
0b444cdb | 276 | 'git update-index'. |
63e0c8b3 SP |
277 | + |
278 | If separate `author` and `committer` commands are used in a `commit` | |
279 | the timestamps may not match, as the system clock will be polled | |
280 | twice (once for each command). The only way to ensure that both | |
281 | author and committer identity information has the same timestamp | |
282 | is to omit `author` (thus copying from `committer`) or to use a | |
283 | date format other than `now`. | |
284 | ||
6e411d20 SP |
285 | Commands |
286 | ~~~~~~~~ | |
882227f1 | 287 | fast-import accepts several commands to update the current repository |
6e411d20 SP |
288 | and control the current import process. More detailed discussion |
289 | (with examples) of each command follows later. | |
290 | ||
291 | `commit`:: | |
292 | Creates a new branch or updates an existing branch by | |
293 | creating a new commit and updating the branch to point at | |
294 | the newly created commit. | |
295 | ||
296 | `tag`:: | |
297 | Creates an annotated tag object from an existing commit or | |
298 | branch. Lightweight tags are not supported by this command, | |
299 | as they are not recommended for recording meaningful points | |
300 | in time. | |
301 | ||
302 | `reset`:: | |
303 | Reset an existing branch (or a new branch) to a specific | |
304 | revision. This command must be used to change a branch to | |
305 | a specific revision without making a commit on it. | |
306 | ||
307 | `blob`:: | |
308 | Convert raw file data into a blob, for future use in a | |
309 | `commit` command. This command is optional and is not | |
310 | needed to perform an import. | |
311 | ||
312 | `checkpoint`:: | |
882227f1 | 313 | Forces fast-import to close the current packfile, generate its |
6e411d20 SP |
314 | unique SHA-1 checksum and index, and start a new packfile. |
315 | This command is optional and is not needed to perform | |
316 | an import. | |
317 | ||
ac053c02 SP |
318 | `progress`:: |
319 | Causes fast-import to echo the entire line to its own | |
320 | standard output. This command is optional and is not needed | |
321 | to perform an import. | |
322 | ||
f963bd5d SR |
323 | `feature`:: |
324 | Require that fast-import supports the specified feature, or | |
325 | abort if it does not. | |
326 | ||
9c8398f0 SR |
327 | `option`:: |
328 | Specify any of the options listed under OPTIONS that do not | |
329 | change stream semantic to suit the frontend's needs. This | |
330 | command is optional and is not needed to perform an import. | |
331 | ||
6e411d20 SP |
332 | `commit` |
333 | ~~~~~~~~ | |
334 | Create or update a branch with a new commit, recording one logical | |
335 | change to the project. | |
336 | ||
337 | .... | |
338 | 'commit' SP <ref> LF | |
339 | mark? | |
74fbd118 SP |
340 | ('author' (SP <name>)? SP LT <email> GT SP <when> LF)? |
341 | 'committer' (SP <name>)? SP LT <email> GT SP <when> LF | |
6e411d20 SP |
342 | data |
343 | ('from' SP <committish> LF)? | |
344 | ('merge' SP <committish> LF)? | |
a8dd2e7d | 345 | (filemodify | filedelete | filecopy | filerename | filedeleteall | notemodify)* |
1fdb649c | 346 | LF? |
6e411d20 SP |
347 | .... |
348 | ||
349 | where `<ref>` is the name of the branch to make the commit on. | |
350 | Typically branch names are prefixed with `refs/heads/` in | |
351 | Git, so importing the CVS branch symbol `RELENG-1_0` would use | |
352 | `refs/heads/RELENG-1_0` for the value of `<ref>`. The value of | |
353 | `<ref>` must be a valid refname in Git. As `LF` is not valid in | |
354 | a Git refname, no quoting or escaping syntax is supported here. | |
355 | ||
882227f1 | 356 | A `mark` command may optionally appear, requesting fast-import to save a |
6e411d20 SP |
357 | reference to the newly created commit for future use by the frontend |
358 | (see below for format). It is very common for frontends to mark | |
359 | every commit they create, thereby allowing future branch creation | |
360 | from any imported commit. | |
361 | ||
362 | The `data` command following `committer` must supply the commit | |
363 | message (see below for `data` command syntax). To import an empty | |
364 | commit message use a 0 length data. Commit messages are free-form | |
365 | and are not interpreted by Git. Currently they must be encoded in | |
882227f1 | 366 | UTF-8, as fast-import does not permit other encodings to be specified. |
6e411d20 | 367 | |
a8dd2e7d JH |
368 | Zero or more `filemodify`, `filedelete`, `filecopy`, `filerename`, |
369 | `filedeleteall` and `notemodify` commands | |
825769a8 SP |
370 | may be included to update the contents of the branch prior to |
371 | creating the commit. These commands may be supplied in any order. | |
02783075 | 372 | However it is recommended that a `filedeleteall` command precede |
a8dd2e7d JH |
373 | all `filemodify`, `filecopy`, `filerename` and `notemodify` commands in |
374 | the same commit, as `filedeleteall` wipes the branch clean (see below). | |
6e411d20 | 375 | |
1fdb649c SP |
376 | The `LF` after the command is optional (it used to be required). |
377 | ||
6e411d20 SP |
378 | `author` |
379 | ^^^^^^^^ | |
380 | An `author` command may optionally appear, if the author information | |
381 | might differ from the committer information. If `author` is omitted | |
882227f1 | 382 | then fast-import will automatically use the committer's information for |
6e411d20 SP |
383 | the author portion of the commit. See below for a description of |
384 | the fields in `author`, as they are identical to `committer`. | |
385 | ||
386 | `committer` | |
387 | ^^^^^^^^^^^ | |
388 | The `committer` command indicates who made this commit, and when | |
389 | they made it. | |
390 | ||
391 | Here `<name>` is the person's display name (for example | |
392 | ``Com M Itter'') and `<email>` is the person's email address | |
393 | (``cm@example.com''). `LT` and `GT` are the literal less-than (\x3c) | |
394 | and greater-than (\x3e) symbols. These are required to delimit | |
395 | the email address from the other fields in the line. Note that | |
396 | `<name>` is free-form and may contain any sequence of bytes, except | |
397 | `LT` and `LF`. It is typically UTF-8 encoded. | |
398 | ||
63e0c8b3 | 399 | The time of the change is specified by `<when>` using the date format |
c499d768 | 400 | that was selected by the \--date-format=<fmt> command line option. |
63e0c8b3 SP |
401 | See ``Date Formats'' above for the set of supported formats, and |
402 | their syntax. | |
6e411d20 SP |
403 | |
404 | `from` | |
405 | ^^^^^^ | |
ea5e370a SP |
406 | The `from` command is used to specify the commit to initialize |
407 | this branch from. This revision will be the first ancestor of the | |
408 | new commit. | |
409 | ||
410 | Omitting the `from` command in the first commit of a new branch | |
411 | will cause fast-import to create that commit with no ancestor. This | |
412 | tends to be desired only for the initial commit of a project. | |
9b33fa08 EB |
413 | If the frontend creates all files from scratch when making a new |
414 | branch, a `merge` command may be used instead of `from` to start | |
415 | the commit with an empty tree. | |
ea5e370a SP |
416 | Omitting the `from` command on existing branches is usually desired, |
417 | as the current commit on that branch is automatically assumed to | |
418 | be the first ancestor of the new commit. | |
6e411d20 SP |
419 | |
420 | As `LF` is not valid in a Git refname or SHA-1 expression, no | |
421 | quoting or escaping syntax is supported within `<committish>`. | |
422 | ||
423 | Here `<committish>` is any of the following: | |
424 | ||
882227f1 | 425 | * The name of an existing branch already in fast-import's internal branch |
6a5d0b0a | 426 | table. If fast-import doesn't know the name, it's treated as a SHA-1 |
6e411d20 SP |
427 | expression. |
428 | ||
429 | * A mark reference, `:<idnum>`, where `<idnum>` is the mark number. | |
430 | + | |
882227f1 | 431 | The reason fast-import uses `:` to denote a mark reference is this character |
6e411d20 | 432 | is not legal in a Git branch name. The leading `:` makes it easy |
02783075 | 433 | to distinguish between the mark 42 (`:42`) and the branch 42 (`42` |
6e411d20 SP |
434 | or `refs/heads/42`), or an abbreviated SHA-1 which happened to |
435 | consist only of base-10 digits. | |
436 | + | |
437 | Marks must be declared (via `mark`) before they can be used. | |
438 | ||
439 | * A complete 40 byte or abbreviated commit SHA-1 in hex. | |
440 | ||
441 | * Any valid Git SHA-1 expression that resolves to a commit. See | |
f028cdae | 442 | ``SPECIFYING REVISIONS'' in linkgit:gitrevisions[1] for details. |
6e411d20 SP |
443 | |
444 | The special case of restarting an incremental import from the | |
445 | current branch value should be written as: | |
446 | ---- | |
447 | from refs/heads/branch^0 | |
448 | ---- | |
882227f1 | 449 | The `{caret}0` suffix is necessary as fast-import does not permit a branch to |
6e411d20 | 450 | start from itself, and the branch is created in memory before the |
209f1298 | 451 | `from` command is even read from the input. Adding `{caret}0` will force |
882227f1 | 452 | fast-import to resolve the commit through Git's revision parsing library, |
6e411d20 SP |
453 | rather than its internal branch table, thereby loading in the |
454 | existing value of the branch. | |
455 | ||
456 | `merge` | |
457 | ^^^^^^^ | |
9b33fa08 EB |
458 | Includes one additional ancestor commit. If the `from` command is |
459 | omitted when creating a new branch, the first `merge` commit will be | |
460 | the first ancestor of the current commit, and the branch will start | |
461 | out with no files. An unlimited number of `merge` commands per | |
882227f1 | 462 | commit are permitted by fast-import, thereby establishing an n-way merge. |
6e411d20 SP |
463 | However Git's other tools never create commits with more than 15 |
464 | additional ancestors (forming a 16-way merge). For this reason | |
465 | it is suggested that frontends do not use more than 15 `merge` | |
9b33fa08 | 466 | commands per commit; 16, if starting a new, empty branch. |
6e411d20 SP |
467 | |
468 | Here `<committish>` is any of the commit specification expressions | |
469 | also accepted by `from` (see above). | |
470 | ||
471 | `filemodify` | |
ef94edb5 | 472 | ^^^^^^^^^^^^ |
6e411d20 SP |
473 | Included in a `commit` command to add a new file or change the |
474 | content of an existing file. This command has two different means | |
475 | of specifying the content of the file. | |
476 | ||
477 | External data format:: | |
478 | The data content for the file was already supplied by a prior | |
479 | `blob` command. The frontend just needs to connect it. | |
480 | + | |
481 | .... | |
482 | 'M' SP <mode> SP <dataref> SP <path> LF | |
483 | .... | |
484 | + | |
485 | Here `<dataref>` can be either a mark reference (`:<idnum>`) | |
486 | set by a prior `blob` command, or a full 40-byte SHA-1 of an | |
487 | existing Git blob object. | |
488 | ||
489 | Inline data format:: | |
490 | The data content for the file has not been supplied yet. | |
491 | The frontend wants to supply it as part of this modify | |
492 | command. | |
493 | + | |
494 | .... | |
495 | 'M' SP <mode> SP 'inline' SP <path> LF | |
496 | data | |
497 | .... | |
498 | + | |
499 | See below for a detailed description of the `data` command. | |
500 | ||
501 | In both formats `<mode>` is the type of file entry, specified | |
502 | in octal. Git only supports the following modes: | |
503 | ||
504 | * `100644` or `644`: A normal (not-executable) file. The majority | |
505 | of files in most projects use this mode. If in doubt, this is | |
506 | what you want. | |
507 | * `100755` or `755`: A normal, but executable, file. | |
9981b6d9 | 508 | * `120000`: A symlink, the content of the file will be the link target. |
03db4525 AG |
509 | * `160000`: A gitlink, SHA-1 of the object refers to a commit in |
510 | another repository. Git links can only be specified by SHA or through | |
511 | a commit mark. They are used to implement submodules. | |
6e411d20 SP |
512 | |
513 | In both formats `<path>` is the complete path of the file to be added | |
514 | (if not already existing) or modified (if already existing). | |
515 | ||
c4431d38 | 516 | A `<path>` string must use UNIX-style directory separators (forward |
6e411d20 SP |
517 | slash `/`), may contain any byte other than `LF`, and must not |
518 | start with double quote (`"`). | |
519 | ||
520 | If an `LF` or double quote must be encoded into `<path>` shell-style | |
521 | quoting should be used, e.g. `"path/with\n and \" in it"`. | |
522 | ||
02783075 | 523 | The value of `<path>` must be in canonical form. That is it must not: |
6e411d20 SP |
524 | |
525 | * contain an empty directory component (e.g. `foo//bar` is invalid), | |
c4431d38 JK |
526 | * end with a directory separator (e.g. `foo/` is invalid), |
527 | * start with a directory separator (e.g. `/foo` is invalid), | |
6e411d20 SP |
528 | * contain the special component `.` or `..` (e.g. `foo/./bar` and |
529 | `foo/../bar` are invalid). | |
530 | ||
531 | It is recommended that `<path>` always be encoded using UTF-8. | |
532 | ||
6e411d20 | 533 | `filedelete` |
ef94edb5 | 534 | ^^^^^^^^^^^^ |
512e44b2 SP |
535 | Included in a `commit` command to remove a file or recursively |
536 | delete an entire directory from the branch. If the file or directory | |
537 | removal makes its parent directory empty, the parent directory will | |
6e411d20 SP |
538 | be automatically removed too. This cascades up the tree until the |
539 | first non-empty directory or the root is reached. | |
540 | ||
541 | .... | |
542 | 'D' SP <path> LF | |
543 | .... | |
544 | ||
512e44b2 SP |
545 | here `<path>` is the complete path of the file or subdirectory to |
546 | be removed from the branch. | |
6e411d20 SP |
547 | See `filemodify` above for a detailed description of `<path>`. |
548 | ||
b6f3481b SP |
549 | `filecopy` |
550 | ^^^^^^^^^^^^ | |
551 | Recursively copies an existing file or subdirectory to a different | |
552 | location within the branch. The existing file or directory must | |
553 | exist. If the destination exists it will be completely replaced | |
554 | by the content copied from the source. | |
555 | ||
556 | .... | |
557 | 'C' SP <path> SP <path> LF | |
558 | .... | |
559 | ||
560 | here the first `<path>` is the source location and the second | |
561 | `<path>` is the destination. See `filemodify` above for a detailed | |
562 | description of what `<path>` may look like. To use a source path | |
563 | that contains SP the path must be quoted. | |
564 | ||
565 | A `filecopy` command takes effect immediately. Once the source | |
566 | location has been copied to the destination any future commands | |
567 | applied to the source location will not impact the destination of | |
568 | the copy. | |
569 | ||
f39a946a SP |
570 | `filerename` |
571 | ^^^^^^^^^^^^ | |
572 | Renames an existing file or subdirectory to a different location | |
573 | within the branch. The existing file or directory must exist. If | |
574 | the destination exists it will be replaced by the source directory. | |
575 | ||
576 | .... | |
577 | 'R' SP <path> SP <path> LF | |
578 | .... | |
579 | ||
580 | here the first `<path>` is the source location and the second | |
581 | `<path>` is the destination. See `filemodify` above for a detailed | |
582 | description of what `<path>` may look like. To use a source path | |
583 | that contains SP the path must be quoted. | |
584 | ||
585 | A `filerename` command takes effect immediately. Once the source | |
586 | location has been renamed to the destination any future commands | |
587 | applied to the source location will create new files there and not | |
588 | impact the destination of the rename. | |
589 | ||
b6f3481b SP |
590 | Note that a `filerename` is the same as a `filecopy` followed by a |
591 | `filedelete` of the source location. There is a slight performance | |
592 | advantage to using `filerename`, but the advantage is so small | |
593 | that it is never worth trying to convert a delete/add pair in | |
594 | source material into a rename for fast-import. This `filerename` | |
595 | command is provided just to simplify frontends that already have | |
596 | rename information and don't want bother with decomposing it into a | |
597 | `filecopy` followed by a `filedelete`. | |
598 | ||
825769a8 SP |
599 | `filedeleteall` |
600 | ^^^^^^^^^^^^^^^ | |
601 | Included in a `commit` command to remove all files (and also all | |
602 | directories) from the branch. This command resets the internal | |
603 | branch structure to have no files in it, allowing the frontend | |
604 | to subsequently add all interesting files from scratch. | |
605 | ||
606 | .... | |
607 | 'deleteall' LF | |
608 | .... | |
609 | ||
610 | This command is extremely useful if the frontend does not know | |
611 | (or does not care to know) what files are currently on the branch, | |
612 | and therefore cannot generate the proper `filedelete` commands to | |
613 | update the content. | |
614 | ||
615 | Issuing a `filedeleteall` followed by the needed `filemodify` | |
616 | commands to set the correct content will produce the same results | |
617 | as sending only the needed `filemodify` and `filedelete` commands. | |
882227f1 | 618 | The `filedeleteall` approach may however require fast-import to use slightly |
825769a8 SP |
619 | more memory per active branch (less than 1 MiB for even most large |
620 | projects); so frontends that can easily obtain only the affected | |
621 | paths for a commit are encouraged to do so. | |
622 | ||
a8dd2e7d JH |
623 | `notemodify` |
624 | ^^^^^^^^^^^^ | |
625 | Included in a `commit` command to add a new note (annotating a given | |
626 | commit) or change the content of an existing note. This command has | |
627 | two different means of specifying the content of the note. | |
628 | ||
629 | External data format:: | |
630 | The data content for the note was already supplied by a prior | |
631 | `blob` command. The frontend just needs to connect it to the | |
632 | commit that is to be annotated. | |
633 | + | |
634 | .... | |
635 | 'N' SP <dataref> SP <committish> LF | |
636 | .... | |
637 | + | |
638 | Here `<dataref>` can be either a mark reference (`:<idnum>`) | |
639 | set by a prior `blob` command, or a full 40-byte SHA-1 of an | |
640 | existing Git blob object. | |
641 | ||
642 | Inline data format:: | |
643 | The data content for the note has not been supplied yet. | |
644 | The frontend wants to supply it as part of this modify | |
645 | command. | |
646 | + | |
647 | .... | |
648 | 'N' SP 'inline' SP <committish> LF | |
649 | data | |
650 | .... | |
651 | + | |
652 | See below for a detailed description of the `data` command. | |
653 | ||
654 | In both formats `<committish>` is any of the commit specification | |
655 | expressions also accepted by `from` (see above). | |
656 | ||
6e411d20 SP |
657 | `mark` |
658 | ~~~~~~ | |
882227f1 | 659 | Arranges for fast-import to save a reference to the current object, allowing |
6e411d20 SP |
660 | the frontend to recall this object at a future point in time, without |
661 | knowing its SHA-1. Here the current object is the object creation | |
662 | command the `mark` command appears within. This can be `commit`, | |
663 | `tag`, and `blob`, but `commit` is the most common usage. | |
664 | ||
665 | .... | |
666 | 'mark' SP ':' <idnum> LF | |
667 | .... | |
668 | ||
669 | where `<idnum>` is the number assigned by the frontend to this mark. | |
ef94edb5 SP |
670 | The value of `<idnum>` is expressed as an ASCII decimal integer. |
671 | The value 0 is reserved and cannot be used as | |
6e411d20 SP |
672 | a mark. Only values greater than or equal to 1 may be used as marks. |
673 | ||
674 | New marks are created automatically. Existing marks can be moved | |
675 | to another object simply by reusing the same `<idnum>` in another | |
676 | `mark` command. | |
677 | ||
678 | `tag` | |
679 | ~~~~~ | |
680 | Creates an annotated tag referring to a specific commit. To create | |
681 | lightweight (non-annotated) tags see the `reset` command below. | |
682 | ||
683 | .... | |
684 | 'tag' SP <name> LF | |
685 | 'from' SP <committish> LF | |
74fbd118 | 686 | 'tagger' (SP <name>)? SP LT <email> GT SP <when> LF |
6e411d20 | 687 | data |
6e411d20 SP |
688 | .... |
689 | ||
690 | where `<name>` is the name of the tag to create. | |
691 | ||
692 | Tag names are automatically prefixed with `refs/tags/` when stored | |
693 | in Git, so importing the CVS branch symbol `RELENG-1_0-FINAL` would | |
882227f1 | 694 | use just `RELENG-1_0-FINAL` for `<name>`, and fast-import will write the |
6e411d20 SP |
695 | corresponding ref as `refs/tags/RELENG-1_0-FINAL`. |
696 | ||
697 | The value of `<name>` must be a valid refname in Git and therefore | |
698 | may contain forward slashes. As `LF` is not valid in a Git refname, | |
699 | no quoting or escaping syntax is supported here. | |
700 | ||
701 | The `from` command is the same as in the `commit` command; see | |
702 | above for details. | |
703 | ||
704 | The `tagger` command uses the same format as `committer` within | |
705 | `commit`; again see above for details. | |
706 | ||
707 | The `data` command following `tagger` must supply the annotated tag | |
708 | message (see below for `data` command syntax). To import an empty | |
709 | tag message use a 0 length data. Tag messages are free-form and are | |
710 | not interpreted by Git. Currently they must be encoded in UTF-8, | |
882227f1 | 711 | as fast-import does not permit other encodings to be specified. |
6e411d20 | 712 | |
882227f1 | 713 | Signing annotated tags during import from within fast-import is not |
6e411d20 SP |
714 | supported. Trying to include your own PGP/GPG signature is not |
715 | recommended, as the frontend does not (easily) have access to the | |
716 | complete set of bytes which normally goes into such a signature. | |
882227f1 | 717 | If signing is required, create lightweight tags from within fast-import with |
6e411d20 | 718 | `reset`, then create the annotated versions of those tags offline |
0b444cdb | 719 | with the standard 'git tag' process. |
6e411d20 SP |
720 | |
721 | `reset` | |
722 | ~~~~~~~ | |
723 | Creates (or recreates) the named branch, optionally starting from | |
724 | a specific revision. The reset command allows a frontend to issue | |
725 | a new `from` command for an existing branch, or to create a new | |
726 | branch from an existing commit without creating a new commit. | |
727 | ||
728 | .... | |
729 | 'reset' SP <ref> LF | |
730 | ('from' SP <committish> LF)? | |
1fdb649c | 731 | LF? |
6e411d20 SP |
732 | .... |
733 | ||
734 | For a detailed description of `<ref>` and `<committish>` see above | |
735 | under `commit` and `from`. | |
736 | ||
1fdb649c SP |
737 | The `LF` after the command is optional (it used to be required). |
738 | ||
6e411d20 SP |
739 | The `reset` command can also be used to create lightweight |
740 | (non-annotated) tags. For example: | |
741 | ||
742 | ==== | |
743 | reset refs/tags/938 | |
744 | from :938 | |
745 | ==== | |
746 | ||
747 | would create the lightweight tag `refs/tags/938` referring to | |
748 | whatever commit mark `:938` references. | |
749 | ||
750 | `blob` | |
751 | ~~~~~~ | |
752 | Requests writing one file revision to the packfile. The revision | |
753 | is not connected to any commit; this connection must be formed in | |
754 | a subsequent `commit` command by referencing the blob through an | |
755 | assigned mark. | |
756 | ||
757 | .... | |
758 | 'blob' LF | |
759 | mark? | |
760 | data | |
761 | .... | |
762 | ||
763 | The mark command is optional here as some frontends have chosen | |
764 | to generate the Git SHA-1 for the blob on their own, and feed that | |
6a5d0b0a | 765 | directly to `commit`. This is typically more work than it's worth |
6e411d20 SP |
766 | however, as marks are inexpensive to store and easy to use. |
767 | ||
768 | `data` | |
769 | ~~~~~~ | |
770 | Supplies raw data (for use as blob/file content, commit messages, or | |
882227f1 | 771 | annotated tag messages) to fast-import. Data can be supplied using an exact |
6e411d20 SP |
772 | byte count or delimited with a terminating line. Real frontends |
773 | intended for production-quality conversions should always use the | |
774 | exact byte count format, as it is more robust and performs better. | |
882227f1 | 775 | The delimited format is intended primarily for testing fast-import. |
6e411d20 | 776 | |
401d53fa SP |
777 | Comment lines appearing within the `<raw>` part of `data` commands |
778 | are always taken to be part of the body of the data and are therefore | |
779 | never ignored by fast-import. This makes it safe to import any | |
780 | file/message content whose lines might start with `#`. | |
781 | ||
ef94edb5 SP |
782 | Exact byte count format:: |
783 | The frontend must specify the number of bytes of data. | |
784 | + | |
6e411d20 SP |
785 | .... |
786 | 'data' SP <count> LF | |
2c570cde | 787 | <raw> LF? |
6e411d20 | 788 | .... |
ef94edb5 | 789 | + |
6e411d20 | 790 | where `<count>` is the exact number of bytes appearing within |
ef94edb5 SP |
791 | `<raw>`. The value of `<count>` is expressed as an ASCII decimal |
792 | integer. The `LF` on either side of `<raw>` is not | |
6e411d20 | 793 | included in `<count>` and will not be included in the imported data. |
2c570cde SP |
794 | + |
795 | The `LF` after `<raw>` is optional (it used to be required) but | |
796 | recommended. Always including it makes debugging a fast-import | |
797 | stream easier as the next command always starts in column 0 | |
798 | of the next line, even if `<raw>` did not end with an `LF`. | |
6e411d20 | 799 | |
ef94edb5 SP |
800 | Delimited format:: |
801 | A delimiter string is used to mark the end of the data. | |
882227f1 | 802 | fast-import will compute the length by searching for the delimiter. |
02783075 | 803 | This format is primarily useful for testing and is not |
ef94edb5 SP |
804 | recommended for real data. |
805 | + | |
6e411d20 SP |
806 | .... |
807 | 'data' SP '<<' <delim> LF | |
808 | <raw> LF | |
809 | <delim> LF | |
2c570cde | 810 | LF? |
6e411d20 | 811 | .... |
ef94edb5 | 812 | + |
6e411d20 SP |
813 | where `<delim>` is the chosen delimiter string. The string `<delim>` |
814 | must not appear on a line by itself within `<raw>`, as otherwise | |
882227f1 | 815 | fast-import will think the data ends earlier than it really does. The `LF` |
6e411d20 SP |
816 | immediately trailing `<raw>` is part of `<raw>`. This is one of |
817 | the limitations of the delimited format, it is impossible to supply | |
818 | a data chunk which does not have an LF as its last byte. | |
2c570cde SP |
819 | + |
820 | The `LF` after `<delim> LF` is optional (it used to be required). | |
6e411d20 SP |
821 | |
822 | `checkpoint` | |
823 | ~~~~~~~~~~~~ | |
882227f1 | 824 | Forces fast-import to close the current packfile, start a new one, and to |
820b9310 | 825 | save out all current branch refs, tags and marks. |
6e411d20 SP |
826 | |
827 | .... | |
828 | 'checkpoint' LF | |
1fdb649c | 829 | LF? |
6e411d20 SP |
830 | .... |
831 | ||
882227f1 | 832 | Note that fast-import automatically switches packfiles when the current |
820b9310 | 833 | packfile reaches \--max-pack-size, or 4 GiB, whichever limit is |
882227f1 | 834 | smaller. During an automatic packfile switch fast-import does not update |
820b9310 SP |
835 | the branch refs, tags or marks. |
836 | ||
837 | As a `checkpoint` can require a significant amount of CPU time and | |
838 | disk IO (to compute the overall pack SHA-1 checksum, generate the | |
839 | corresponding index file, and update the refs) it can easily take | |
840 | several minutes for a single `checkpoint` command to complete. | |
841 | ||
842 | Frontends may choose to issue checkpoints during extremely large | |
843 | and long running imports, or when they need to allow another Git | |
844 | process access to a branch. However given that a 30 GiB Subversion | |
882227f1 | 845 | repository can be loaded into Git through fast-import in about 3 hours, |
820b9310 SP |
846 | explicit checkpointing may not be necessary. |
847 | ||
1fdb649c | 848 | The `LF` after the command is optional (it used to be required). |
820b9310 | 849 | |
ac053c02 SP |
850 | `progress` |
851 | ~~~~~~~~~~ | |
852 | Causes fast-import to print the entire `progress` line unmodified to | |
853 | its standard output channel (file descriptor 1) when the command is | |
854 | processed from the input stream. The command otherwise has no impact | |
855 | on the current import, or on any of fast-import's internal state. | |
856 | ||
857 | .... | |
858 | 'progress' SP <any> LF | |
859 | LF? | |
860 | .... | |
861 | ||
862 | The `<any>` part of the command may contain any sequence of bytes | |
863 | that does not contain `LF`. The `LF` after the command is optional. | |
864 | Callers may wish to process the output through a tool such as sed to | |
865 | remove the leading part of the line, for example: | |
866 | ||
867 | ==== | |
b1889c36 | 868 | frontend | git fast-import | sed 's/^progress //' |
ac053c02 SP |
869 | ==== |
870 | ||
871 | Placing a `progress` command immediately after a `checkpoint` will | |
872 | inform the reader when the `checkpoint` has been completed and it | |
873 | can safely access the refs that fast-import updated. | |
874 | ||
f963bd5d SR |
875 | `feature` |
876 | ~~~~~~~~~ | |
877 | Require that fast-import supports the specified feature, or abort if | |
878 | it does not. | |
879 | ||
880 | .... | |
881 | 'feature' SP <feature> LF | |
882 | .... | |
883 | ||
884 | The <feature> part of the command may be any string matching | |
885 | ^[a-zA-Z][a-zA-Z-]*$ and should be understood by fast-import. | |
886 | ||
081751c8 SR |
887 | Feature work identical as their option counterparts with the |
888 | exception of the import-marks feature, see below. | |
f963bd5d SR |
889 | |
890 | The following features are currently supported: | |
891 | ||
892 | * date-format | |
893 | * import-marks | |
894 | * export-marks | |
bc3c79ae SR |
895 | * relative-marks |
896 | * no-relative-marks | |
f963bd5d SR |
897 | * force |
898 | ||
081751c8 SR |
899 | The import-marks behaves differently from when it is specified as |
900 | commandline option in that only one "feature import-marks" is allowed | |
901 | per stream. Also, any --import-marks= specified on the commandline | |
902 | will override those from the stream (if any). | |
903 | ||
9c8398f0 SR |
904 | `option` |
905 | ~~~~~~~~ | |
906 | Processes the specified option so that git fast-import behaves in a | |
907 | way that suits the frontend's needs. | |
908 | Note that options specified by the frontend are overridden by any | |
909 | options the user may specify to git fast-import itself. | |
910 | ||
911 | .... | |
912 | 'option' SP <option> LF | |
913 | .... | |
914 | ||
915 | The `<option>` part of the command may contain any of the options | |
916 | listed in the OPTIONS section that do not change import semantics, | |
917 | without the leading '--' and is treated in the same way. | |
918 | ||
919 | Option commands must be the first commands on the input (not counting | |
920 | feature commands), to give an option command after any non-option | |
921 | command is an error. | |
922 | ||
923 | The following commandline options change import semantics and may therefore | |
924 | not be passed as option: | |
925 | ||
926 | * date-format | |
927 | * import-marks | |
928 | * export-marks | |
929 | * force | |
930 | ||
e7e5170f SP |
931 | Crash Reports |
932 | ------------- | |
933 | If fast-import is supplied invalid input it will terminate with a | |
934 | non-zero exit status and create a crash report in the top level of | |
935 | the Git repository it was importing into. Crash reports contain | |
936 | a snapshot of the internal fast-import state as well as the most | |
937 | recent commands that lead up to the crash. | |
938 | ||
939 | All recent commands (including stream comments, file changes and | |
940 | progress commands) are shown in the command history within the crash | |
941 | report, but raw file data and commit messages are excluded from the | |
942 | crash report. This exclusion saves space within the report file | |
943 | and reduces the amount of buffering that fast-import must perform | |
944 | during execution. | |
945 | ||
946 | After writing a crash report fast-import will close the current | |
947 | packfile and export the marks table. This allows the frontend | |
948 | developer to inspect the repository state and resume the import from | |
949 | the point where it crashed. The modified branches and tags are not | |
950 | updated during a crash, as the import did not complete successfully. | |
951 | Branch and tag information can be found in the crash report and | |
952 | must be applied manually if the update is needed. | |
953 | ||
954 | An example crash: | |
955 | ||
956 | ==== | |
957 | $ cat >in <<END_OF_INPUT | |
958 | # my very first test commit | |
959 | commit refs/heads/master | |
960 | committer Shawn O. Pearce <spearce> 19283 -0400 | |
961 | # who is that guy anyway? | |
962 | data <<EOF | |
963 | this is my commit | |
964 | EOF | |
965 | M 644 inline .gitignore | |
966 | data <<EOF | |
967 | .gitignore | |
968 | EOF | |
969 | M 777 inline bob | |
970 | END_OF_INPUT | |
971 | ||
b1889c36 | 972 | $ git fast-import <in |
e7e5170f SP |
973 | fatal: Corrupt mode: M 777 inline bob |
974 | fast-import: dumping crash report to .git/fast_import_crash_8434 | |
975 | ||
976 | $ cat .git/fast_import_crash_8434 | |
977 | fast-import crash report: | |
978 | fast-import process: 8434 | |
979 | parent process : 1391 | |
980 | at Sat Sep 1 00:58:12 2007 | |
981 | ||
982 | fatal: Corrupt mode: M 777 inline bob | |
983 | ||
984 | Most Recent Commands Before Crash | |
985 | --------------------------------- | |
986 | # my very first test commit | |
987 | commit refs/heads/master | |
988 | committer Shawn O. Pearce <spearce> 19283 -0400 | |
989 | # who is that guy anyway? | |
990 | data <<EOF | |
991 | M 644 inline .gitignore | |
992 | data <<EOF | |
993 | * M 777 inline bob | |
994 | ||
995 | Active Branch LRU | |
996 | ----------------- | |
997 | active_branches = 1 cur, 5 max | |
998 | ||
999 | pos clock name | |
1000 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
1001 | 1) 0 refs/heads/master | |
1002 | ||
1003 | Inactive Branches | |
1004 | ----------------- | |
1005 | refs/heads/master: | |
1006 | status : active loaded dirty | |
1007 | tip commit : 0000000000000000000000000000000000000000 | |
1008 | old tree : 0000000000000000000000000000000000000000 | |
1009 | cur tree : 0000000000000000000000000000000000000000 | |
1010 | commit clock: 0 | |
1011 | last pack : | |
1012 | ||
1013 | ||
1014 | ------------------- | |
1015 | END OF CRASH REPORT | |
1016 | ==== | |
1017 | ||
bdd9f424 SP |
1018 | Tips and Tricks |
1019 | --------------- | |
1020 | The following tips and tricks have been collected from various | |
882227f1 | 1021 | users of fast-import, and are offered here as suggestions. |
bdd9f424 SP |
1022 | |
1023 | Use One Mark Per Commit | |
1024 | ~~~~~~~~~~~~~~~~~~~~~~~ | |
1025 | When doing a repository conversion, use a unique mark per commit | |
1026 | (`mark :<n>`) and supply the \--export-marks option on the command | |
882227f1 | 1027 | line. fast-import will dump a file which lists every mark and the Git |
bdd9f424 SP |
1028 | object SHA-1 that corresponds to it. If the frontend can tie |
1029 | the marks back to the source repository, it is easy to verify the | |
1030 | accuracy and completeness of the import by comparing each Git | |
1031 | commit to the corresponding source revision. | |
1032 | ||
1033 | Coming from a system such as Perforce or Subversion this should be | |
882227f1 | 1034 | quite simple, as the fast-import mark can also be the Perforce changeset |
bdd9f424 SP |
1035 | number or the Subversion revision number. |
1036 | ||
1037 | Freely Skip Around Branches | |
1038 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
1039 | Don't bother trying to optimize the frontend to stick to one branch | |
1040 | at a time during an import. Although doing so might be slightly | |
882227f1 | 1041 | faster for fast-import, it tends to increase the complexity of the frontend |
bdd9f424 SP |
1042 | code considerably. |
1043 | ||
882227f1 | 1044 | The branch LRU builtin to fast-import tends to behave very well, and the |
bdd9f424 SP |
1045 | cost of activating an inactive branch is so low that bouncing around |
1046 | between branches has virtually no impact on import performance. | |
1047 | ||
c7346156 SP |
1048 | Handling Renames |
1049 | ~~~~~~~~~~~~~~~~ | |
1050 | When importing a renamed file or directory, simply delete the old | |
1051 | name(s) and modify the new name(s) during the corresponding commit. | |
1052 | Git performs rename detection after-the-fact, rather than explicitly | |
1053 | during a commit. | |
1054 | ||
bdd9f424 SP |
1055 | Use Tag Fixup Branches |
1056 | ~~~~~~~~~~~~~~~~~~~~~~ | |
1057 | Some other SCM systems let the user create a tag from multiple | |
1058 | files which are not from the same commit/changeset. Or to create | |
1059 | tags which are a subset of the files available in the repository. | |
1060 | ||
1061 | Importing these tags as-is in Git is impossible without making at | |
1062 | least one commit which ``fixes up'' the files to match the content | |
882227f1 | 1063 | of the tag. Use fast-import's `reset` command to reset a dummy branch |
bdd9f424 SP |
1064 | outside of your normal branch space to the base commit for the tag, |
1065 | then commit one or more file fixup commits, and finally tag the | |
1066 | dummy branch. | |
1067 | ||
1068 | For example since all normal branches are stored under `refs/heads/` | |
1069 | name the tag fixup branch `TAG_FIXUP`. This way it is impossible for | |
1070 | the fixup branch used by the importer to have namespace conflicts | |
1071 | with real branches imported from the source (the name `TAG_FIXUP` | |
1072 | is not `refs/heads/TAG_FIXUP`). | |
1073 | ||
1074 | When committing fixups, consider using `merge` to connect the | |
1075 | commit(s) which are supplying file revisions to the fixup branch. | |
0b444cdb | 1076 | Doing so will allow tools such as 'git blame' to track |
bdd9f424 SP |
1077 | through the real commit history and properly annotate the source |
1078 | files. | |
1079 | ||
882227f1 | 1080 | After fast-import terminates the frontend will need to do `rm .git/TAG_FIXUP` |
bdd9f424 SP |
1081 | to remove the dummy branch. |
1082 | ||
1083 | Import Now, Repack Later | |
1084 | ~~~~~~~~~~~~~~~~~~~~~~~~ | |
882227f1 | 1085 | As soon as fast-import completes the Git repository is completely valid |
02783075 | 1086 | and ready for use. Typically this takes only a very short time, |
bdd9f424 SP |
1087 | even for considerably large projects (100,000+ commits). |
1088 | ||
1089 | However repacking the repository is necessary to improve data | |
1090 | locality and access performance. It can also take hours on extremely | |
1091 | large projects (especially if -f and a large \--window parameter is | |
1092 | used). Since repacking is safe to run alongside readers and writers, | |
1093 | run the repack in the background and let it finish when it finishes. | |
1094 | There is no reason to wait to explore your new Git project! | |
1095 | ||
1096 | If you choose to wait for the repack, don't try to run benchmarks | |
882227f1 | 1097 | or performance tests until repacking is completed. fast-import outputs |
bdd9f424 SP |
1098 | suboptimal packfiles that are simply never seen in real use |
1099 | situations. | |
1100 | ||
1101 | Repacking Historical Data | |
1102 | ~~~~~~~~~~~~~~~~~~~~~~~~~ | |
1103 | If you are repacking very old imported data (e.g. older than the | |
1104 | last year), consider expending some extra CPU time and supplying | |
0b444cdb | 1105 | \--window=50 (or higher) when you run 'git repack'. |
bdd9f424 SP |
1106 | This will take longer, but will also produce a smaller packfile. |
1107 | You only need to expend the effort once, and everyone using your | |
1108 | project will benefit from the smaller repository. | |
1109 | ||
ac053c02 SP |
1110 | Include Some Progress Messages |
1111 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
1112 | Every once in a while have your frontend emit a `progress` message | |
1113 | to fast-import. The contents of the messages are entirely free-form, | |
1114 | so one suggestion would be to output the current month and year | |
1115 | each time the current commit date moves into the next month. | |
1116 | Your users will feel better knowing how much of the data stream | |
1117 | has been processed. | |
1118 | ||
bdd9f424 | 1119 | |
6e411d20 SP |
1120 | Packfile Optimization |
1121 | --------------------- | |
882227f1 | 1122 | When packing a blob fast-import always attempts to deltify against the last |
6e411d20 SP |
1123 | blob written. Unless specifically arranged for by the frontend, |
1124 | this will probably not be a prior version of the same file, so the | |
1125 | generated delta will not be the smallest possible. The resulting | |
1126 | packfile will be compressed, but will not be optimal. | |
1127 | ||
1128 | Frontends which have efficient access to all revisions of a | |
1129 | single file (for example reading an RCS/CVS ,v file) can choose | |
1130 | to supply all revisions of that file as a sequence of consecutive | |
882227f1 | 1131 | `blob` commands. This allows fast-import to deltify the different file |
6e411d20 SP |
1132 | revisions against each other, saving space in the final packfile. |
1133 | Marks can be used to later identify individual file revisions during | |
1134 | a sequence of `commit` commands. | |
1135 | ||
882227f1 SP |
1136 | The packfile(s) created by fast-import do not encourage good disk access |
1137 | patterns. This is caused by fast-import writing the data in the order | |
6e411d20 SP |
1138 | it is received on standard input, while Git typically organizes |
1139 | data within packfiles to make the most recent (current tip) data | |
1140 | appear before historical data. Git also clusters commits together, | |
1141 | speeding up revision traversal through better cache locality. | |
1142 | ||
1143 | For this reason it is strongly recommended that users repack the | |
882227f1 | 1144 | repository with `git repack -a -d` after fast-import completes, allowing |
6e411d20 SP |
1145 | Git to reorganize the packfiles for faster data access. If blob |
1146 | deltas are suboptimal (see above) then also adding the `-f` option | |
1147 | to force recomputation of all deltas can significantly reduce the | |
1148 | final packfile size (30-50% smaller can be quite typical). | |
1149 | ||
bdd9f424 | 1150 | |
6e411d20 SP |
1151 | Memory Utilization |
1152 | ------------------ | |
882227f1 | 1153 | There are a number of factors which affect how much memory fast-import |
6e411d20 | 1154 | requires to perform an import. Like critical sections of core |
02783075 BH |
1155 | Git, fast-import uses its own memory allocators to amortize any overheads |
1156 | associated with malloc. In practice fast-import tends to amortize any | |
6e411d20 SP |
1157 | malloc overheads to 0, due to its use of large block allocations. |
1158 | ||
1159 | per object | |
1160 | ~~~~~~~~~~ | |
882227f1 | 1161 | fast-import maintains an in-memory structure for every object written in |
6e411d20 SP |
1162 | this execution. On a 32 bit system the structure is 32 bytes, |
1163 | on a 64 bit system the structure is 40 bytes (due to the larger | |
1164 | pointer sizes). Objects in the table are not deallocated until | |
882227f1 | 1165 | fast-import terminates. Importing 2 million objects on a 32 bit system |
6e411d20 SP |
1166 | will require approximately 64 MiB of memory. |
1167 | ||
1168 | The object table is actually a hashtable keyed on the object name | |
882227f1 | 1169 | (the unique SHA-1). This storage configuration allows fast-import to reuse |
6e411d20 SP |
1170 | an existing or already written object and avoid writing duplicates |
1171 | to the output packfile. Duplicate blobs are surprisingly common | |
1172 | in an import, typically due to branch merges in the source. | |
1173 | ||
1174 | per mark | |
1175 | ~~~~~~~~ | |
1176 | Marks are stored in a sparse array, using 1 pointer (4 bytes or 8 | |
1177 | bytes, depending on pointer size) per mark. Although the array | |
1178 | is sparse, frontends are still strongly encouraged to use marks | |
1179 | between 1 and n, where n is the total number of marks required for | |
1180 | this import. | |
1181 | ||
1182 | per branch | |
1183 | ~~~~~~~~~~ | |
1184 | Branches are classified as active and inactive. The memory usage | |
1185 | of the two classes is significantly different. | |
1186 | ||
1187 | Inactive branches are stored in a structure which uses 96 or 120 | |
1188 | bytes (32 bit or 64 bit systems, respectively), plus the length of | |
882227f1 | 1189 | the branch name (typically under 200 bytes), per branch. fast-import will |
6e411d20 SP |
1190 | easily handle as many as 10,000 inactive branches in under 2 MiB |
1191 | of memory. | |
1192 | ||
1193 | Active branches have the same overhead as inactive branches, but | |
1194 | also contain copies of every tree that has been recently modified on | |
1195 | that branch. If subtree `include` has not been modified since the | |
1196 | branch became active, its contents will not be loaded into memory, | |
1197 | but if subtree `src` has been modified by a commit since the branch | |
1198 | became active, then its contents will be loaded in memory. | |
1199 | ||
1200 | As active branches store metadata about the files contained on that | |
1201 | branch, their in-memory storage size can grow to a considerable size | |
1202 | (see below). | |
1203 | ||
882227f1 | 1204 | fast-import automatically moves active branches to inactive status based on |
6e411d20 SP |
1205 | a simple least-recently-used algorithm. The LRU chain is updated on |
1206 | each `commit` command. The maximum number of active branches can be | |
c499d768 | 1207 | increased or decreased on the command line with \--active-branches=. |
6e411d20 SP |
1208 | |
1209 | per active tree | |
1210 | ~~~~~~~~~~~~~~~ | |
1211 | Trees (aka directories) use just 12 bytes of memory on top of the | |
1212 | memory required for their entries (see ``per active file'' below). | |
02783075 | 1213 | The cost of a tree is virtually 0, as its overhead amortizes out |
6e411d20 SP |
1214 | over the individual file entries. |
1215 | ||
1216 | per active file entry | |
1217 | ~~~~~~~~~~~~~~~~~~~~~ | |
1218 | Files (and pointers to subtrees) within active trees require 52 or 64 | |
1219 | bytes (32/64 bit platforms) per entry. To conserve space, file and | |
1220 | tree names are pooled in a common string table, allowing the filename | |
1221 | ``Makefile'' to use just 16 bytes (after including the string header | |
1222 | overhead) no matter how many times it occurs within the project. | |
1223 | ||
1224 | The active branch LRU, when coupled with the filename string pool | |
882227f1 | 1225 | and lazy loading of subtrees, allows fast-import to efficiently import |
6e411d20 SP |
1226 | projects with 2,000+ branches and 45,114+ files in a very limited |
1227 | memory footprint (less than 2.7 MiB per active branch). | |
1228 | ||
1229 | ||
1230 | Author | |
1231 | ------ | |
1232 | Written by Shawn O. Pearce <spearce@spearce.org>. | |
1233 | ||
1234 | Documentation | |
1235 | -------------- | |
1236 | Documentation by Shawn O. Pearce <spearce@spearce.org>. | |
1237 | ||
1238 | GIT | |
1239 | --- | |
9e1f0a85 | 1240 | Part of the linkgit:git[1] suite |