frontend | git fast-import [options]
This program is usually not what the end user wants to run directly. Most end users want to use one of the existing frontend programs, which parses a specific type of foreign source and feeds the contents stored there to git-fast-import.
fast-import reads a mixed command/data stream from standard input and writes one or more packfiles directly into the current repository. When EOF is received on standard input, fast import writes out updated branch and tag refs, fully updating the current repository with the newly imported data.
The fast-import backend itself can import into an empty repository (one that has already been initialized by git-init) or incrementally update an existing populated repository. Whether or not incremental imports are supported from a particular foreign source depends on the frontend program in use.
--date-format=<fmt>
--force
--max-pack-size=<n>
--depth=<n>
--active-branches=<n>
--export-marks=<file>
--import-marks=<file>
--export-pack-edges=<file>
--quiet
--stats
The design of fast-import allows it to import large projects in a minimum amount of memory usage and processing time. Assuming the frontend is able to keep up with fast-import and feed it a constant stream of data, import times for projects holding 10+ years of history and containing 100,000+ individual commits are generally completed in just 1-2 hours on quite modest (~$2,000 USD) hardware.
Most bottlenecks appear to be in foreign source data access (the source just cannot extract revisions fast enough) or disk IO (fast-import writes as fast as the disk will take the data). Imports will run faster if the source data is stored on a different drive than the destination Git repository (due to less IO contention).
A typical frontend for fast-import tends to weigh in at approximately 200 lines of Perl/Python/Ruby code. Most developers have been able to create working importers in just a couple of hours, even though it is their first exposure to fast-import, and sometimes even to Git. This is an ideal situation, given that most conversion tools are throw-away (use once, and never look back).
Like git-push or git-fetch, imports handled by fast-import are safe to run alongside parallel git repack -a -d or git gc invocations, or any other Git operation (including git-prune, as loose objects are never used by fast-import).
fast-import does not lock the branch or tag refs it is actively importing. After the import, during its ref update phase, fast-import tests each existing branch ref to verify the update will be a fast-forward update (the commit stored in the ref is contained in the new history of the commit to be written). If the update is not a fast-forward update, fast-import will skip updating that ref and instead prints a warning message. fast-import will always attempt to update all branch refs, and does not stop on the first failure.
Branch updates can be forced with --force, but its recommended that this only be used on an otherwise quiet repository. Using --force is not necessary for an initial import into an empty repository.
fast-import tracks a set of branches in memory. Any branch can be created or modified at any point during the import process by sending a commit command on the input stream. This design allows a frontend program to process an unlimited number of branches simultaneously, generating commits in the order they are available from the source data. It also simplifies the frontend programs considerably.
fast-import does not use or alter the current working directory, or any file within it. (It does however update the current Git repository, as referenced by GIT_DIR.) Therefore an import frontend may use the working directory for its own purposes, such as extracting file revisions from the foreign source. This ignorance of the working directory also allows fast-import to run very quickly, as it does not need to perform any costly file update operations when switching between branches.
With the exception of raw file data (which Git does not interpret) the fast-import input format is text (ASCII) based. This text based format simplifies development and debugging of frontend programs, especially when a higher level language such as Perl, Python or Ruby is being used.
fast-import is very strict about its input. Where we say SP below we mean exactly one space. Likewise LF means one (and only one) linefeed. Supplying additional whitespace characters will cause unexpected results, such as branch names or file names with leading or trailing spaces in their name, or early termination of fast-import when it encounters unexpected input.
To aid in debugging frontends fast-import ignores any line that begins with # (ASCII pound/hash) up to and including the line ending LF. A comment line may contain any sequence of bytes that does not contain an LF and therefore may be used to include any detailed debugging information that might be specific to the frontend and useful when inspecting a fast-import data stream.
The following date formats are supported. A frontend should select the format it will use for this import by passing the format name in the --date-format=<fmt> command line option.
raw
The time of the event is specified by <time> as the number of seconds since the UNIX epoch (midnight, Jan 1, 1970, UTC) and is written as an ASCII decimal integer.
The local offset is specified by <offutc> as a positive or negative offset from UTC. For example EST (which is 5 hours behind UTC) would be expressed in <tz> by lq-0500rq while UTC is lq+0000rq. The local offset does not affect <time>; it is used only as an advisement to help formatting routines display the timestamp.
If the local offset is not available in the source material, use lq+0000rq, or the most common local offset. For example many organizations have a CVS repository which has only ever been accessed by users who are located in the same location and timezone. In this case a reasonable offset from UTC could be assumed.
Unlike the rfc2822 format, this format is very strict. Any variation in formatting will cause fast-import to reject the value.
rfc2822
An example value is lqTue Feb 6 11:22:18 2007 -0500rq. The Git parser is accurate, but a little on the lenient side. It is the same parser used by git-am when applying patches received from email.
Some malformed strings may be accepted as valid dates. In some of these cases Git will still be able to obtain the correct date from the malformed string. There are also some types of malformed strings which Git will parse wrong, and yet consider valid. Seriously malformed strings will be rejected.
Unlike the raw format above, the timezone/UTC offset information contained in an RFC 2822 date string is used to adjust the date value to UTC prior to storage. Therefore it is important that this information be as accurate as possible.
If the source material uses RFC 2822 style dates, the frontend should let fast-import handle the parsing and conversion (rather than attempting to do it itself) as the Git parser has been well tested in the wild.
Frontends should prefer the raw format if the source material already uses UNIX-epoch format, can be coaxed to give dates in that format, or its format is easily convertible to it, as there is no ambiguity in parsing.
now
This is a toy format. The current time and timezone of this system is always copied into the identity string at the time it is being created by fast-import. There is no way to specify a different time or timezone.
This particular format is supplied as its short to implement and may be useful to a process that wants to create a new commit right now, without needing to use a working directory or git-update-index.
If separate author and committer commands are used in a commit the timestamps may not match, as the system clock will be polled twice (once for each command). The only way to ensure that both author and committer identity information has the same timestamp is to omit author (thus copying from committer) or to use a date format other than now.
fast-import accepts several commands to update the current repository and control the current import process. More detailed discussion (with examples) of each command follows later.
commit
tag
reset
blob
checkpoint
progress
Create or update a branch with a new commit, recording one logical change to the project.
'commit' SP <ref> LF
mark?
('author' SP <name> SP LT <email> GT SP <when> LF)?
'committer' SP <name> SP LT <email> GT SP <when> LF
data
('from' SP <committish> LF)?
('merge' SP <committish> LF)?
(filemodify | filedelete | filecopy | filerename | filedeleteall)*
LF?
where <ref> is the name of the branch to make the commit on. Typically branch names are prefixed with refs/heads/ in Git, so importing the CVS branch symbol RELENG-1_0 would use refs/heads/RELENG-1_0 for the value of <ref>. The value of <ref> must be a valid refname in Git. As LF is not valid in a Git refname, no quoting or escaping syntax is supported here.
A mark command may optionally appear, requesting fast-import to save a reference to the newly created commit for future use by the frontend (see below for format). It is very common for frontends to mark every commit they create, thereby allowing future branch creation from any imported commit.
The data command following committer must supply the commit message (see below for data command syntax). To import an empty commit message use a 0 length data. Commit messages are free-form and are not interpreted by Git. Currently they must be encoded in UTF-8, as fast-import does not permit other encodings to be specified.
Zero or more filemodify, filedelete, filecopy, filerename and filedeleteall commands may be included to update the contents of the branch prior to creating the commit. These commands may be supplied in any order. However it is recommended that a filedeleteall command precede all filemodify, filecopy and filerename commands in the same commit, as filedeleteall wipes the branch clean (see below).
The LF after the command is optional (it used to be required).
An author command may optionally appear, if the author information might differ from the committer information. If author is omitted then fast-import will automatically use the committercqs information for the author portion of the commit. See below for a description of the fields in author, as they are identical to committer.
The committer command indicates who made this commit, and when they made it.
Here <name> is the personcqs display name (for example lqCom M Itterrq) and <email> is the personcqs email address (lqm[blue]cm@example.comm[]
The time of the change is specified by <when> using the date format that was selected by the --date-format=<fmt> command line option. See lqDate Formatsrq above for the set of supported formats, and their syntax.
The from command is used to specify the commit to initialize this branch from. This revision will be the first ancestor of the new commit.
Omitting the from command in the first commit of a new branch will cause fast-import to create that commit with no ancestor. This tends to be desired only for the initial commit of a project. If the frontend creates all files from scratch when making a new branch, a merge command may be used instead of from to start the commit with an empty tree. Omitting the from command on existing branches is usually desired, as the current commit on that branch is automatically assumed to be the first ancestor of the new commit.
As LF is not valid in a Git refname or SHA-1 expression, no quoting or escaping syntax is supported within <committish>.
Here <committish> is any of the following:
The reason fast-import uses
:
to denote a mark reference is this character is not legal in a Git branch name. The leading
:
makes it easy to distinguish between the mark 42 (:42) and the branch 42 (42
or
refs/heads/42), or an abbreviated SHA-1 which happened to consist only of base-10 digits.
Marks must be declared (via
mark) before they can be used.
The special case of restarting an incremental import from the current branch value should be written as:
author
[1]rq). LT and GT are the literal less-than (\x3c) and greater-than (\x3e) symbols. These are required to delimit the email address from the other fields in the line. Note that <name> is free-form and may contain any sequence of bytes, except LT and LF. It is typically UTF-8 encoded.
from refs/heads/branch^0
The ^0 suffix is necessary as fast-import does not permit a branch to start from itself, and the branch is created in memory before the from command is even read from the input. Adding ^0 will force fast-import to resolve the commit through Gitcqs revision parsing library, rather than its internal branch table, thereby loading in the existing value of the branch.
Includes one additional ancestor commit. If the from command is omitted when creating a new branch, the first merge commit will be the first ancestor of the current commit, and the branch will start out with no files. An unlimited number of merge commands per commit are permitted by fast-import, thereby establishing an n-way merge. However Gitcqs other tools never create commits with more than 15 additional ancestors (forming a 16-way merge). For this reason it is suggested that frontends do not use more than 15 merge commands per commit; 16, if starting a new, empty branch.
Here <committish> is any of the commit specification expressions also accepted by from (see above).
Included in a commit command to add a new file or change the content of an existing file. This command has two different means of specifying the content of the file.
External data format
merge
'M' SP <mode> SP <dataref> SP <path> LF
Here <dataref> can be either a mark reference (:<idnum>) set by a prior blob command, or a full 40-byte SHA-1 of an existing Git blob object.
Inline data format
'M' SP <mode> SP 'inline' SP <path> LF
data
See below for a detailed description of the data command.
In both formats <mode> is the type of file entry, specified in octal. Git only supports the following modes:
100644 or 644: A normal (not-executable) file. The majority of files in most projects use this mode. If in doubt, this is what you want.
100755 or 755: A normal, but executable, file.
120000: A symlink, the content of the file will be the link target.
160000: A gitlink, SHA-1 of the object refers to a commit in another repository. Git links can only be specified by SHA or through a commit mark. They are used to implement submodules.
In both formats <path> is the complete path of the file to be added (if not already existing) or modified (if already existing).
A <path> string must use UNIX-style directory separators (forward slash /), may contain any byte other than LF, and must not start with double quote (").
If an LF or double quote must be encoded into <path> shell-style quoting should be used, e.g. "path/with\n and \" in it".
The value of <path> must be in canonical form. That is it must not:
It is recommended that <path> always be encoded using UTF-8.
Included in a commit command to remove a file or recursively delete an entire directory from the branch. If the file or directory removal makes its parent directory empty, the parent directory will be automatically removed too. This cascades up the tree until the first non-empty directory or the root is reached.
filedelete
'D' SP <path> LF
here <path> is the complete path of the file or subdirectory to be removed from the branch. See filemodify above for a detailed description of <path>.
Recursively copies an existing file or subdirectory to a different location within the branch. The existing file or directory must exist. If the destination exists it will be completely replaced by the content copied from the source.
filecopy
'C' SP <path> SP <path> LF
here the first <path> is the source location and the second <path> is the destination. See filemodify above for a detailed description of what <path> may look like. To use a source path that contains SP the path must be quoted.
A filecopy command takes effect immediately. Once the source location has been copied to the destination any future commands applied to the source location will not impact the destination of the copy.
Renames an existing file or subdirectory to a different location within the branch. The existing file or directory must exist. If the destination exists it will be replaced by the source directory.
filerename
'R' SP <path> SP <path> LF
here the first <path> is the source location and the second <path> is the destination. See filemodify above for a detailed description of what <path> may look like. To use a source path that contains SP the path must be quoted.
A filerename command takes effect immediately. Once the source location has been renamed to the destination any future commands applied to the source location will create new files there and not impact the destination of the rename.
Note that a filerename is the same as a filecopy followed by a filedelete of the source location. There is a slight performance advantage to using filerename, but the advantage is so small that it is never worth trying to convert a delete/add pair in source material into a rename for fast-import. This filerename command is provided just to simplify frontends that already have rename information and doncqt want bother with decomposing it into a filecopy followed by a filedelete.
Included in a commit command to remove all files (and also all directories) from the branch. This command resets the internal branch structure to have no files in it, allowing the frontend to subsequently add all interesting files from scratch.
filedeleteall
'deleteall' LF
This command is extremely useful if the frontend does not know (or does not care to know) what files are currently on the branch, and therefore cannot generate the proper filedelete commands to update the content.
Issuing a filedeleteall followed by the needed filemodify commands to set the correct content will produce the same results as sending only the needed filemodify and filedelete commands. The filedeleteall approach may however require fast-import to use slightly more memory per active branch (less than 1 MiB for even most large projects); so frontends that can easily obtain only the affected paths for a commit are encouraged to do so.
Arranges for fast-import to save a reference to the current object, allowing the frontend to recall this object at a future point in time, without knowing its SHA-1. Here the current object is the object creation command the mark command appears within. This can be commit, tag, and blob, but commit is the most common usage.
'mark' SP ':' <idnum> LF
where <idnum> is the number assigned by the frontend to this mark. The value of <idnum> is expressed as an ASCII decimal integer. The value 0 is reserved and cannot be used as a mark. Only values greater than or equal to 1 may be used as marks.
New marks are created automatically. Existing marks can be moved to another object simply by reusing the same <idnum> in another mark command.
Creates an annotated tag referring to a specific commit. To create lightweight (non-annotated) tags see the reset command below.
'tag' SP <name> LF
'from' SP <committish> LF
'tagger' SP <name> SP LT <email> GT SP <when> LF
data
where <name> is the name of the tag to create.
Tag names are automatically prefixed with refs/tags/ when stored in Git, so importing the CVS branch symbol RELENG-1_0-FINAL would use just RELENG-1_0-FINAL for <name>, and fast-import will write the corresponding ref as refs/tags/RELENG-1_0-FINAL.
The value of <name> must be a valid refname in Git and therefore may contain forward slashes. As LF is not valid in a Git refname, no quoting or escaping syntax is supported here.
The from command is the same as in the commit command; see above for details.
The tagger command uses the same format as committer within commit; again see above for details.
The data command following tagger must supply the annotated tag message (see below for data command syntax). To import an empty tag message use a 0 length data. Tag messages are free-form and are not interpreted by Git. Currently they must be encoded in UTF-8, as fast-import does not permit other encodings to be specified.
Signing annotated tags during import from within fast-import is not supported. Trying to include your own PGP/GPG signature is not recommended, as the frontend does not (easily) have access to the complete set of bytes which normally goes into such a signature. If signing is required, create lightweight tags from within fast-import with reset, then create the annotated versions of those tags offline with the standard git-tag process.
Creates (or recreates) the named branch, optionally starting from a specific revision. The reset command allows a frontend to issue a new from command for an existing branch, or to create a new branch from an existing commit without creating a new commit.
'reset' SP <ref> LF
('from' SP <committish> LF)?
LF?
For a detailed description of <ref> and <committish> see above under commit and from.
The LF after the command is optional (it used to be required).
The reset command can also be used to create lightweight (non-annotated) tags. For example:
reset refs/tags/938 from :938
would create the lightweight tag refs/tags/938 referring to whatever commit mark :938 references.
Requests writing one file revision to the packfile. The revision is not connected to any commit; this connection must be formed in a subsequent commit command by referencing the blob through an assigned mark.
'blob' LF
mark?
data
The mark command is optional here as some frontends have chosen to generate the Git SHA-1 for the blob on their own, and feed that directly to commit. This is typically more work than its worth however, as marks are inexpensive to store and easy to use.
Supplies raw data (for use as blob/file content, commit messages, or annotated tag messages) to fast-import. Data can be supplied using an exact byte count or delimited with a terminating line. Real frontends intended for production-quality conversions should always use the exact byte count format, as it is more robust and performs better. The delimited format is intended primarily for testing fast-import.
Comment lines appearing within the <raw> part of data commands are always taken to be part of the body of the data and are therefore never ignored by fast-import. This makes it safe to import any file/message content whose lines might start with #.
Exact byte count format
'data' SP <count> LF
<raw> LF?
where <count> is the exact number of bytes appearing within <raw>. The value of <count> is expressed as an ASCII decimal integer. The LF on either side of <raw> is not included in <count> and will not be included in the imported data.
The LF after <raw> is optional (it used to be required) but recommended. Always including it makes debugging a fast-import stream easier as the next command always starts in column 0 of the next line, even if <raw> did not end with an LF.
Delimited format
'data' SP '<<' <delim> LF
<raw> LF
<delim> LF
LF?
where <delim> is the chosen delimiter string. The string <delim> must not appear on a line by itself within <raw>, as otherwise fast-import will think the data ends earlier than it really does. The LF immediately trailing <raw> is part of <raw>. This is one of the limitations of the delimited format, it is impossible to supply a data chunk which does not have an LF as its last byte.
The LF after <delim> LF is optional (it used to be required).
Forces fast-import to close the current packfile, start a new one, and to save out all current branch refs, tags and marks.
'checkpoint' LF
LF?
Note that fast-import automatically switches packfiles when the current packfile reaches --max-pack-size, or 4 GiB, whichever limit is smaller. During an automatic packfile switch fast-import does not update the branch refs, tags or marks.
As a checkpoint can require a significant amount of CPU time and disk IO (to compute the overall pack SHA-1 checksum, generate the corresponding index file, and update the refs) it can easily take several minutes for a single checkpoint command to complete.
Frontends may choose to issue checkpoints during extremely large and long running imports, or when they need to allow another Git process access to a branch. However given that a 30 GiB Subversion repository can be loaded into Git through fast-import in about 3 hours, explicit checkpointing may not be necessary.
The LF after the command is optional (it used to be required).
Causes fast-import to print the entire progress line unmodified to its standard output channel (file descriptor 1) when the command is processed from the input stream. The command otherwise has no impact on the current import, or on any of fast-importcqs internal state.
'progress' SP <any> LF
LF?
The <any> part of the command may contain any sequence of bytes that does not contain LF. The LF after the command is optional. Callers may wish to process the output through a tool such as sed to remove the leading part of the line, for example:
frontend | git fast-import | sed 's/^progress //'
Placing a progress command immediately after a checkpoint will inform the reader when the checkpoint has been completed and it can safely access the refs that fast-import updated.
If fast-import is supplied invalid input it will terminate with a non-zero exit status and create a crash report in the top level of the Git repository it was importing into. Crash reports contain a snapshot of the internal fast-import state as well as the most recent commands that lead up to the crash.
All recent commands (including stream comments, file changes and progress commands) are shown in the command history within the crash report, but raw file data and commit messages are excluded from the crash report. This exclusion saves space within the report file and reduces the amount of buffering that fast-import must perform during execution.
After writing a crash report fast-import will close the current packfile and export the marks table. This allows the frontend developer to inspect the repository state and resume the import from the point where it crashed. The modified branches and tags are not updated during a crash, as the import did not complete successfully. Branch and tag information can be found in the crash report and must be applied manually if the update is needed.
An example crash:
$ cat >in <<END_OF_INPUT # my very first test commit commit refs/heads/master committer Shawn O. Pearce <spearce> 19283 -0400 # who is that guy anyway? data <<EOF this is my commit EOF M 644 inline .gitignore data <<EOF .gitignore EOF M 777 inline bob END_OF_INPUT
$ git fast-import <in fatal: Corrupt mode: M 777 inline bob fast-import: dumping crash report to .git/fast_import_crash_8434
$ cat .git/fast_import_crash_8434
fast-import crash report:
fast-import process: 8434
parent process : 1391
at Sat Sep 1 00:58:12 2007
fatal: Corrupt mode: M 777 inline bob
Most Recent Commands Before Crash --------------------------------- # my very first test commit commit refs/heads/master committer Shawn O. Pearce <spearce> 19283 -0400 # who is that guy anyway? data <<EOF M 644 inline .gitignore data <<EOF * M 777 inline bob
Active Branch LRU
-----------------
active_branches = 1 cur, 5 max
pos clock name ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 1) 0 refs/heads/master
Inactive Branches ----------------- refs/heads/master: status : active loaded dirty tip commit : 0000000000000000000000000000000000000000 old tree : 0000000000000000000000000000000000000000 cur tree : 0000000000000000000000000000000000000000 commit clock: 0 last pack :
------------------- END OF CRASH REPORT
The following tips and tricks have been collected from various users of fast-import, and are offered here as suggestions.
When doing a repository conversion, use a unique mark per commit (mark :<n>) and supply the --export-marks option on the command line. fast-import will dump a file which lists every mark and the Git object SHA-1 that corresponds to it. If the frontend can tie the marks back to the source repository, it is easy to verify the accuracy and completeness of the import by comparing each Git commit to the corresponding source revision.
Coming from a system such as Perforce or Subversion this should be quite simple, as the fast-import mark can also be the Perforce changeset number or the Subversion revision number.
Doncqt bother trying to optimize the frontend to stick to one branch at a time during an import. Although doing so might be slightly faster for fast-import, it tends to increase the complexity of the frontend code considerably.
The branch LRU builtin to fast-import tends to behave very well, and the cost of activating an inactive branch is so low that bouncing around between branches has virtually no impact on import performance.
When importing a renamed file or directory, simply delete the old name(s) and modify the new name(s) during the corresponding commit. Git performs rename detection after-the-fact, rather than explicitly during a commit.
Some other SCM systems let the user create a tag from multiple files which are not from the same commit/changeset. Or to create tags which are a subset of the files available in the repository.
Importing these tags as-is in Git is impossible without making at least one commit which lqfixes uprq the files to match the content of the tag. Use fast-importcqs reset command to reset a dummy branch outside of your normal branch space to the base commit for the tag, then commit one or more file fixup commits, and finally tag the dummy branch.
For example since all normal branches are stored under refs/heads/ name the tag fixup branch TAG_FIXUP. This way it is impossible for the fixup branch used by the importer to have namespace conflicts with real branches imported from the source (the name TAG_FIXUP is not refs/heads/TAG_FIXUP).
When committing fixups, consider using merge to connect the commit(s) which are supplying file revisions to the fixup branch. Doing so will allow tools such as git-blame to track through the real commit history and properly annotate the source files.
After fast-import terminates the frontend will need to do rm .git/TAG_FIXUP to remove the dummy branch.
As soon as fast-import completes the Git repository is completely valid and ready for use. Typically this takes only a very short time, even for considerably large projects (100,000+ commits).
However repacking the repository is necessary to improve data locality and access performance. It can also take hours on extremely large projects (especially if -f and a large --window parameter is used). Since repacking is safe to run alongside readers and writers, run the repack in the background and let it finish when it finishes. There is no reason to wait to explore your new Git project!
If you choose to wait for the repack, doncqt try to run benchmarks or performance tests until repacking is completed. fast-import outputs suboptimal packfiles that are simply never seen in real use situations.
If you are repacking very old imported data (e.g. older than the last year), consider expending some extra CPU time and supplying --window=50 (or higher) when you run git-repack. This will take longer, but will also produce a smaller packfile. You only need to expend the effort once, and everyone using your project will benefit from the smaller repository.
Every once in a while have your frontend emit a progress message to fast-import. The contents of the messages are entirely free-form, so one suggestion would be to output the current month and year each time the current commit date moves into the next month. Your users will feel better knowing how much of the data stream has been processed.
When packing a blob fast-import always attempts to deltify against the last blob written. Unless specifically arranged for by the frontend, this will probably not be a prior version of the same file, so the generated delta will not be the smallest possible. The resulting packfile will be compressed, but will not be optimal.
Frontends which have efficient access to all revisions of a single file (for example reading an RCS/CVS ,v file) can choose to supply all revisions of that file as a sequence of consecutive blob commands. This allows fast-import to deltify the different file revisions against each other, saving space in the final packfile. Marks can be used to later identify individual file revisions during a sequence of commit commands.
The packfile(s) created by fast-import do not encourage good disk access patterns. This is caused by fast-import writing the data in the order it is received on standard input, while Git typically organizes data within packfiles to make the most recent (current tip) data appear before historical data. Git also clusters commits together, speeding up revision traversal through better cache locality.
For this reason it is strongly recommended that users repack the repository with git repack -a -d after fast-import completes, allowing Git to reorganize the packfiles for faster data access. If blob deltas are suboptimal (see above) then also adding the -f option to force recomputation of all deltas can significantly reduce the final packfile size (30-50% smaller can be quite typical).
There are a number of factors which affect how much memory fast-import requires to perform an import. Like critical sections of core Git, fast-import uses its own memory allocators to amortize any overheads associated with malloc. In practice fast-import tends to amortize any malloc overheads to 0, due to its use of large block allocations.
fast-import maintains an in-memory structure for every object written in this execution. On a 32 bit system the structure is 32 bytes, on a 64 bit system the structure is 40 bytes (due to the larger pointer sizes). Objects in the table are not deallocated until fast-import terminates. Importing 2 million objects on a 32 bit system will require approximately 64 MiB of memory.
The object table is actually a hashtable keyed on the object name (the unique SHA-1). This storage configuration allows fast-import to reuse an existing or already written object and avoid writing duplicates to the output packfile. Duplicate blobs are surprisingly common in an import, typically due to branch merges in the source.
Marks are stored in a sparse array, using 1 pointer (4 bytes or 8 bytes, depending on pointer size) per mark. Although the array is sparse, frontends are still strongly encouraged to use marks between 1 and n, where n is the total number of marks required for this import.
Branches are classified as active and inactive. The memory usage of the two classes is significantly different.
Inactive branches are stored in a structure which uses 96 or 120 bytes (32 bit or 64 bit systems, respectively), plus the length of the branch name (typically under 200 bytes), per branch. fast-import will easily handle as many as 10,000 inactive branches in under 2 MiB of memory.
Active branches have the same overhead as inactive branches, but also contain copies of every tree that has been recently modified on that branch. If subtree include has not been modified since the branch became active, its contents will not be loaded into memory, but if subtree src has been modified by a commit since the branch became active, then its contents will be loaded in memory.
As active branches store metadata about the files contained on that branch, their in-memory storage size can grow to a considerable size (see below).
fast-import automatically moves active branches to inactive status based on a simple least-recently-used algorithm. The LRU chain is updated on each commit command. The maximum number of active branches can be increased or decreased on the command line with --active-branches=.
Trees (aka directories) use just 12 bytes of memory on top of the memory required for their entries (see lqper active filerq below). The cost of a tree is virtually 0, as its overhead amortizes out over the individual file entries.
Files (and pointers to subtrees) within active trees require 52 or 64 bytes (32/64 bit platforms) per entry. To conserve space, file and tree names are pooled in a common string table, allowing the filename lqMakefilerq to use just 16 bytes (after including the string header overhead) no matter how many times it occurs within the project.
The active branch LRU, when coupled with the filename string pool and lazy loading of subtrees, allows fast-import to efficiently import projects with 2,000+ branches and 45,114+ files in a very limited memory footprint (less than 2.7 MiB per active branch).
Written by Shawn O. Pearce <m[blue]spearce@spearce.orgm[][2]>.
Documentation by Shawn O. Pearce <m[blue]spearce@spearce.orgm[][2]>.
Part of the git(1) suite