Protocol Documentation

repository is the repository that generated files are being checked for.

blobs is a set of blobs that the generated file check is being performed on.

revision is a Git revision that resolves to a blob.

path is the file path of the blob and is used to gain insight as to whether the blob is generated.

blobs is a set of blobs that the generated file check has been performed on.

revision is the Git revision of the checked blob provided in the request.

generated is the result of the file generation check for a particular blob.

repository is the repository that shall be searched.

oid is the object ID of the blob we want to get.

limit is the maximum number of bytes we want to receive. Use '-1' to get the full blob no matter how big. Setting this to `0` will return no data.

size is the size of the blob. Present only in first response message.

data is a chunk of data.

oid of the actual blob returned. Empty if no blob was found.

repository is the repository that shall be searched.

revision_paths identifies the set of revision/path pairs that shall be searched for blobs.

limit is the maximum number of bytes we want to receive. Use '-1' to get the full blobs no matter how big.

revision is the revision that identifies the tree-ish. Must not be empty.

path is the path relative to the treeish revision that shall be searched for a blob. If the path is empty the root directory of the tree-ish will be searched.

size is the size of the blob. Present only on the first message per blob

data is a chunk of blob data, which could span over multiple messages.

oid is the object ID of the current blob. Only present on the first message per blob. Empty if no blob was found.

is_submodule indicates whether the blob is a submodule.

mode is the file mode of blob as present in the tree. It is typically one of: - 0o100644 for non-executable files. - 0o100755 for executable files. - 0o160000 for submodules. - 0o040000 for subtrees.

revision is the revision that this blob has been traversed from.

path is the path of the blob inside of the tree.

type is the type of the "blob".

repository is the repository for which LFS pointers should be retrieved from.

blob_ids is the list of blobs to retrieve LFS pointers from. Must be a non-empty list of blobs IDs to fetch.

lfs_pointers is the list of LFS pointers which were requested.

size is the size of the blob. This is not the size of the LFS object pointed to.

data is the bare data of the LFS pointer blob. It contains the pointer to the LFS data in the format specified by the LFS project.

oid is the object ID of the blob.

file_size is the size given when parsing the LFS pointer spec.

file_oid is the object id given when parsing the LFS pointer spec.

repository is the repository in which blobs should be enumerated.

limit is the maximum number of blobs to return. If set to its default (`0`), then all found blobs will be returned.

bytes_limit is the maximum number of bytes to receive for each blob. If set to `0`, then no blob data will be sent. If `-1`, then all blob data will be sent without any limits.

blobs is the blobs which have been found. In case blob contents were requested and contents of a blob exceed the maximum gRPC message size, then this blob will be split up into multiple blob messages which span across multiple responses. In that case, metadata of the blob will only be sent on the first such message for this specific blob.

oid is the object ID of the blob. Will only be set for the first message of each specific blob.

size is the size of the blob. Will only be set for the first message of each specific blob.

data is the contents of the blob. This field is optional and depends on the BytesLimit in the original request.

repository is the repository for which LFS pointers should be retrieved from.

limit limits the number of LFS pointers returned.

lfs_pointers is the list of LFS pointers which were requested.

repository is the repository in which blobs should be enumerated.

revisions is the list of revisions to retrieve blobs from. These revisions will be walked. Supports pseudo-revisions `--all` and `--not` as well as negated revisions via `^revision`. Revisions cannot start with a leading dash. Please consult gitrevisions(7) for more info. Must not be empty.

limit is the maximum number of blobs to return. If set to its default (`0`), then all found blobs will be returned.

bytes_limit is the maximum number of bytes to receive for each blob. If set to `0`, then no blob data will be sent. If `-1`, then all blob data will be sent without any limits.

with_paths determines whether paths of blobs should be returned. When set to `true`, paths are returned on a best-effort basis: a path will only exist if the blob was traversed via a tree.

blobs is the blobs which have been found. In case blob contents were requested and contents of a blob exceed the maximum gRPC message size, then this blob will be split up into multiple blob messages which span across multiple responses. In that case, metadata of the blob will only be sent on the first such message for this specific blob.

oid is the object ID of the blob. Will only be set for the first message of each specific blob.

size is the size of the blob. Will only be set for the first message of each specific blob.

data is the contents of the blob. This field is optional and depends on the BytesLimit in the original request.

path is the path of the blob. May be unset depending on how the blob had been traversed.

repository is the repository for which LFS pointers should be retrieved from.

revisions is the list of revisions to retrieve LFS pointers from. Must be a non-empty list.

limit limits the number of LFS pointers returned.

lfs_pointers is the list of LFS pointers which were requested.

repository ... Only available on the first message

object_map is a raw object-map file as generated by BFG: https://rtyley.github.io/bfg-repo-cleaner Each line in the file has two object SHAs, space-separated - the original SHA of the object, and the SHA after BFG has rewritten the object.

entries ...

type ...

old_oid ...

new_oid ...

repository is the repository that shall be rewritten. Must be sent on the first request only.

blobs is the list of blob object ids that will be removed from history.

redactions is the list of literals or patterns that will be replaced with "***REMOVED***". Items cannot contain newline characters. See https://htmlpreview.github.io/?https://github.com/newren/git-filter-repo/blob/docs/html/git-filter-repo.html for a full explanation of what patterns are supported.

authority_name represents the storage that created the partition.

partition_id is the local incrementing ID of a specific partition within a storage. Together with `authority_name`, this forms a unique identifier for a partition across the cluster. A partition belongs to a Raft group.

id is the unique identifier for the Raft message. This ID is generated by an in-memory event registry. Raft uses this ID to track the commit status of a log entry.

data represents packed and serialized log data.

local_path is the path to the staging log entry directory. Before the request is sent to other nodes, this whole directory of the entry is serialized. So, this field is only relevant to the primary node who issues the request.

packed is the serialized form of the log entry data. Transport implementations populates this field before sending out messages to other members of a Raft group.

cluster_id is the identifier of the Raft cluster to which this message belongs.

replica_id uniquely identifies a replica in the Raft cluster.

message is the Raft message to be delivered.

raft_msg is a wrapper for raftpb.Message

chunk of snapshot data

destination is where the snapshot is saved.

snapshot_size is the snapshot size in bytes.

partition_key identifies which partition this replica belongs to.

member_id is the unique identifier assigned by etcd/raft.

storage_name is the name of the storage where this replica is hosted.

metadata contains replica information.

type indicates whether the replica is learner or voter.

address is the network address of the replica.

repository is the repository in which existence of objects and refs are checked. Only the initial request must contain a repository, the repository of all subsequent requests will be ignored.

revisions contains the revisions that shall be checked for existence. This accepts all revisions as documented in gitrevisions(7).

revisions is the list of RevisionExistence objects used to indicate which revisions exist.

name refers to the revision.

exists is true if the revision exists in the repository.

repository is the repository for which we need to check the ancestry.

ancestor_id is the object ID of the commit which needs to be checked as ancestor.

child_id is the object ID of the commit whose ancestor needs to be confirmed.

value denotes whether the provided commit is the ancestor or not.

repository is the repository where to detect the languages in.

revision tells for which commitish the languages should be detected. If it's omitted, the HEAD commit of the default branch is used.

languages is a set of all the detected languages and their statistics.

name is the name of the detected language, for example: Ruby, Go, HTML A full list of language names can be found at: https://github.com/github/linguist/blob/master/lib/linguist/languages.yml

share is the percentual share (value between 0 and 100) of this language in relation to other languages that exist in the given revision.

color specifies the associated color for this language, for example #3fd5e0.

bytes is the total amount of bytes written in this language

repository is the repository in which the commit is present.

revision is a commitish for which we want to get the stats.

oid is the commit object ID. It is empty if the commit is not found.

additions refers to the number of lines added in the commit.

deletions refers to the number of lines deleted in the commit.

files refers to the number of files changed in the commit.

repository is the repository to run the query in.

revision is the commitish to retrieve commits against.

offset is the offset from which to list commits.

limit is the maximum number of commits to list.

path is the relative path to filter the commits to.

query is the search param to search commit messages against.

global_options contains the global options used to modify the behaviour of Git.

commits is the list of commits which match the query from the request.

repository is the repository in which we want to count the number of commits.

revision is a commitish which is the start point for the traversal of commits.

after is used to filter commits more recent than a specific date.

before is used to filter commits older than a specific date.

path is used to filter commits which modify the provided path.

max_count is used to cap the number of commits.

all is used to consider all refs (including HEAD) as the start point for the traversal. all and Revision options are mutually exclusive.

first_parent ensures that only the first parent commit is followed in the traversal.

global_options contains the global options used to modify the behaviour of Git.

count denotes the number of commits found as per the given filters.

repository is the repository in which we want to find the number of diverging commits.

from is the object ID of one of the commits against which we want to check the number of diverging commits. The From and To fields are interchangeable.

to is the object ID of one of the commits against which we want to check the number of diverging commits. The To and From fields are interchangeable.

max_count denotes the cap for the number of diverging commits to be reported.

left_count denotes the number of diverging commits present in the 'From' commit provided.

right_count denotes the number of diverging commits present in the 'To' commit provided.

repository is the repository to retrieve the commits from.

shas is a set of commitish object IDs. If there is no corresponding commit for a provided sha, no error will be thrown. Only signed commits will be returned in the response.

shas is the filtered list of shas from the request which have an associated signed commit.

repository is the repository from which we will list the commits.

revision is the commitish which is the starting point for the traversal. When nil, return all commits reachable by any branch in the repo.

max_count is the maximum number of commits to list.

skip is the number of commits to skip before starting to show the commit output.

order is the order in which commits should be traversed. Refer to the enum's documentation for further information.

commits is a list of commits produced as per the request.

repository is the repository in which we want to find the commit.

revision is a commitish which is to be resolved to a commit.

trailers if set, parses and adds the trailing information of the commit.

commit is the requested commit, it is nil when the commit was not found.

bad_object is returned when the revision cannot be found in the git repository.

ambiguous_ref is returned when the given revision cannot be found.

invalid_range is returned when the provided revision-range cannot be found in the git repository.

repository is the repository from which the commits are listed.

revision is a commitish against which the commits for the provided paths are listed. If there is no revision provided, the default branch is considered. It is mutually exclusive with the `All` option.

limit is maximum number of commits to list.

offset is the number of commits to skip before starting the traversal.

paths is a list of non-empty paths for which we want to list commits.

follow is used to list the history of a file beyond renames (works only for a single file).

skip_merges is used to skip merge commits.

Deprecated. disable_walk is no longer supported.

after will only list commits which are more recent than the specified date.

before will only list commits which are older than the specified date.

all is used to consider all refs (including HEAD) as the start point for the traversal. `all` and `revision` options are mutually exclusive.

first_parent ensures that only the first parent commit is followed in the traversal.

author will only list commits whose author matches the given pattern, which is a regular expression.

order is the order in which commits should be traversed. Please refer to the enum's documentation for further information.

global_options contains the global options used to modify the behaviour of Git.

trailers if set, parses and adds the trailing information of the commit.

include_shortstat determines whether to include the number of lines and files changed in the commits. Populates the `ShortStats` field.

include_referenced_by determines whether to include references that point to a commit in the response. The field contains patterns that are matched against references. Only matching references are included in the response. For example, one can get tags pointing to a commit with `refs/tags`, or branches respectively with `refs/heads`. By sending both, the response will include all references that match either pattern. protolint:disable:this REPEATED_FIELD_NAMES_PLURALIZED

message_regex is used to select only commits with messages matching the provided pattern. The pattern is a case-insensitive basic Posix regular expression.

commits is a list of commits produced as per the request.

repository is the repository to retrieve the commits from.

commit_ids is the list of commitish object IDs for whom we want to retrieve the commit message.

commit_id is the commit object ID. A response with commit_id set, will be followed with response with the message field set with the commit message for that commit.

message is the commit message for the commit.

repository is the repository to retrieve the commits from.

commit_ids is the list of commitish object IDs for whom we want to retrieve signature information. If a commit doesn't have associated signature information it will be omitted from the results.

commit_id of the signature.

signature of the commit (GPG or SSH).

signed_text is text that is used to verify the signature.

signer of the commit

author provides the original commit author information without 'mailmap'.

resolve_tree is set when the provided revision ID could not be resolved.

path is set when the provided path is not valid.

repository is the repository to get the tree entries from.

revision is the commitish at which the tree entries is to be read.

path is the path of the entry that shall be read, relative to the tree of the specified revision.

recursive denotes wether to recursively fetch sub-trees.

sort defines the sorting parameter.

pagination_params controls paging. Refer to PaginationParameter documentation for further info.

skip_flat_paths is an option to skip the expensive operation of populating flat paths.

entries denotes a list of tree entries.

pagination_cursor defines the offset for subsequent requests. Refer to PaginationCursor documentation for further info.

repository is the repository to run the query in.

revision is the commitish that is used as the start commit to perform the search.

path is the path for which the last commit should be searched. This path can either point to a blob or to a tree. The path must be relative and must not escape the repository root. If the path is empty or "/", then the repository root will be searched instead.

literal_pathspec will treat the path literally. No globbing or pathspec magic is performed. This option is deprecated in favor of GlobalOptions.

global_options contains the global options used to modify the behaviour of Git.

commit is the commit that has last modified the given path. Unset in case the path could not be found.

repository is the repository in which commits should be searched for.

pagination_params controls paging. Refer to PaginationParameter documentation for further info.

commits is the list of commits found.

repository is the repository to list commits from.

oid is a set of commitish object IDs to list commits for. If there is no commit against a provided object ID, no error is thrown. It is simply ignored. protolint:disable:this REPEATED_FIELD_NAMES_PLURALIZED

commits are the list of commits for the provided object IDs from the request.

repository is the repository to list commits from.

ref_names is a set of references to obtain commits for. If there is no commit against a provided reference, no error is thrown. It is simply ignored.

commit_refs is a list of CommitForRef objects which provides the commits against the requested references.

commit is the commit object against the reference.

ref_name is a reference from the repository.

repository is the repository in which commits should be searched for.

revisions is the set of revisions which should be walked to enumerate commits. Accepts all notation as documented in gitrevisions(7) as well as the pseudo-revisions `--not` and `--all` as documented in git-rev-list(1). Must not be empty.

pagination_params controls paging. Refer to PaginationParameter documentation for further info.

order is the order in which commits should be traversed. Please refer to the enum's documentation for further information.

reverse will cause all commits to be listed in reverse.

max_parents will skip all commits which have more than the specified number of parents. If set to `0`, no filtering by parents will happen. If set to `1`, all merge commits will be omitted.

disable_walk will disable walking the graph. As a result, only commits which are immediately referenced by Revisions will be returned.

first_parent will cause the graph walk to only go down the first-parent chain of commits. Merge commits will thus only cause the mainline to be enumerated.

after will only list commits which are more recent than the specified date.

before will only list commits which are older than the specified date.

author will only list commits whose author matches the given pattern, which is a regular expression.

ignore_case will apply case-sensitive behaviour while regex matching.

commit_message_patterns will only list commits whose commit message matches any of the given patterns.

skip will skip a number commits before starting to show the commit output.

commits is the list of commits found.

repository is the repository to list the files from.

revision is a commitish against which we list all files.

paths is the list of paths found for the requested revision. Clients must remember to force utf-8 encoding.

repository is the repository to run the query in.

revision is the treeish to retrieve commits against.

path is the relative path to further filter against the treeish revision. This path can either point to a blob or to a tree. The path must be relative and must not escape the repository root. If the path is empty or "/", then it will be replaced with the root of of the repository.

limit is the number of tree entries to limit the response to.

offset is the offset of the tree entries from which to start the response.

Deprecated. literal_pathspec is deprecated.

global_options contains the global options used to modify the behaviour of Git.

commits is a set of CommitForTree objects.

commit is the commit that has last modified the path.

path_bytes is the path associated with the commit.

path_not_found is returned when the blamed path cannot be found in the revision.

out_of_range is returned when the specified blamed range exceeds the file length.

resolve_ignore_revs indicates that resolving the specified file containing ignored revisions has failed.

invalid_ignore_revs_format indicates that the specified file containing ignored revisions has an invalid format.

content is the content of the line that is invalid. May be truncated.

actual_lines contains the actual number of lines that can be blamed in the file.

ignore_revisions_blob is provided revision that does not refer to a blob

repository is the repository to perform the blame on.

revision is the commitish at which to start the blame.

path is the path of the blob that should be blamed.

range is the comma-separated range of line numbers to perform the blame on, e.g. "1,1000". Optional - if no range is provided, the whole file will be blamed.

ignore_revisions_blob contains a revision like `refs/heads/master:.git-ignore-revs-file` that resolves to a blob. If set, the file will be used as input to git-blame(1) to ignore certain commits during blame. Please refer to the git-blame(1), more specifically the `--ignore-revs-file` parameter, for more information. If unset, no commits will be ignored. If set, it must resolve to a valid blob or otherwise the RPC call will fail.

data is the raw data as generated by git-blame(1).

oid of the object this tree entry points to.

path is the path of the entry relative to the tree of the specified revision.

type denotes the type of the tree entry.

mode is the mode of the tree entry.

commit_oid is the commit object via which this entry was retrieved.

flat_path is the relative path of the first subdir that doesn't have only one directory descendant.

repository is the repository for which to read the tree entry.

revision is the commitish at which the tree entry is to be read.

path is the path of the entry that shall be read, relative to the tree of the specified revision.

limit is the maximum number of bytes to fetch. If object is bigger, remaining bytes are not sent 0 means there is no limit.

max_size is the maximum allowed object size. If bigger, a FailedPrecondition error is returned 0 means there is no maximum size.

type is the object type of the tree entry.

oid is the object ID of the tree entry. In case of submodules, it contains the commit ID that the submodule currently refers to.

size is the size of the tree entry.

mode is the mode of the tree entry.

data contains the raw object contents. This data may be split up across multiple messages.

header contains the header information about a conflicted file.

content is the content data from the conflicted file. The file data may be broken into chunks and sent in the stream following a file header. if skip_content is set in the request, conflicted file content is not streamed.

commit_oid is the tree-ish revision being merged into.

their_path is the path of the conflicted file being merged from.

our_path is the path of the conflicted file being merged into.

our_mode is the mode of the conflicted file being merged into.

ancestor_path is path of the conflicted file for the merge base.

repository is the repository for which we want to list the conflicted files.

our_commit_oid is the tree-ish OID to merge into.

their_commit_oid is the tree-ish OID to merge from.

allow_tree_conflicts will not cause the request to fail in case there are tree conflicts. If set to true, then responses may contain conflict files where some of the paths are unset.

skip_content will skip the parsing and streaming of conflicted file's content. This can be useful when we only want to know if there is a conflict and which files are conflicted but don't care about the contents of the conflicted files.

files is a list of conflicted file information being sent in the stream. This information contains conflicted file headers and optionally the conflicted file content.

header is the initial message specifying parameters of the RPC call.

files_json is a JSON-encoded list of conflicts resolutions.

repository is the repository in which conflicts shall be resolved and where SourceBranch shall be updated with the resolved conflict.

our_commit_oid is the OID of the commit representing the local commit.

target_repository is the repository from which TheirCommitOid shall be retrieved.

their_commit_oid is the OID of the commit representing the remote commit which is to be merged into the local commit.

source_branch is the branch on which the new commit shall be created.

target_branch identifies the branch which will be fetched from target_repository in case TheirCommitOid does not exist in Repository.

commit_message is the message of the newly created merge commit.

user is the user used as author and committer of the newly created merge commit.

timestamp is the optional timestamp to use for the commit as committer date. If it's not set, the current time will be used.

resolution_error contains a description of why conflict resolution has failed.

base is the merge base revision.

rev1 is the first revision.

rev2 is the the second revision.

path is the path of the change in question.

status is the type of change.

old_mode is the mode of the changed path previous to the change. May be one of the following values: - 0o000000 if the path does not exist. - 0o100644 if the path refers to a normal file. - 0o100755 if the path refers to an executable file. - 0o040000 if the path refers to a tree entry. - 0o160000 if the path refers to a submodule.

new_mode is the mode of the changed path after the change. Please refer to `old_mode` for a list of potential values.

old_blob_id is the blob oid of the changed path previous to the change.

new_blob_id is the blob oid of the changed path after the changed.

old_path is the pre-image path of a renamed or copied object. This field is empty for all other change types.

score denotes the similarity between the source and target. Only present for renames and copies.

from_path is the path that to_path is getting compared to.

to_path is the path that from_path is getting compared against.

from_id is the id of the blob that is getting compared to.

to_id is the id of the blob that is getting compared against.

old_mode is the mode of the blob getting compared against.

new_mode is the mode of the blob getting compared to.

repository is the one to get the commit deltas from.

left_commit_id is the left commit ID in <left commit>..<right commit>.

right_commit_id is the right commit ID in <left commit>..<right commit>.

paths is a list of paths that limits the diff to those specific paths.

deltas is a list of CommitDeltas

repository is the one from which to get the diff.

left_commit_id is the left commit ID in <left commit>..<right commit>.

right_commit_id is the right commit ID in <left commit>..<right commit>.

paths is a list of paths that limits the diff to those specific paths.

collapse_diffs causes patches to be emptied after safe_max_files, safe_max_files, or safe_max_lines is reached.

enforce_limits causes parsing of diffs to stop if max_files, max_lines, or max_bytes is reached.

max_files is the maximum number of files in a diff. Once reached, parsing is stopped if enforce_limits is true.

max_lines is the maximum number of lines in a diff. Once reached, parsing is stopped if enforce_limits is true.

max_bytes is the maximum number of bytes in a diff. Once reached, parsing is stopped if enforce_limits is true.

max_patch_bytes is the limitation of a single diff patch, patches surpassing this limit are pruned by default. If this is 0 you will get back empty patches.

safe_max_files is the maximum number of files in a diff. Once reached, patches are emptied if collapse_diffs is true.

safe_max_lines is the maximum number of lines in a diff. Once reached, patches are emptied if collapse_diffs is true.

safe_max_bytes is the maximum number of bytes in a diff. Once reached, patches are emptied if collapse_diffs is true.

diff_mode is the mode used for generating the diff. Please refer to the enum declaration for supported modes.

max_patch_bytes_for_file_extension is a map of file extension to max patch bytes for that extension. It overrides max patch bytes for file extension. If the file does not have an extension (eg Dockerfile), then the file name is used for matching. For files with more than on extension (eg file.html.tmpl) only the last extension is matched.

whitespace_changes states which whitespace changes should be included in the diff. Please refer to the enum declaration for supported modes.

collect_all_paths can be set when you want to get back all files, even when the enforced limits are hit. Responses sent when the limits were hit will have the path filled in, have the `overflow_marker` set, but have no patch data.

from_path is the path that to_path is getting compared to.

to_path is the path that from_path is getting compared against.

from_id is the id of the blob that is getting compared against.

to_id is the id of the blob that is getting compared to.

old_mode is the mode of the file getting compared against.

new_mode is the mode of the file getting compared to.

binary indicates whether or not binary data is getting compared.

raw_patch_data is the unparsed data of the diff.

end_of_patch indicates if this message represents the end of the diff.

overflow_marker Indicates if the diff file exceeded limitations, in which case there will be no patch data sent, only information about the patch.

collapsed indicates the patch surpassed a "safe" limit and was therefore pruned, but the client may still request the full patch on a separate request.

too_large indicates the patch was pruned since it surpassed a hard limit, and can therefore not be expanded.

repository is the repository the blob diff is generated for.

blob_pairs defines the set of blob pairs to generate diffs for.

diff_mode is the mode used for generating the diff. Refer to the enum declaration above for supported modes.

whitespace_changes states which whitespace changes should be included in the diff. Refer to the enum declaration above for supported modes.

patch_bytes_limit is the max size in bytes of a single patch generated from a pair of blobs. Patches surpassing this limit are pruned. If set to zero, no limit is enforced.

raw_info is the set of file pairs in raw format used to compute diffs for.

left_blob is the left blob in <left blob>..<right blob>. Can be either a path-scoped revision or blob ID.

right_blob is the right blob in <left blob>..<right blob>. Can be either a path-scoped revision or blob ID.

left_blob_id is the left blob ID in in <left blob>..<right blob> of the diffed blob.

right_blob_id is the right blob ID in in <left blob>..<right blob> of the diffed blob.

patch is the raw generated diff output.

status is the current state of the patch being transmitted.

binary indicates whether or not the diff compares binary data. If Git considers either blob to be binary, Git only reports whether the files differ.

over_patch_bytes_limit indicates the patch was pruned since it surpassed the set limit.

patch_size is the size of the patch in bytes and is always returned even if a patch is pruned. To improve performance, a client may make an initial DiffBlobs request with a lower "soft" patch_bytes_limit to avoid receiving large patches. If a patch exceeds this byte limit and gets pruned, the patch size can be used by the client to determine whether it is sensible to make a follow-up DiffBlobs request by increasing the patch_bytes_limit to a "hard" limit.

path is the path of the change.

additions is the number of additions in a diff.

deletions is the number of deletions in a diff.

old_path is the original path in the event of a rename.

repository is the one to get diff stats from.

left_commit_id is the left commit ID in <left commit>..<right commit>

right_commit_id is the right commit ID in <left commit>..<right commit>

stats is a list of DiffStats.

repository is the repository the changed paths are computed in.

Deprecated. commits is the list of commits to compare to their parents. This field is deprecated. To adapt to the new calling convention you can create one `CommitRequest` per commit, where each `CommitRequest` has only the `commit_revision` field.

requests specifies the requests of what to compare.

merge_commit_diff_mode controls how merge commits are treated.

find_renames toggles rename detection on the files in the diff.

tree_request is a request comparing two trees with each other.

commit_request is a request comparing one or more commits with each other.

commit_revision is the revision of the commit that should be compared. If no `parent_commit_revisions` are given, then the commit will be compared against its parents. The revision needs to peel to a commit object.

parent_commit_revisions are the revisions of commits to treat as the commit's parents. This is an optional field: if not specified, the actual parents of the commit referred to by `commit_revision` are used.

left_tree_revision is the revision of the left tree to compare. Accepts any revision that peels to a tree object.

right_tree_revision is the revision of the right tree to compare. Accepts any revision that peels to a tree object.

paths contains the attributes for one changed file. In case of merge commits, or when comparing three or more commits, a file might be included more than once if it was changed between multiple commits.

repository is the repository the patch ID shall be computed in.

old_revision is the old revision that shall be used to compute the patch from that will then be passed to git-patch-id(1). Accepts revisions as specified in gitrevisions(5).

new_revision is the new revision that shall be used to compute the patch from that will then be passed to git-patch-id(1). Accepts revisions as specified in gitrevisions(5).

patch_id is the patch ID that was generated by hashing the diff of the given old and new revision.

repository is the one to get range diff output from.

range_pair specifies format: <range1> <range2>.

revision_range specifies format: <rev1>...<rev2>.

base_with_revisions specifies format: <base> <rev1> <rev2>.

from_commit_id is the left side commit id in the commit pair line.

to_commit_id is the right side commit id in the commit pair line.

comparison is the comparison result of the two commits.

commit_message_title is the commit message title of the commit.

patch_data is the range diff patch data if two commits are not equal.

end_of_patch check if the patch data is the last one response of this commit pair.

range1 is the first range, e.g. <base>..<rev>, <rev>^!

range2 is the second range.e.g. <base>..<rev>, <rev>^!

repository is the one to get the diff from.

left_commit_id is the left commit ID in <left commit>..<right commit>.

right_commit_id is the right commit ID in <left commit>..<right commit>.

data is a sequence of bytes representing the unmodified diff patch data returned from git-diff(1).

repository is the one to get the patch from.

left_commit_id is the left commit ID in <left commit>..<right commit>.

right_commit_id is the right commit ID in <left commit>..<right commit>

data is a sequence of bytes representing the unmodified diff patch data returned from git-format-patch(1).

repository is the one to get range diff output from.

range_pair specifies format: <range1> <range2>.

revision_range specifies format: <rev1>...<rev2>.

base_with_revisions specifies format: <base> <rev1> <rev2>.

data is the raw range diff data.

rev1 is the first revision.

rev2 is the the second revision.

error_message is the error message as returned by the endpoint.

protocol is the protocol used.

user_id is the user ID as which changes had been pushed.

changes is the set of changes which have failed the access check.

reference is the reference passed to the git command.

bad_oid is the bad object id passed to the git command.

stdout is the standard output of the hook that has failed, if any. Data may be truncated.

stderr is the standard error of the hook that has failed, if any. Data may be truncated.

hook_type is the type of the hook.

path is the file or directory path that triggered the error.

error_type is the type of index error that occurred.

refs are the offending refs with invalid formats.

range is the commit range passed to git log command.

error_message provides context into why a limit was enforced.

retry_after provides the duration after which a retry is safe. 0 indicates non-retryable.

conflicting_files is the set of files which have been conflicting. If this field is empty, then there has still been a merge conflict, but it wasn't able to determine which files have been conflicting.

conflicting_commit_ids is the set of commit IDs that caused the conflict. In the general case, this should be set to two commit IDs.

parent_revision is the revision checked against ChildRevision for whether it is an ancestor of ChildRevision

child_revision is the revision checked against ParentRevision for whether it is a descendent of ChildRevision.

path is the file or directory path that triggered the error. The path may be truncated due to size limits.

error_type is the type of path error that occurred.

path is the path that could not be found.

reference_name is the name of the reference that exists already.

oid is the object ID of the reference that preexists already.

reference_name is the name of the reference that does not exist.

reference_name is the name of the reference that was failed to update.

expected_object_id is the object ID that the reference was expected to point to but didn't.

actual_object_id is the object ID that the reference actually pointed to at the time when it was tried to be updated.

reference_name is the name of the reference that failed to be updated.

old_oid is the object ID the reference should have pointed to before the update.

new_oid is the object ID the reference should have pointed to after the update.

refs are the references that could not be locked.

revision is the name of the revision that was tried to be resolved.

repository ...

args contains the arguments passed to the pack-objects hook, without the leading "git"

gl_id is the user id of the initator of the fetch

gl_username is the username of the initator of the fetch

git_protocol is the protocol used for the fetch

remote_ip is the original IP of the initator of the fetch. This field is used as the key for concurrency limiting before we trigger git-pack-objects command

environment_variables contains the environment variables passed to the hook invocation.

repository ...

environment_variables ...

stdin ...

git_push_options ...

stdout ...

stderr ...

exit_status ...

repository ...

environment_variables ...

stdin ...

git_push_options ...

stdout ...

stderr ...

exit_status ...

repository is the repository the hook operates on. This field is unused currently since we gather all information from stdin.

environment_variables is the set of env variables to be passed to the RPC.

stdin is a chunk of raw data which the hook receives.

stdout is a chunk of raw data to be output to stdout of the hook.

stderr is a chunk of raw data to be output to stderr of the hook.

exit_status is the exit status that the hooks outputs, this dictates the success/failure of git-receive-pack(1).

repository ...

environment_variables ...

stdin ...

state ...

stdout ...

stderr ...

exit_status ...

repository ...

environment_variables ...

ref ...

old_value ...

new_value ...

stdout ...

stderr ...

exit_status ...

storage_name ...

relative_path ...

modification_time is the modification time of the repository directory. This can be used as a proxy for when the repository was last modified.

op ...

scope_level indicates what level an RPC interacts with a server: - REPOSITORY: scoped to only a single repo - SERVER: affects the entire server and potentially all repos - STORAGE: scoped to a specific storage location and all repos within - PARTITION: scoped to a specific partition and all repos within