]>
Commit | Line | Data |
---|---|---|
b31222cf SC |
1 | Git Protocol Capabilities |
2 | ========================= | |
3 | ||
4 | Servers SHOULD support all capabilities defined in this document. | |
5 | ||
6 | On the very first line of the initial server response of either | |
7 | receive-pack and upload-pack the first reference is followed by | |
8 | a NUL byte and then a list of space delimited server capabilities. | |
9 | These allow the server to declare what it can and cannot support | |
10 | to the client. | |
11 | ||
12 | Client will then send a space separated list of capabilities it wants | |
13 | to be in effect. The client MUST NOT ask for capabilities the server | |
14 | did not say it supports. | |
15 | ||
16 | Server MUST diagnose and abort if capabilities it does not understand | |
17 | was sent. Server MUST NOT ignore capabilities that client requested | |
18 | and server advertised. As a consequence of these rules, server MUST | |
19 | NOT advertise capabilities it does not understand. | |
20 | ||
69fb9603 | 21 | The 'report-status', 'delete-refs', and 'quiet' capabilities are sent and |
b31222cf SC |
22 | recognized by the receive-pack (push to server) process. |
23 | ||
9354b9a4 | 24 | The 'ofs-delta' and 'side-band-64k' capabilities are sent and recognized |
af608260 JK |
25 | by both upload-pack and receive-pack protocols. The 'agent' capability |
26 | may optionally be sent in both protocols. | |
b31222cf SC |
27 | |
28 | All other capabilities are only recognized by the upload-pack (fetch | |
29 | from server) process. | |
30 | ||
31 | multi_ack | |
32 | --------- | |
33 | ||
34 | The 'multi_ack' capability allows the server to return "ACK obj-id | |
35 | continue" as soon as it finds a commit that it can use as a common | |
36 | base, between the client's wants and the client's have set. | |
37 | ||
38 | By sending this early, the server can potentially head off the client | |
39 | from walking any further down that particular branch of the client's | |
40 | repository history. The client may still need to walk down other | |
41 | branches, sending have lines for those, until the server has a | |
42 | complete cut across the DAG, or the client has said "done". | |
43 | ||
44 | Without multi_ack, a client sends have lines in --date-order until | |
45 | the server has found a common base. That means the client will send | |
46 | have lines that are already known by the server to be common, because | |
47 | they overlap in time with another branch that the server hasn't found | |
48 | a common base on yet. | |
49 | ||
50 | For example suppose the client has commits in caps that the server | |
51 | doesn't and the server has commits in lower case that the client | |
52 | doesn't, as in the following diagram: | |
53 | ||
54 | +---- u ---------------------- x | |
55 | / +----- y | |
56 | / / | |
57 | a -- b -- c -- d -- E -- F | |
58 | \ | |
59 | +--- Q -- R -- S | |
60 | ||
61 | If the client wants x,y and starts out by saying have F,S, the server | |
62 | doesn't know what F,S is. Eventually the client says "have d" and | |
63 | the server sends "ACK d continue" to let the client know to stop | |
6a5d0b0a | 64 | walking down that line (so don't send c-b-a), but it's not done yet, |
b31222cf SC |
65 | it needs a base for x. The client keeps going with S-R-Q, until a |
66 | gets reached, at which point the server has a clear base and it all | |
67 | ends. | |
68 | ||
69 | Without multi_ack the client would have sent that c-b-a chain anyway, | |
70 | interleaved with S-R-Q. | |
71 | ||
72 | thin-pack | |
73 | --------- | |
74 | ||
1ba98a79 CMN |
75 | A thin pack is one with deltas which reference base objects not |
76 | contained within the pack (but are known to exist at the receiving | |
77 | end). This can reduce the network traffic significantly, but it | |
78 | requires the receiving end to know how to "thicken" these packs by | |
79 | adding the missing bases to the pack. | |
80 | ||
81 | The upload-pack server advertises 'thin-pack' when it can generate | |
82 | and send a thin pack. A client requests the 'thin-pack' capability | |
83 | when it understands how to "thicken" it, notifying the server that | |
84 | it can receive such a pack. A client MUST NOT request the | |
85 | 'thin-pack' capability if it cannot turn a thin pack into a | |
86 | self-contained pack. | |
87 | ||
88 | Receive-pack, on the other hand, is assumed by default to be able to | |
89 | handle thin packs, but can ask the client not to use the feature by | |
90 | advertising the 'no-thin' capability. A client MUST NOT send a thin | |
91 | pack if the server advertises the 'no-thin' capability. | |
92 | ||
93 | The reasons for this asymmetry are historical. The receive-pack | |
94 | program did not exist until after the invention of thin packs, so | |
95 | historically the reference implementation of receive-pack always | |
96 | understood thin packs. Adding 'no-thin' later allowed receive-pack | |
97 | to disable the feature in a backwards-compatible manner. | |
b31222cf SC |
98 | |
99 | ||
100 | side-band, side-band-64k | |
101 | ------------------------ | |
102 | ||
103 | This capability means that server can send, and client understand multiplexed | |
104 | progress reports and error info interleaved with the packfile itself. | |
105 | ||
106 | These two options are mutually exclusive. A modern client always | |
107 | favors 'side-band-64k'. | |
108 | ||
109 | Either mode indicates that the packfile data will be streamed broken | |
110 | up into packets of up to either 1000 bytes in the case of 'side_band', | |
111 | or 65520 bytes in the case of 'side_band_64k'. Each packet is made up | |
112 | of a leading 4-byte pkt-line length of how much data is in the packet, | |
113 | followed by a 1-byte stream code, followed by the actual data. | |
114 | ||
115 | The stream code can be one of: | |
116 | ||
117 | 1 - pack data | |
118 | 2 - progress messages | |
119 | 3 - fatal error message just before stream aborts | |
120 | ||
121 | The "side-band-64k" capability came about as a way for newer clients | |
122 | that can handle much larger packets to request packets that are | |
123 | actually crammed nearly full, while maintaining backward compatibility | |
124 | for the older clients. | |
125 | ||
126 | Further, with side-band and its up to 1000-byte messages, it's actually | |
127 | 999 bytes of payload and 1 byte for the stream code. With side-band-64k, | |
128 | same deal, you have up to 65519 bytes of data and 1 byte for the stream | |
129 | code. | |
130 | ||
131 | The client MUST send only maximum of one of "side-band" and "side- | |
132 | band-64k". Server MUST diagnose it as an error if client requests | |
133 | both. | |
134 | ||
135 | ofs-delta | |
136 | --------- | |
137 | ||
5d1e3415 | 138 | Server can send, and client understand PACKv2 with delta referring to |
b31222cf SC |
139 | its base by position in pack rather than by an obj-id. That is, they can |
140 | send/read OBJ_OFS_DELTA (aka type 6) in a packfile. | |
141 | ||
af608260 JK |
142 | agent |
143 | ----- | |
144 | ||
145 | The server may optionally send a capability of the form `agent=X` to | |
146 | notify the client that the server is running version `X`. The client may | |
147 | optionally return its own agent string by responding with an `agent=Y` | |
148 | capability (but it MUST NOT do so if the server did not mention the | |
149 | agent capability). The `X` and `Y` strings may contain any printable | |
150 | ASCII characters except space (i.e., the byte range 32 < x < 127), and | |
151 | are typically of the form "package/version" (e.g., "git/1.8.3.1"). The | |
152 | agent strings are purely informative for statistics and debugging | |
153 | purposes, and MUST NOT be used to programatically assume the presence | |
154 | or absence of particular features. | |
155 | ||
b31222cf SC |
156 | shallow |
157 | ------- | |
158 | ||
159 | This capability adds "deepen", "shallow" and "unshallow" commands to | |
160 | the fetch-pack/upload-pack protocol so clients can request shallow | |
161 | clones. | |
162 | ||
163 | no-progress | |
164 | ----------- | |
165 | ||
166 | The client was started with "git clone -q" or something, and doesn't | |
167 | want that side band 2. Basically the client just says "I do not | |
168 | wish to receive stream 2 on sideband, so do not send it to me, and if | |
169 | you did, I will drop it on the floor anyway". However, the sideband | |
170 | channel 3 is still used for error responses. | |
171 | ||
172 | include-tag | |
173 | ----------- | |
174 | ||
175 | The 'include-tag' capability is about sending annotated tags if we are | |
176 | sending objects they point to. If we pack an object to the client, and | |
177 | a tag object points exactly at that object, we pack the tag object too. | |
178 | In general this allows a client to get all new annotated tags when it | |
179 | fetches a branch, in a single network connection. | |
180 | ||
181 | Clients MAY always send include-tag, hardcoding it into a request when | |
182 | the server advertises this capability. The decision for a client to | |
183 | request include-tag only has to do with the client's desires for tag | |
184 | data, whether or not a server had advertised objects in the | |
185 | refs/tags/* namespace. | |
186 | ||
187 | Servers MUST pack the tags if their referrant is packed and the client | |
188 | has requested include-tags. | |
189 | ||
190 | Clients MUST be prepared for the case where a server has ignored | |
191 | include-tag and has not actually sent tags in the pack. In such | |
192 | cases the client SHOULD issue a subsequent fetch to acquire the tags | |
193 | that include-tag would have otherwise given the client. | |
194 | ||
195 | The server SHOULD send include-tag, if it supports it, regardless | |
196 | of whether or not there are tags available. | |
197 | ||
198 | report-status | |
199 | ------------- | |
200 | ||
9a621ad0 | 201 | The receive-pack process can receive a 'report-status' capability, |
b31222cf SC |
202 | which tells it that the client wants a report of what happened after |
203 | a packfile upload and reference update. If the pushing client requests | |
204 | this capability, after unpacking and updating references the server | |
205 | will respond with whether the packfile unpacked successfully and if | |
206 | each reference was updated successfully. If any of those were not | |
207 | successful, it will send back an error message. See pack-protocol.txt | |
208 | for example messages. | |
209 | ||
210 | delete-refs | |
211 | ----------- | |
212 | ||
213 | If the server sends back the 'delete-refs' capability, it means that | |
6a5d0b0a | 214 | it is capable of accepting a zero-id value as the target |
b31222cf SC |
215 | value of a reference update. It is not sent back by the client, it |
216 | simply informs the client that it can be sent zero-id values | |
217 | to delete references. | |
69fb9603 JK |
218 | |
219 | quiet | |
220 | ----- | |
221 | ||
222 | If the receive-pack server advertises the 'quiet' capability, it is | |
223 | capable of silencing human-readable progress output which otherwise may | |
224 | be shown when processing the received pack. A send-pack client should | |
225 | respond with the 'quiet' capability to suppress server-side progress | |
226 | reporting if the local progress reporting is also being suppressed | |
227 | (e.g., via `push -q`, or if stderr does not go to a tty). | |
4acbe91a NTND |
228 | |
229 | allow-tip-sha1-in-want | |
230 | ---------------------- | |
231 | ||
232 | If the upload-pack server advertises this capability, fetch-pack may | |
233 | send "want" lines with SHA-1s that exist at the server but are not | |
234 | advertised by upload-pack. |