Merge branch 'jt/namespaced-ls-refs-fix'
[git/git.git] / Documentation / technical / pack-protocol.txt
1 Packfile transfer protocols
2 ===========================
4 Git supports transferring data in packfiles over the ssh://, git://, http:// and
5 file:// transports. There exist two sets of protocols, one for pushing
6 data from a client to a server and another for fetching data from a
7 server to a client. The three transports (ssh, git, file) use the same
8 protocol to transfer data. http is documented in http-protocol.txt.
10 The processes invoked in the canonical Git implementation are 'upload-pack'
11 on the server side and 'fetch-pack' on the client side for fetching data;
12 then 'receive-pack' on the server and 'send-pack' on the client for pushing
13 data. The protocol functions to have a server tell a client what is
14 currently on the server, then for the two to negotiate the smallest amount
15 of data to send in order to fully update one or the other.
17 pkt-line Format
18 ---------------
20 The descriptions below build on the pkt-line format described in
21 protocol-common.txt. When the grammar indicate `PKT-LINE(...)`, unless
22 otherwise noted the usual pkt-line LF rules apply: the sender SHOULD
23 include a LF, but the receiver MUST NOT complain if it is not present.
25 An error packet is a special pkt-line that contains an error string.
27 ----
28 error-line = PKT-LINE("ERR" SP explanation-text)
29 ----
31 Throughout the protocol, where `PKT-LINE(...)` is expected, an error packet MAY
32 be sent. Once this packet is sent by a client or a server, the data transfer
33 process defined in this protocol is terminated.
35 Transports
36 ----------
37 There are three transports over which the packfile protocol is
38 initiated. The Git transport is a simple, unauthenticated server that
39 takes the command (almost always 'upload-pack', though Git
40 servers can be configured to be globally writable, in which 'receive-
41 pack' initiation is also allowed) with which the client wishes to
42 communicate and executes it and connects it to the requesting
43 process.
45 In the SSH transport, the client just runs the 'upload-pack'
46 or 'receive-pack' process on the server over the SSH protocol and then
47 communicates with that invoked process over the SSH connection.
49 The file:// transport runs the 'upload-pack' or 'receive-pack'
50 process locally and communicates with it over a pipe.
52 Extra Parameters
53 ----------------
55 The protocol provides a mechanism in which clients can send additional
56 information in its first message to the server. These are called "Extra
57 Parameters", and are supported by the Git, SSH, and HTTP protocols.
59 Each Extra Parameter takes the form of `<key>=<value>` or `<key>`.
61 Servers that receive any such Extra Parameters MUST ignore all
62 unrecognized keys. Currently, the only Extra Parameter recognized is
63 "version" with a value of '1' or '2'. See protocol-v2.txt for more
64 information on protocol version 2.
66 Git Transport
67 -------------
69 The Git transport starts off by sending the command and repository
70 on the wire using the pkt-line format, followed by a NUL byte and a
71 hostname parameter, terminated by a NUL byte.
73 0033git-upload-pack /project.git\\0
75 The transport may send Extra Parameters by adding an additional NUL
76 byte, and then adding one or more NUL-terminated strings:
78 003egit-upload-pack /project.git\\0\0version=1\0
80 --
81 git-proto-request = request-command SP pathname NUL
82 [ host-parameter NUL ] [ NUL extra-parameters ]
83 request-command = "git-upload-pack" / "git-receive-pack" /
84 "git-upload-archive" ; case sensitive
85 pathname = *( %x01-ff ) ; exclude NUL
86 host-parameter = "host=" hostname [ ":" port ]
87 extra-parameters = 1*extra-parameter
88 extra-parameter = 1*( %x01-ff ) NUL
89 --
91 host-parameter is used for the
92 git-daemon name based virtual hosting. See --interpolated-path
93 option to git daemon, with the %H/%CH format characters.
95 Basically what the Git client is doing to connect to an 'upload-pack'
96 process on the server side over the Git protocol is this:
98 $ echo -e -n \
99 "0039git-upload-pack /schacon/gitbook.git\\0" |
100 nc -v 9418
103 SSH Transport
104 -------------
106 Initiating the upload-pack or receive-pack processes over SSH is
107 executing the binary on the server via SSH remote execution.
108 It is basically equivalent to running this:
110 $ ssh "git-upload-pack '/project.git'"
112 For a server to support Git pushing and pulling for a given user over
113 SSH, that user needs to be able to execute one or both of those
114 commands via the SSH shell that they are provided on login. On some
115 systems, that shell access is limited to only being able to run those
116 two commands, or even just one of them.
118 In an ssh:// format URI, it's absolute in the URI, so the '/' after
119 the host name (or port number) is sent as an argument, which is then
120 read by the remote git-upload-pack exactly as is, so it's effectively
121 an absolute path in the remote filesystem.
123 git clone ssh://
124 |
125 v
126 ssh "git-upload-pack '/project.git'"
128 In a "user@host:path" format URI, its relative to the user's home
129 directory, because the Git client will run:
131 git clone
132 |
133 v
134 ssh "git-upload-pack 'project.git'"
136 The exception is if a '~' is used, in which case
137 we execute it without the leading '/'.
139 ssh://,
140 |
141 v
142 ssh "git-upload-pack '~alice/project.git'"
144 Depending on the value of the `protocol.version` configuration variable,
145 Git may attempt to send Extra Parameters as a colon-separated string in
146 the GIT_PROTOCOL environment variable. This is done only if
147 the `ssh.variant` configuration variable indicates that the ssh command
148 supports passing environment variables as an argument.
150 A few things to remember here:
152 - The "command name" is spelled with dash (e.g. git-upload-pack), but
153 this can be overridden by the client;
155 - The repository path is always quoted with single quotes.
157 Fetching Data From a Server
158 ---------------------------
160 When one Git repository wants to get data that a second repository
161 has, the first can 'fetch' from the second. This operation determines
162 what data the server has that the client does not then streams that
163 data down to the client in packfile format.
166 Reference Discovery
167 -------------------
169 When the client initially connects the server will immediately respond
170 with a version number (if "version=1" is sent as an Extra Parameter),
171 and a listing of each reference it has (all branches and tags) along
172 with the object name that each reference currently points to.
174 $ echo -e -n "0044git-upload-pack /schacon/gitbook.git\\0\0version=1\0" |
175 nc -v 9418
176 000aversion 1
177 00887217a7c7e582c46cec22a130adf4b9d7d950fba0 HEAD\0multi_ack thin-pack
178 side-band side-band-64k ofs-delta shallow no-progress include-tag
179 00441d3fcd5ced445d1abc402225c0b8a1299641f497 refs/heads/integration
180 003f7217a7c7e582c46cec22a130adf4b9d7d950fba0 refs/heads/master
181 003cb88d2441cac0977faf98efc80305012112238d9d refs/tags/v0.9
182 003c525128480b96c89e6418b1e40909bf6c5b2d580f refs/tags/v1.0
183 003fe92df48743b7bc7d26bcaabfddde0a1e20cae47c refs/tags/v1.0^{}
184 0000
186 The returned response is a pkt-line stream describing each ref and
187 its current value. The stream MUST be sorted by name according to
188 the C locale ordering.
190 If HEAD is a valid ref, HEAD MUST appear as the first advertised
191 ref. If HEAD is not a valid ref, HEAD MUST NOT appear in the
192 advertisement list at all, but other refs may still appear.
194 The stream MUST include capability declarations behind a NUL on the
195 first ref. The peeled value of a ref (that is "ref^{}") MUST be
196 immediately after the ref itself, if presented. A conforming server
197 MUST peel the ref if it's an annotated tag.
199 ----
200 advertised-refs = *1("version 1")
201 (no-refs / list-of-refs)
202 *shallow
203 flush-pkt
205 no-refs = PKT-LINE(zero-id SP "capabilities^{}"
206 NUL capability-list)
208 list-of-refs = first-ref *other-ref
209 first-ref = PKT-LINE(obj-id SP refname
210 NUL capability-list)
212 other-ref = PKT-LINE(other-tip / other-peeled)
213 other-tip = obj-id SP refname
214 other-peeled = obj-id SP refname "^{}"
216 shallow = PKT-LINE("shallow" SP obj-id)
218 capability-list = capability *(SP capability)
219 capability = 1*(LC_ALPHA / DIGIT / "-" / "_")
220 LC_ALPHA = %x61-7A
221 ----
223 Server and client MUST use lowercase for obj-id, both MUST treat obj-id
224 as case-insensitive.
226 See protocol-capabilities.txt for a list of allowed server capabilities
227 and descriptions.
229 Packfile Negotiation
230 --------------------
231 After reference and capabilities discovery, the client can decide to
232 terminate the connection by sending a flush-pkt, telling the server it can
233 now gracefully terminate, and disconnect, when it does not need any pack
234 data. This can happen with the ls-remote command, and also can happen when
235 the client already is up to date.
237 Otherwise, it enters the negotiation phase, where the client and
238 server determine what the minimal packfile necessary for transport is,
239 by telling the server what objects it wants, its shallow objects
240 (if any), and the maximum commit depth it wants (if any). The client
241 will also send a list of the capabilities it wants to be in effect,
242 out of what the server said it could do with the first 'want' line.
244 ----
245 upload-request = want-list
246 *shallow-line
247 *1depth-request
248 [filter-request]
249 flush-pkt
251 want-list = first-want
252 *additional-want
254 shallow-line = PKT-LINE("shallow" SP obj-id)
256 depth-request = PKT-LINE("deepen" SP depth) /
257 PKT-LINE("deepen-since" SP timestamp) /
258 PKT-LINE("deepen-not" SP ref)
260 first-want = PKT-LINE("want" SP obj-id SP capability-list)
261 additional-want = PKT-LINE("want" SP obj-id)
263 depth = 1*DIGIT
265 filter-request = PKT-LINE("filter" SP filter-spec)
266 ----
268 Clients MUST send all the obj-ids it wants from the reference
269 discovery phase as 'want' lines. Clients MUST send at least one
270 'want' command in the request body. Clients MUST NOT mention an
271 obj-id in a 'want' command which did not appear in the response
272 obtained through ref discovery.
274 The client MUST write all obj-ids which it only has shallow copies
275 of (meaning that it does not have the parents of a commit) as
276 'shallow' lines so that the server is aware of the limitations of
277 the client's history.
279 The client now sends the maximum commit history depth it wants for
280 this transaction, which is the number of commits it wants from the
281 tip of the history, if any, as a 'deepen' line. A depth of 0 is the
282 same as not making a depth request. The client does not want to receive
283 any commits beyond this depth, nor does it want objects needed only to
284 complete those commits. Commits whose parents are not received as a
285 result are defined as shallow and marked as such in the server. This
286 information is sent back to the client in the next step.
288 The client can optionally request that pack-objects omit various
289 objects from the packfile using one of several filtering techniques.
290 These are intended for use with partial clone and partial fetch
291 operations. An object that does not meet a filter-spec value is
292 omitted unless explicitly requested in a 'want' line. See `rev-list`
293 for possible filter-spec values.
295 Once all the 'want's and 'shallow's (and optional 'deepen') are
296 transferred, clients MUST send a flush-pkt, to tell the server side
297 that it is done sending the list.
299 Otherwise, if the client sent a positive depth request, the server
300 will determine which commits will and will not be shallow and
301 send this information to the client. If the client did not request
302 a positive depth, this step is skipped.
304 ----
305 shallow-update = *shallow-line
306 *unshallow-line
307 flush-pkt
309 shallow-line = PKT-LINE("shallow" SP obj-id)
311 unshallow-line = PKT-LINE("unshallow" SP obj-id)
312 ----
314 If the client has requested a positive depth, the server will compute
315 the set of commits which are no deeper than the desired depth. The set
316 of commits start at the client's wants.
318 The server writes 'shallow' lines for each
319 commit whose parents will not be sent as a result. The server writes
320 an 'unshallow' line for each commit which the client has indicated is
321 shallow, but is no longer shallow at the currently requested depth
322 (that is, its parents will now be sent). The server MUST NOT mark
323 as unshallow anything which the client has not indicated was shallow.
325 Now the client will send a list of the obj-ids it has using 'have'
326 lines, so the server can make a packfile that only contains the objects
327 that the client needs. In multi_ack mode, the canonical implementation
328 will send up to 32 of these at a time, then will send a flush-pkt. The
329 canonical implementation will skip ahead and send the next 32 immediately,
330 so that there is always a block of 32 "in-flight on the wire" at a time.
332 ----
333 upload-haves = have-list
334 compute-end
336 have-list = *have-line
337 have-line = PKT-LINE("have" SP obj-id)
338 compute-end = flush-pkt / PKT-LINE("done")
339 ----
341 If the server reads 'have' lines, it then will respond by ACKing any
342 of the obj-ids the client said it had that the server also has. The
343 server will ACK obj-ids differently depending on which ack mode is
344 chosen by the client.
346 In multi_ack mode:
348 * the server will respond with 'ACK obj-id continue' for any common
349 commits.
351 * once the server has found an acceptable common base commit and is
352 ready to make a packfile, it will blindly ACK all 'have' obj-ids
353 back to the client.
355 * the server will then send a 'NAK' and then wait for another response
356 from the client - either a 'done' or another list of 'have' lines.
358 In multi_ack_detailed mode:
360 * the server will differentiate the ACKs where it is signaling
361 that it is ready to send data with 'ACK obj-id ready' lines, and
362 signals the identified common commits with 'ACK obj-id common' lines.
364 Without either multi_ack or multi_ack_detailed:
366 * upload-pack sends "ACK obj-id" on the first common object it finds.
367 After that it says nothing until the client gives it a "done".
369 * upload-pack sends "NAK" on a flush-pkt if no common object
370 has been found yet. If one has been found, and thus an ACK
371 was already sent, it's silent on the flush-pkt.
373 After the client has gotten enough ACK responses that it can determine
374 that the server has enough information to send an efficient packfile
375 (in the canonical implementation, this is determined when it has received
376 enough ACKs that it can color everything left in the --date-order queue
377 as common with the server, or the --date-order queue is empty), or the
378 client determines that it wants to give up (in the canonical implementation,
379 this is determined when the client sends 256 'have' lines without getting
380 any of them ACKed by the server - meaning there is nothing in common and
381 the server should just send all of its objects), then the client will send
382 a 'done' command. The 'done' command signals to the server that the client
383 is ready to receive its packfile data.
385 However, the 256 limit *only* turns on in the canonical client
386 implementation if we have received at least one "ACK %s continue"
387 during a prior round. This helps to ensure that at least one common
388 ancestor is found before we give up entirely.
390 Once the 'done' line is read from the client, the server will either
391 send a final 'ACK obj-id' or it will send a 'NAK'. 'obj-id' is the object
392 name of the last commit determined to be common. The server only sends
393 ACK after 'done' if there is at least one common base and multi_ack or
394 multi_ack_detailed is enabled. The server always sends NAK after 'done'
395 if there is no common base found.
397 Instead of 'ACK' or 'NAK', the server may send an error message (for
398 example, if it does not recognize an object in a 'want' line received
399 from the client).
401 Then the server will start sending its packfile data.
403 ----
404 server-response = *ack_multi ack / nak
405 ack_multi = PKT-LINE("ACK" SP obj-id ack_status)
406 ack_status = "continue" / "common" / "ready"
407 ack = PKT-LINE("ACK" SP obj-id)
408 nak = PKT-LINE("NAK")
409 ----
411 A simple clone may look like this (with no 'have' lines):
413 ----
414 C: 0054want 74730d410fcb6603ace96f1dc55ea6196122532d multi_ack \
415 side-band-64k ofs-delta\n
416 C: 0032want 7d1665144a3a975c05f1f43902ddaf084e784dbe\n
417 C: 0032want 5a3f6be755bbb7deae50065988cbfa1ffa9ab68a\n
418 C: 0032want 7e47fe2bd8d01d481f44d7af0531bd93d3b21c01\n
419 C: 0032want 74730d410fcb6603ace96f1dc55ea6196122532d\n
420 C: 0000
421 C: 0009done\n
423 S: 0008NAK\n
425 ----
427 An incremental update (fetch) response might look like this:
429 ----
430 C: 0054want 74730d410fcb6603ace96f1dc55ea6196122532d multi_ack \
431 side-band-64k ofs-delta\n
432 C: 0032want 7d1665144a3a975c05f1f43902ddaf084e784dbe\n
433 C: 0032want 5a3f6be755bbb7deae50065988cbfa1ffa9ab68a\n
434 C: 0000
435 C: 0032have 7e47fe2bd8d01d481f44d7af0531bd93d3b21c01\n
436 C: [30 more have lines]
437 C: 0032have 74730d410fcb6603ace96f1dc55ea6196122532d\n
438 C: 0000
440 S: 003aACK 7e47fe2bd8d01d481f44d7af0531bd93d3b21c01 continue\n
441 S: 003aACK 74730d410fcb6603ace96f1dc55ea6196122532d continue\n
442 S: 0008NAK\n
444 C: 0009done\n
446 S: 0031ACK 74730d410fcb6603ace96f1dc55ea6196122532d\n
448 ----
451 Packfile Data
452 -------------
454 Now that the client and server have finished negotiation about what
455 the minimal amount of data that needs to be sent to the client is, the server
456 will construct and send the required data in packfile format.
458 See pack-format.txt for what the packfile itself actually looks like.
460 If 'side-band' or 'side-band-64k' capabilities have been specified by
461 the client, the server will send the packfile data multiplexed.
463 Each packet starting with the packet-line length of the amount of data
464 that follows, followed by a single byte specifying the sideband the
465 following data is coming in on.
467 In 'side-band' mode, it will send up to 999 data bytes plus 1 control
468 code, for a total of up to 1000 bytes in a pkt-line. In 'side-band-64k'
469 mode it will send up to 65519 data bytes plus 1 control code, for a
470 total of up to 65520 bytes in a pkt-line.
472 The sideband byte will be a '1', '2' or a '3'. Sideband '1' will contain
473 packfile data, sideband '2' will be used for progress information that the
474 client will generally print to stderr and sideband '3' is used for error
475 information.
477 If no 'side-band' capability was specified, the server will stream the
478 entire packfile without multiplexing.
481 Pushing Data To a Server
482 ------------------------
484 Pushing data to a server will invoke the 'receive-pack' process on the
485 server, which will allow the client to tell it which references it should
486 update and then send all the data the server will need for those new
487 references to be complete. Once all the data is received and validated,
488 the server will then update its references to what the client specified.
490 Authentication
491 --------------
493 The protocol itself contains no authentication mechanisms. That is to be
494 handled by the transport, such as SSH, before the 'receive-pack' process is
495 invoked. If 'receive-pack' is configured over the Git transport, those
496 repositories will be writable by anyone who can access that port (9418) as
497 that transport is unauthenticated.
499 Reference Discovery
500 -------------------
502 The reference discovery phase is done nearly the same way as it is in the
503 fetching protocol. Each reference obj-id and name on the server is sent
504 in packet-line format to the client, followed by a flush-pkt. The only
505 real difference is that the capability listing is different - the only
506 possible values are 'report-status', 'delete-refs', 'ofs-delta' and
507 'push-options'.
509 Reference Update Request and Packfile Transfer
510 ----------------------------------------------
512 Once the client knows what references the server is at, it can send a
513 list of reference update requests. For each reference on the server
514 that it wants to update, it sends a line listing the obj-id currently on
515 the server, the obj-id the client would like to update it to and the name
516 of the reference.
518 This list is followed by a flush-pkt.
520 ----
521 update-requests = *shallow ( command-list | push-cert )
523 shallow = PKT-LINE("shallow" SP obj-id)
525 command-list = PKT-LINE(command NUL capability-list)
526 *PKT-LINE(command)
527 flush-pkt
529 command = create / delete / update
530 create = zero-id SP new-id SP name
531 delete = old-id SP zero-id SP name
532 update = old-id SP new-id SP name
534 old-id = obj-id
535 new-id = obj-id
537 push-cert = PKT-LINE("push-cert" NUL capability-list LF)
538 PKT-LINE("certificate version 0.1" LF)
539 PKT-LINE("pusher" SP ident LF)
540 PKT-LINE("pushee" SP url LF)
541 PKT-LINE("nonce" SP nonce LF)
542 *PKT-LINE("push-option" SP push-option LF)
544 *PKT-LINE(command LF)
545 *PKT-LINE(gpg-signature-lines LF)
546 PKT-LINE("push-cert-end" LF)
548 push-option = 1*( VCHAR | SP )
549 ----
551 If the server has advertised the 'push-options' capability and the client has
552 specified 'push-options' as part of the capability list above, the client then
553 sends its push options followed by a flush-pkt.
555 ----
556 push-options = *PKT-LINE(push-option) flush-pkt
557 ----
559 For backwards compatibility with older Git servers, if the client sends a push
560 cert and push options, it MUST send its push options both embedded within the
561 push cert and after the push cert. (Note that the push options within the cert
562 are prefixed, but the push options after the cert are not.) Both these lists
563 MUST be the same, modulo the prefix.
565 After that the packfile that
566 should contain all the objects that the server will need to complete the new
567 references will be sent.
569 ----
570 packfile = "PACK" 28*(OCTET)
571 ----
573 If the receiving end does not support delete-refs, the sending end MUST
574 NOT ask for delete command.
576 If the receiving end does not support push-cert, the sending end
577 MUST NOT send a push-cert command. When a push-cert command is
578 sent, command-list MUST NOT be sent; the commands recorded in the
579 push certificate is used instead.
581 The packfile MUST NOT be sent if the only command used is 'delete'.
583 A packfile MUST be sent if either create or update command is used,
584 even if the server already has all the necessary objects. In this
585 case the client MUST send an empty packfile. The only time this
586 is likely to happen is if the client is creating
587 a new branch or a tag that points to an existing obj-id.
589 The server will receive the packfile, unpack it, then validate each
590 reference that is being updated that it hasn't changed while the request
591 was being processed (the obj-id is still the same as the old-id), and
592 it will run any update hooks to make sure that the update is acceptable.
593 If all of that is fine, the server will then update the references.
595 Push Certificate
596 ----------------
598 A push certificate begins with a set of header lines. After the
599 header and an empty line, the protocol commands follow, one per
600 line. Note that the trailing LF in push-cert PKT-LINEs is _not_
601 optional; it must be present.
603 Currently, the following header fields are defined:
605 `pusher` ident::
606 Identify the GPG key in "Human Readable Name <email@address>"
607 format.
609 `pushee` url::
610 The repository URL (anonymized, if the URL contains
611 authentication material) the user who ran `git push`
612 intended to push into.
614 `nonce` nonce::
615 The 'nonce' string the receiving repository asked the
616 pushing user to include in the certificate, to prevent
617 replay attacks.
619 The GPG signature lines are a detached signature for the contents
620 recorded in the push certificate before the signature block begins.
621 The detached signature is used to certify that the commands were
622 given by the pusher, who must be the signer.
624 Report Status
625 -------------
627 After receiving the pack data from the sender, the receiver sends a
628 report if 'report-status' capability is in effect.
629 It is a short listing of what happened in that update. It will first
630 list the status of the packfile unpacking as either 'unpack ok' or
631 'unpack [error]'. Then it will list the status for each of the references
632 that it tried to update. Each line is either 'ok [refname]' if the
633 update was successful, or 'ng [refname] [error]' if the update was not.
635 ----
636 report-status = unpack-status
637 1*(command-status)
638 flush-pkt
640 unpack-status = PKT-LINE("unpack" SP unpack-result)
641 unpack-result = "ok" / error-msg
643 command-status = command-ok / command-fail
644 command-ok = PKT-LINE("ok" SP refname)
645 command-fail = PKT-LINE("ng" SP refname SP error-msg)
647 error-msg = 1*(OCTECT) ; where not "ok"
648 ----
650 Updates can be unsuccessful for a number of reasons. The reference can have
651 changed since the reference discovery phase was originally sent, meaning
652 someone pushed in the meantime. The reference being pushed could be a
653 non-fast-forward reference and the update hooks or configuration could be
654 set to not allow that, etc. Also, some references can be updated while others
655 can be rejected.
657 An example client/server communication might look like this:
659 ----
660 S: 007c74730d410fcb6603ace96f1dc55ea6196122532d refs/heads/local\0report-status delete-refs ofs-delta\n
661 S: 003e7d1665144a3a975c05f1f43902ddaf084e784dbe refs/heads/debug\n
662 S: 003f74730d410fcb6603ace96f1dc55ea6196122532d refs/heads/master\n
663 S: 003f74730d410fcb6603ace96f1dc55ea6196122532d refs/heads/team\n
664 S: 0000
666 C: 003e7d1665144a3a975c05f1f43902ddaf084e784dbe 74730d410fcb6603ace96f1dc55ea6196122532d refs/heads/debug\n
667 C: 003e74730d410fcb6603ace96f1dc55ea6196122532d 5a3f6be755bbb7deae50065988cbfa1ffa9ab68a refs/heads/master\n
668 C: 0000
671 S: 000eunpack ok\n
672 S: 0018ok refs/heads/debug\n
673 S: 002ang refs/heads/master non-fast-forward\n
674 ----