| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
|
|
|
|
|
|
|
|
| |
On session.ReceivePack error the gororutine doing the encoding got
blocked either writing objects to the pipe or sending error to the done
channel. The problem did not cause a perceived problem but left blocked
goroutines.
Signed-off-by: Javi Fontan <jfontan@gmail.com>
|
|
|
|
|
| |
Issue: #909
Signed-off-by: Jeremy Stribling <strib@alum.mit.edu>
|
|
|
|
|
|
| |
support offset deltas
Signed-off-by: Benjamin Ash <bash@intelerad.com>
|
|
|
|
|
|
| |
Relates to #870
Signed-off-by: Fedor Korotkov <fedor.korotkov@gmail.com>
|
|
|
|
| |
Signed-off-by: Máximo Cuadros <mcuadros@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The current code iterates all the references in the remote to check if
they match the refspec. This is OK when the refspec is a wildcard but
is a waste of time when they are not.
A hash with references is generated for fast access before starting the
update and used only when the refspec is not a wildcard.
In a repository with 7800 references this meant 7800 * 7800 checks. With
the current code it took 8m30s to update the references. With the new
code it takes less than 0.5s.
References are already extensively tested in remote_test.go.
Signed-off-by: Javi Fontan <jfontan@gmail.com>
|
|
|
|
|
|
|
|
|
|
| |
updateShallow substituted the previous shallow list with the one
returned by the UploadPackResponse. If the repository had previous
shallow commits these are deleted from the list.
This change adds the new shallow hashes to the old ones.
Signed-off-by: Javi Fontan <jfontan@gmail.com>
|
|
|
|
|
|
|
|
| |
Previously some close errors were losts. This is specially problematic
in go-git as lots of work is done here like generating indexes and
moving packfiles.
Signed-off-by: Javi Fontan <jfontan@gmail.com>
|
|
|
|
|
|
| |
It always returns a nil error.
Signed-off-by: Daniel Martí <mvdan@mvdan.cc>
|
|
|
|
|
|
|
|
|
|
| |
- no length for map initialization
- don't check for boolean/error return
- don't format string
- use string method of bytes buffer instead of converting bytes to
string
- use `strings.Contains` instead of `strings.Index`
- use `bytes.Equal` instead of `bytes.Compare`
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
Signed-off-by: Máximo Cuadros <mcuadros@gmail.com>
|
|\
| |
| | |
fix: a range loop can break in advance
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
If the local ref is not an ancestor of the remote ref being fetched,
then when we send an UploadPack request with that local ref as one of
the Haves, the remote will not recognize it, and will think we are
asking for the entire history of the repo, even if there's a common
ancestor.
To do this right, we need to support the multi-ack protocol so we can
negotiate a common commit. That's hard though; this is a quick fix
just to include the previous 100 commits for each local ref in the
Haves list, and hope that one of them is the common commit.
|
|\ \
| | |
| | | |
remote: add support for ls-remote
|
| | | |
|
| |/ |
|
|\ \
| | |
| | | |
plumbing: the commit walker can skip externally-seen commits
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
When the revlist is computing the set of hashes needed to transfer, it
doesn't need to walk over commits it has already processed. So, it
can instruct the commit walker not to walk those commits by passing in
its own `seen` map.
For a 36K object repo, this brought the time for `revlist.Objects`
down from 50s to 30s.
|
|/ /
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
One use of go-git is to transfer git data from a non-standard git repo
(not stored in a file system, for example) to a "remote" backed by a
standard, local .git repo.
In this scenario, delta compression is not needed to reduce transfer
time over the "network", because there is no network. The underlying
storage layer has already taken care of the data tranfer, and sending
the objects to local .git storage doesn't require compression. So this
PR gives the user the option to turn off compression when it isn't
needed.
Of course, this results in a larger, uncompressed local .git repo, but
the user can then run git gc or git repack on that repo if they care
about the storage costs.
Turning the pack window to 0 on reduces total push time of a 36K repo
by 50 seconds (out of a pre-PR total of 3m26s).
|
|/
|
|
| |
Signed-off-by: Miguel Molina <miguel@erizocosmi.co>
|
| |
|
| |
|
| |
|
|\
| |
| | |
plumbing: moved `Reference.Is*` methods to `ReferenceName.Is*`
|
| | |
|
|/
|
|
|
|
|
|
| |
* Change `URL string` to `URL []string` in `RemoteConfig`, since
git allows multiple URLs per remote. See:
http://marc.info/?l=git&m=116231242118202&w=2
* Fix marshalling of multiple fetch refspecs.
|
| |
|
|\
| |
| | |
transport: context package support allowing cancellation of any network operation
|
| | |
|
|/ |
|
| |
|
|\
| |
| | |
remote: fetch, correct behavior on tags
|
| | |
|
|/ |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* server: implement delete-refs and announce it.
* remote: check if server announced delete-refs before trying
to delete and fail fast if it does not.
Note that the client does not need no send 'delete-refs' back
to the server to be able to delete references:
```
delete-refs
-----------
If the server sends back the 'delete-refs' capability, it means that
it is capable of accepting a zero-id value as the target
value of a reference update. It is not sent back by the client, it
simply informs the client that it can be sent zero-id values
to delete references.
```
So our server implementation does not check if the client sent
delete-refs back, it just accepts deletes if it receives them.
|
|
|
|
|
| |
Refspecs for deletes were not being used to produce
delete commands on the update request.
|
| |
|
|\
| |
| | |
fix naming of NewCommit{Pre,Post}Iterator
|
| |
| |
| |
| |
| | |
Use Iter suffix, just as all other iterators in the project.
Use Preorder and Postorder to be more clear.
|
|/
|
|
|
|
| |
* at low level, ReceivePack must close its stream to the
server to signal it has finished.
* remote.go: Close() must be called on session.
|
|\
| |
| | |
add test for tags push, closes #354
|
| | |
|
|\ \
| |/
|/| |
transport: make Endpoint an interface, fixes #362
|
| |
| |
| |
| |
| | |
* add internal *url.URL implementation for regular URLs.
* add internal implementation for SCP-like URLs.
|
|/
|
|
|
| |
* add support for force push
* add support for push of new references
|