All users already use repo.local() to test for local, which is false
in the repository base class. statichttprepository never derived from
this class anyway.
Moves ssh argument building to platform specific utils code.
The win32 version looks for plink in ssh command string and
uses '-P' in lieu of '-p' for specifying a port
We weren't reading all the data sent by the server. Depending on
the system, the remote hg (actually, the remote python) could send a
"close failed: [Errno 32] Broken pipe", making some tests fail.
The __del__ method of sshrepo reads the stderr of the remote process
until EOF and prints it.
If an exception is raised, this method ends up being called:
- on Linux: after the "abort: ..." message is printed
- on OS X: before the "abort: ..." message is printed
This patch explicitly flushes the stderr of the remote process before
raising a RepoError.
- This removes the "repo not found" error when cloning or init-ing a remote
repo.
- Since the remote hg will abort if the repo already exists we don't need to
validate it.
existing clone code uses pull to get changes from remote repo. is very
slow, uses lots of memory and cpu.
new clone code has server write file data straight to client, client
writes file data straight to disk. memory and cpu used are very low,
clone is much faster over lan.
new client can still clone with pull, can still clone from older servers.
new server can still serve older clients.
Rearranged the clone command a good bit to make sure it validates that
the source does exist and that the destination doesn't before doing anything.
Before I moved the source repo check it would create the destination
repository before it verified the source existed.
Moved the responsibility for creating the destination repo root directory
entirly into the localrepo class so that local to local cloning doesn't break.
This also simplifies the code a bit since it's no longer being done in both
clone and init.
Changed the names of the 'repo' and 'other' variables to 'dest_repo' and
'src_repo' to maintain my sanity.
Passes 82/83 tests. The only failure is the version number test, which I
suspect is supposed to fail since it comes from a generated file.
now all repositories have capabilities slot, tuple with list of names.
if 'unbundle' capability present, repo supports push where client does
not need to lock server. repository classes that have unbundle capability
also have unbundle method.
implemented for ssh now, will be base for push over http.
unbundle protocol acts this way. server tells client what heads it
has during normal negotiate step. client starts unbundle by repeat
server's heads back to it. if server has new heads, abort immediately.
otherwise, transfer changes to server. once data transferred, server
locks and checks heads again. if heads same, changes can be added.
else someone else added heads, and server aborts.
if client wants to force server to add heads, sends special heads list of
'force'.
Because older servers don't return any output for unknown commands,
it's tricky to add new commands. The approach is this: we add a
"hello" command that reports any interesting capabilities (and other
things that might be of interest in the future). To detect whether
this new command is supported, we issue both it and our startup
detection command ("between") at the beginning of a connection.
repo.addchangegroup method now returns number of heads modified and added,
so command line can tell whether update or merge needed. this makes
tiny change to ssh wire protocol, but change is backwards compatible.
pull command now returns 0 if no changes to pull.
