narrowrepo: make repo requirement include the string 'experimental' We can just move to "narrow" or similar when we finalize the format. I'm not sure what the migration process from one requirement to another should look like, but we're about to cross that bridge at Google once this change lands, so hopefully we'll know soon. Differential Revision: https://phab.mercurial-scm.org/D2007

narrow: mark requirement as a constant Differential Revision: https://phab.mercurial-scm.org/D2006

narrowbundle2: mark most constants as module-private This became obvious when preparing the previous change, so there's no reason to not just do it now while it's easy. Differential Revision: https://phab.mercurial-scm.org/D2005

narrowbundle2: make constants ALLCAPS to be a bit more readable Differential Revision: https://phab.mercurial-scm.org/D2004

narrow: remove old version-checking logic and declare internal Differential Revision: https://phab.mercurial-scm.org/D1980

tests: fold narrow treemanifest tests into main test file using testcases These tests predate the testcases functionality in run-tests.py, so it was never done in narrowhg. Differential Revision: https://phab.mercurial-scm.org/D1979

narrow: assume addflagprocessor will always exist on revlog module Differential Revision: https://phab.mercurial-scm.org/D1978

narrow: this code should assume REVIDX_FLAGS_ORDER exists Differential Revision: https://phab.mercurial-scm.org/D1977

narrow: drop legacy support for getsubsetraw Differential Revision: https://phab.mercurial-scm.org/D1976

narrow: remove support for old match modules Differential Revision: https://phab.mercurial-scm.org/D1975

narrow: import experimental extension from narrowhg revision cb51d673e9c5 Adjustments: * renamed src to hgext/narrow * marked extension experimental * added correct copyright header where it was missing * updated hgrc extension enable line in library.sh * renamed library.sh to narrow-library.sh * dropped all files from repo root as they're not interesting * dropped test-pyflakes.t, test-check-code.t and test-check-py3-compat.t * renamed remaining tests to all be test-narrow-* when they didn't already * fixed test-narrow-expanddirstate.t to refer to narrow and not narrowhg * fixed tests that wanted `update -C .` instead of `merge --abort` * corrected a two-space indent in narrowspec.py * added a missing _() in narrowcommands.py * fixed imports to pass the import checker * narrow only adds its --include and --exclude to clone if sparse isn't enabled to avoid breaking test-duplicateoptions.py. This is a kludge, and we'll need to come up with a better solution in the future. These were more or less the minimum to import something that would pass tests and not create a bunch of files we'll never use. Changes I intend to make as followups: * rework the test-narrow-*-tree.t tests to use the new testcases functionality in run-tests.py * remove lots of monkeypatches of core things Differential Revision: https://phab.mercurial-scm.org/D1974

python3: whitelist another four passing tests Differential Revision: https://phab.mercurial-scm.org/D2197

wireprotoserver: define and use parse_qs from urllib The cgi module is deprecated since Python 2.6. Let's replace uses of it in wireprotoserver with a similar function from urllib. Differential Revision: https://phab.mercurial-scm.org/D2094

tests: add tests for sending recognized command before handshake Rounding out our test coverage for the SSH server. Differential Revision: https://phab.mercurial-scm.org/D2093

wireprotoserver: add version to SSH protocol names (API) We recently introduced version 2 of the SSH protocol. Like we did for the peer, we will need to differentiate version 1 and 2 of the server. So, we add version components to the advertised name of the protocol handler. Nothing in core looks for the old value. But an extension may. .. api:: SSH protocol handler now advertises its name internally as "ssh-v1" instead of "ssh." Differential Revision: https://phab.mercurial-scm.org/D2092

wireproto: introduce type for raw byte responses (API) Right now we simply return a str/bytes instance for simple responses. I want all wire protocol response types to be strongly typed. So let's invent and use a type for raw bytes responses. .. api:: Wire protocol command handlers now return a wireprototypes.bytesresponse instead of a raw bytes instance. Protocol handlers will continue handling bytes instances. However, any extensions wrapping wire protocol commands will need to handle the new type. Differential Revision: https://phab.mercurial-scm.org/D2089

wireprototypes: move wire protocol response types to new module We'll be introducing more types as part of wire protocol version 2. These types are shared between the command handling code (in wireproto.py) and the protocol/transport code in wireprotoserver.py. So they need to go in a new module to prevent a cycle. The types are aliased into the wireproto module, so API compatibility is preserved. Differential Revision: https://phab.mercurial-scm.org/D2088

wireprotoserver: move responsetype() out of http handler This is our last public attribute not part of the protocol interface! Differential Revision: https://phab.mercurial-scm.org/D2087

wireproto: remove unused proto argument from supportedcompengines (API) In theory, the protocol should be passed to this function. But the argument isn't being used and it is getting in the way of refactoring. So let's remove it. We can always add it back later if we need it again. Differential Revision: https://phab.mercurial-scm.org/D2086

wireprotoserver: rename getfile() to forwardpayload() (API) "file" can mean a lot of things. Let's rename the interface method to something more descriptive. While I was here, I moved the docs about the payload format to the implementation of the SSH protocol, because it was lying about what the HTTP payload looked like. Differential Revision: https://phab.mercurial-scm.org/D2085

wireprotoserver: rename _client to client (API) This method is called in wireproto.py. It should be part of the public API/interface. .. api:: The ``_client()`` method of the wire protocol handler interface has been renamed to ``client()``. Differential Revision: https://phab.mercurial-scm.org/D2084

wireprotoserver: remove redirect() and restore() (API) These methods on the protocol handler interface are no longer used in core. .. api:: redirect() and restore() have been removed from the wire protocol handler interface. Use mayberedirectstdio() instead. Differential Revision: https://phab.mercurial-scm.org/D2083

wireproto: use maybecapturestdio() for push responses (API) The "pushres" and "pusherr" response types currently call proto.restore() in the HTTP protocol. This completes the pairing with proto.redirect() that occurs in the @wireprotocommand functions. (But since the SSH protocol has a no-op redirect(), it doesn't bother calling restore() because it would also be a no-op.) Having the disconnect between these paired calls is very confusing. Knowing that you must use proto.redirect() if returning a "pushres" or a "pusherr" is even wonkier. We replace this confusing code with our new context manager for [maybe] capturing output. The "pushres" and "pusherr" types have gained an "output" argument to their constructor and an attribute to hold captured data. The HTTP protocol now retrieves output from these objects. .. api:: ``wireproto.pushres`` and ``wireproto.pusherr`` now explicitly track stdio output. Differential Revision: https://phab.mercurial-scm.org/D2082

wireprotoserver: add context manager mechanism for redirecting stdio Today, proto.redirect() sets up redirecting stdio and proto.restore() undoes that. The API is a bit wonky because restore() is only implemented on the HTTP protocol. Furthermore, not all calls to redirect() are obviously paired with calls to restore(). For example, the call to restore() for "unbundle" requests is handled by the response handler for the HTTP protocol. This commit introduces a new method on the protocol handler interface to maybe capture stdio. It emits a file object or None depending on whether stdio capture is used by the transport. To prove it works, the "pushkey" wire protocol command has been updated to use the new API. I'm not convinced this is the best mechanism to capture stdio. I may need to come up with something better once the new wire protocol emerges into existence. But it is strictly better than before because it removes variance in the wire protocol handler interface. It therefore gets us closer to a unified interface between the SSH and HTTP transports. Differential Revision: https://phab.mercurial-scm.org/D2081