Gregory Szorc <gregory.szorc@gmail.com> [Tue, 18 Sep 2018 17:57:36 -0700] rev 39796
filelog: stop proxying compress() (API)
The censoring code was previously relying on this. With a dedicated
censoring API on the interface, no consumers are left and we can
stop proxying this method.
Differential Revision: https://phab.mercurial-scm.org/D4658
Gregory Szorc <gregory.szorc@gmail.com> [Tue, 18 Sep 2018 17:56:15 -0700] rev 39795
filelog: stop proxying start(), end(), and length() (API)
These were needed by the censoring code, which formerly lived in the
censor extension. Now that there is a censoring API on the file storage
interface, nothing uses these methods and we can stop proxying them.
Differential Revision: https://phab.mercurial-scm.org/D4657
Gregory Szorc <gregory.szorc@gmail.com> [Tue, 18 Sep 2018 17:51:43 -0700] rev 39794
revlog: move censor logic out of censor extension
The censor extension is doing very low-level things with revlogs.
It is fundamentally impossible for this logic to remain in the censor
extension while support multiple storage backends: we need each
storage backend to implement censor in its own storage-specific
way.
This commit effectively moves the revlog-specific censoring code to
be a method of revlogs themselves.
We've defined a new API on the file storage interface for censoring
an individual node. Even though the current censoring code doesn't
use it, the API requires a transaction instance because it logically
makes sense for storage backends to require an active transaction
(which implies a held write lock) in order to rewrite storage.
After this commit, the censor extension has been reduced to
boilerplate precondition checking before invoking the generic
storage API.
I tried to keep the code as similar as possible. But some minor
changes were made:
* We use self._io instead of instantiating a new revlogio instance.
* We compare self.version against REVLOGV0 instead of != REVLOGV1
because presumably all future revlog versions will support censoring.
* We use self.opener instead of going through repo.svfs (we don't have
a handle on the repo instance from a revlog).
* "revlog" dropped
* Replace "flog" with "self".
Differential Revision: https://phab.mercurial-scm.org/D4656
Gregory Szorc <gregory.szorc@gmail.com> [Tue, 18 Sep 2018 16:47:09 -0700] rev 39793
global: replace most uses of RevlogError with StorageError (API)
When catching errors in storage, we should be catching
StorageError instead of RevlogError. When throwing errors related
to storage, we shouldn't be using RevlogError unless we know
the error stemmed from revlogs. And we only reliably know that
if we're in revlog.py or are inheriting from a type defined in
revlog.py.
Differential Revision: https://phab.mercurial-scm.org/D4655
Gregory Szorc <gregory.szorc@gmail.com> [Tue, 18 Sep 2018 16:45:13 -0700] rev 39792
error: introduce StorageError
Errors in revlogs are often represented by RevlogError. It's fine
for revlogs to raise a revlog-specific exception. But in the context
of multiple storage backends, it doesn't make sense to be throwing or
catching an exception with "revlog" in its name when revlogs may not
even be in play.
This commit introduces a new generic StorageError type for representing
errors in the storage layer.
RevlogError is an instance of this type.
Interface documentation and tests referencing RevlogError has been
updated to specify StorageError should be used.
.. api::
``error.StorageError`` has been introduced to represent errors in
storage. It should be used in place of ``error.RevlogError`` unless
the error is known to come from a revlog.
Differential Revision: https://phab.mercurial-scm.org/D4654
Gregory Szorc <gregory.szorc@gmail.com> [Tue, 18 Sep 2018 16:28:17 -0700] rev 39791
revlog: drop LookupError alias (API)
This alias is especially bad because it shadows the built-in
LookupError type. This has caused me confusion in the past
when reading revlog code. Qualifying all uses with "error." will
make it obvious that we're using a Mercurial error type.
Differential Revision: https://phab.mercurial-scm.org/D4653
Gregory Szorc <gregory.szorc@gmail.com> [Tue, 18 Sep 2018 16:24:36 -0700] rev 39790
revlog: drop some more error aliases (API)
These should be lightly used and I doubt that will be any
strong objections to removing the aliases.
Note that some uses of ProgrammingError in this file use
translated messages. I'm pretty sure that's a bug. But the
linters don't complain, so meh.
Differential Revision: https://phab.mercurial-scm.org/D4652
Gregory Szorc <gregory.szorc@gmail.com> [Tue, 18 Sep 2018 16:18:37 -0700] rev 39789
revlog: drop RevlogError alias (API)
error.RevlogError was moved from revlog.py in 08cabecfa8a8 in
2009. revlog.RevlogError has remained as an alias ever since.
Let's drop the alias and use error.RevlogError directly.
Differential Revision: https://phab.mercurial-scm.org/D4651
Gregory Szorc <gregory.szorc@gmail.com> [Tue, 18 Sep 2018 16:52:11 -0700] rev 39788
testing: add interface unit tests for file storage
Our strategy for supporting alternate storage backends is to define
interfaces for everything then "code to the interface."
We already have interfaces for various primitives, including file
and manifest storage.
What we don't have is generic unit tests for those interfaces. Up
to this point we've been relying on high-level integration tests
(mainly in the form of existing .t tests) to test alternate storage
backends. And my experience with developing the "simple store" test
extension is that such testing is very tedious: it takes several
minutes to run all tests and when you find a failure, it is often
non-trivial to debug.
This commit starts to change that.
This commit introduces the mercurial.testing.storage module. It
contains testing code for storage. Currently, it defines some
unittest.TestCase classes for testing the file storage interfaces.
It also defines some factory functions that allow a caller to easily
spawn a custom TestCase "bound" to a specific file storage backend
implementation.
A new .py test has been added. It simply defines a callable to produce
filelog and transaction instances on demand and then "registers" the
various test classes so the filelog class can be tested with the
storage interface unit tests.
As part of writing the tests, I identified a couple of apparent
bugs in revlog.py and filelog.py! These are tracked with inline
TODO comments.
Writing the tests makes it more obvious where the storage interface
is lacking. For example, we raise either IndexError or
error.LookupError for missing revisions depending on whether we
use an integer revision or a node. Also, we raise error.RevlogError
in various places when we should be raising a storage-agnostic
error type.
The storage interfaces are currently far from perfect and there is much
work to be done to improve them. But at least with this commit we
finally have the start of unit tests that can be used to "qualify"
the behavior of a storage backend. And when implementing and debugging
new storage backends, we now have an obvious place to define new
tests and have obvious places to insert breakpoints to facilitate
debugging. This should be invaluable when implementing new storage
backends.
I added the mercurial.testing package because these interface
conformance tests are generic and need to be usable by all storage
backends. Having the code live in tests/ would make it difficult for
storage backends implemented in extensions to test their interface
conformance. First, it would require obtaining a copy of Mercurial's
storage test code in order to test. Second, it would make testing
against multiple Mercurial versions difficult, as you would need to
import N copies of the storage testing code in order to achieve test
coverage. By making the test code part of the Mercurial distribution
itself, extensions can `import mercurial.testing.*` to access and run
the test code. The test will run against whatever Mercurial version
is active.
FWIW I've always wanted to move parts of run-tests.py into the
mercurial.* package to make the testing story simpler (e.g. imagine an
`hg debugruntests` command that could invoke the test harness). While I
have no plans to do that in the near future, establishing the
mercurial.testing package does provide a natural home for that code
should someone do this in the future.
Differential Revision: https://phab.mercurial-scm.org/D4650
Gregory Szorc <gregory.szorc@gmail.com> [Tue, 18 Sep 2018 15:32:11 -0700] rev 39787
narrow: remove narrowrevlog
Core now automatically enables ellipsis support on revlogs when
repositories have narrow enabled. So, we no longer need to globally
register the revlog flag as part of activating the narrow extension
and this code can be deleted.
A side effect of this change is that repositories will now raise an
error on encountering an ellipsis flag when the narrow extension is
loaded. Previously, loading the narrow extension on a non-narrow repo
could result in silent usage of the ellipsis flag. This could lead
to undetected bugs. I think the new behavior is more correct.
Differential Revision: https://phab.mercurial-scm.org/D4649