It should mostly still work (except for :Pyimport), but it won't pass all the tests anymore.
Since I'm dropping Python 2 in Jedi anyways, it makes sense to also remove it here.
* pythonx/jedi 3ad3dc08...3b7106ae (203):
> Fix a typo
> Prepare for 0.17.2
> Fix a Python 3.9 issue on travis
> Properly negate with Interpreter, fixes#1636
> Fix issues with generators, fixes#1624
> Introduce py__next__ to have more clear way to use __next__
> Merge pull request #1633 from mrclary/mrclary-fix-wingkinl-patch-python-environ
> Fix dict completions for inherited dicts, fixes#1631
> Refactor Interpeter completions a bit
> Make sure the interpreter completions work better in Jupyter Notebook, fixes#1628
> Fix an issue with interpreter completion, see also #1628
> Making sure to note that Python 2 will not be supported after 0.17.2
> Mention in Changelog that 3.9 is now supported
> Add download badge
> Some test skips
> Revert some of the Decoratee changes
> Some more signature adjustments
> Change Decoratee slightly
> Fix an issue with decoratee names
> Add Project.path, fixes#1622
> Fix the of a signature with a decorator
> Fix docstrings for method decorators, fixes#1621
> Make partial use the __doc__ of its function, fixes#1621
> Merge pull request #1614 from PeterJCLaw/fix-decorator-factory-passthrough
> Merge pull request #1623 from mallamanis/master
> Remove a print
> Typeshed third party libraries should not be loaded if they don't actually exist in the environment, fixes#1620
> Add the env_vars change to CHANGELOG
> Merge pull request #1619 from mrclary/subprocess-env-vars
> Make sure that Callables are properly represented
> Add Python 3.8 to the tox env list (#1618)
> Use Python 3 in the deployment script
> Prepare release 0.17.1
> Make sure the current version doesn't install a parso version that is new
> Small changelog improvement
> Mention Django Manager support for managers/querysets in changelog
> Add SyntaxError.get_message
> Mention 3.9 support better
> Fix decorator param completion
> Merge branch 'django'
> Merge branch 'master' of github.com:davidhalter/jedi
> Fix some issues with sub class matching, fixes#1560
> Better debugging output for is_sub_class_of
> Ignore runtime_checkable, because we don't really need it
> Fix a small extract_variable issue, fixes#1611
> Try to get get the tests for Python 3.9 passing, fixes#1608
> Make it possible to use inheritance on generics without always specifying type vars, see also discussion in #1593
> Update MANIFEST.in
> Make sure that scopes can only be exact values, see #1590
> Merge pull request #1590 from muffinmad/references-scope
> Fix a small issue with doctest completions, fixes#1585
> Exclude more Linux constants in test_import
> Merge pull request #1601 from yuan-xy/patch_3
> Refactoring about checking \r\n (#1603)
> fix typo (#1602)
> fix typo
> Merge pull request #1599 from isidentical/py38-plus-setuppy
> Merge pull request #1595 from PeterJCLaw/operator-not-in
> Import FileNotFoundError from jedi._compatibility
> Merge pull request #1586 from PeterJCLaw/django-more-fields
> Fix potential AttributeError in get_defintion_start_position/get_defintion_end_position, see #1584
> Fix an error of get_definition_end_pos, see #1584
> Refactor test_definition_start_end_position to use parametrize
> definition_end_position -> get_definition_end_position, same for start, see #1584
> Merge pull request #1584 from pappasam/get_definition_position
> Fix tests in Python 2
> Fix a small issue
> Magic methods fixes for reverse methods
> lists and tuples should not be added
> Use __truediv__ instead of __div__
> Fix an issue with Tuple generics
> A few class renames
> Better debugging
> Skip another non-important Python 2 test that fails on Windows
> Fix a namespace issue when getting references
> Fix an about dict completions
> Fix a None issue
> Implement magic method return values, fixes#1577
> Remove an unused cache method
> Undefined api types should not return a random value
> Add support for generic optional parameters (#1559)
> Fix inline case where a name was removed without the code being used, fixes#1582
> Make the diff nicer if there is no ending newline, fixes#1581
> Merge pull request #1579 from muffinmad/pseudotreenameclass
> Merge pull request #1572 from davidhalter/classvar
> Reduce noise in signatures of compiled params (#1564)
> Fix#1573 again; a tree_node can be None
> Fix argument clinic unpacking, remove dynamic bullshit
> This fixes two issues with the caching on Windows:
> Fix getattr completions on very weird cases, fixes#1573
> Fix tests
> Remove the common folder and move it to a common file
> Remove common.value
> Move infer_type_vars to base_value
> Merge pull request #1554 from PeterJCLaw/fix-nested-tuple-argument
> Add .venv to _IGNORE_FOLDERS
> Correct a test
> Speed up signature fetching for MixedName, see discussion in #1422
> Test that the actual signature of a function is used in Interpreter
> Very small refactoring
> Tell sith that goto_assignments is now goto
> Merge pull request #1561 from PeterJCLaw/newtype-pyclass
> Merge pull request #1553 from PeterJCLaw/generic-tuple-return
> Allow files for get_default_project, fixes#1552
> Don't display unnecessary help, fixes#1557
> Pin Django in a different way so tests can work everywhere
> Make sure that Django's values/values_list is tested (though not implemented
> Properly implement inheritance for Django models
> Make sure Django User inference works
> Pin the Django test dependency
> Mention django stubs support in README
> Include Django stubs license in Jedi package
> Some more django query tests
> Infer many to many fields
> Include Django stubs as a third party repo
> Make sure that infering the Django User model works
> Fix a recursion error issue
> Remove unused imports
> Fix typos in api.classes docstrings
> Remove a redundant check from Name.desc_with_module
> Actually fix#1556, forgot to add this in 94d374c9ceeb16ea1c6fe49b8e039d415c7dbe5a
> Start changelog for 0.17.1
> Fix a small issue with the help method, fixes#1556
> Django is not supported for Python 2
> Work with a NameWrapper, so Django goto works better
> Fix another django modelfield issue
> Some nitpicks
> Fix another issue with foreign keys
> Fix ForeignKey issues with invalid values
> Some other small refactorings
> Refactor a few things for django
> Remove old third party django tests
> Use debug instead of print for Django and fix indentation, see #1467
> Make Django test optional
> Merge branch 'master' of https://github.com/ANtlord/jedi
> Better developer tools
> Make the linter completely private
> Make sure that calling Jedi with a random argument in CLI results in errors
> Fix a small _get_annotated_class_object, fixes#1550
> Also don't complete keywords if kwargs only are allowed, see #1541
> Completion: Don't suggest variables when only kwargs are legal, fixes#1541
> fix#1548
> * Use an explicit environment for subprocess to ensure that existing environment variables are not inherited. This ensures more reliable results, see issue #1540. * Attempt to send SYSTEMROOT variable to Windows subprocess
> Sort test_project::test_search results to fix failures
> Inc difference limit in TestSetupReadline::test_import for py3.8
> Project._python_path -> Project.environment_path
> Set the release date in Changelog
> Some README improvements
> Extract now properly validates line/column and those two params are required
> Some minor CHANGELOG changes
> Upgrade parso version
> Fix a small issue
> Some more code quality fixes
> Some code cleanups
> Don't reuse a variable
> Remove unreachable code
> _convert_names kwargs are not needed
> Add pyproject.toml to the list of files to search for projects
> Use the interpreter environment if the executable is not available, fixes#1531
> Fix a potential AttributeError
> Add a comment to clarify the Type case
> Avoid AttributeErrors for generics when a module is passed
> Merge branch 'improve-type-annotation-inference-refactors' of https://github.com/PeterJCLaw/jedi
> Merge branch 'improve-type-annotation-inference' of https://github.com/PeterJCLaw/jedi
> If branch inference should not trigger for things we don't know, fixes#1530
> Fix string completion issue, fixes#1528
> Fix wrong types for iterate, fixes#1524
> Fix stub conversion for Decoratee, so docstrings work, see #117
> Remove unused environment param
> Fix a Python 2 test
> Merge branch 'refactor'
> Merge branch 'project'
> Included statement as a possible return type for BaseDefinition.type
> Fix partialmethod issues
> Make sure partialmethod tests are only executed for Python 3
> Merge branch 'add-partialmethod' of https://github.com/ffe4/jedi
> Merge branch 'expandtab' of https://github.com/Carreau/jedi
> Changed semantics of ClassVar attributes in classes, fixes#1502
> Avoid duplicate definitions for goto, fixes#1514
> Merge _remove_statements and infer_expr_stmt, fixes#1504
> Fix namedtuple docstring/signature issues, fixes#1506
> Fix a getattr_static issue, fixes#1517
> Disable some more tests for Python 2
> Make InterpreterEnvironment public
> is_typeddict should be part of ClassMixin
> Don't continue searching for values if an annotation is found
> TypedDict checking should be at a later point
> Simplify tests a bit
> Merge branch 'typeddict' of https://github.com/pappasam/jedi
> Update sith's module docstring to match the available operations
> Remove the word 'class' from annotation_string
> Add a couple of docstrings
> Tell sith that 'completions' became 'complete'
> Add venv to .gitignore
> Remove additional prefix which seems incorrect
> Update use of _source which no longer exists to _code
> Make sure to move up VSCode, because it's used a lot
> Revert "Make sure to mention that VSCode is using Jedi"
> Make sure to mention that VSCode is using Jedi
> Fix a test issue
> Actually use follow_builtin_imports and improve the goto docstring, fixes#1492
> Catch an error with illegal class instances, fixes#1491
> Make Ellipsis without list in Callable work, fixes#1475
Without this it would not go there with an already open qf window, and
would go to an unexpected buffer line instead of selecting the entry.
This is a follow-up to b689409, where it was decided that it should
always require the user to select an entry.
An exception is when the same usages are used again: then it will select
the nearest/current entry only (via ":cc").
It basically allows to give a warning when '1' is used / set explicitly, and
the 'conceal' feature is not available.
Also defaulting to '2' (cmdline mode) then seems better than not using
it.
This is just some minor clean up before
https://github.com/davidhalter/jedi-vim/pull/652.
* pythonx/jedi afbd8cad...3ad3dc08 (197):
> Run get_type_hint tests only for 3.6+
> Stubs should not become stubs again in the conversion function, fixes#1475
> Make sure to print errors in __main__ completions
> Make sure param hints are working for functions
> Generate type hints, fixes#987
> Bump version to 0.16.1
> Make sure to be able to deal with all kinds of loaders, fixes#1487
> Make generators return more correct values with while loops, fixes#683
> Avoid aborting search for yields when they are still reachable, see #683
> Jedi understand now when you use del, fixes#313
> Remove dead code
> Remove Python 3.3 from environment tests
> Make sure to skip some tests for Python 3.5
> Small Changelog updates
> Fix a small signature issue
> Create a private helper to test completions
> Fix an issue with names
> compiled_object -> compiled_value
> Remove dead code
> CompiledObject -> CompiledValue
> CompiledValue -> ExactValue
> Revisit caching of mixed
> Make MixedName a Namewrapper instead of inheritance
> Remove underscore_memoization caching method
> Refactor MixedName quite a bit
> Fix issue with mixed objects, fixes#1480
> Remove some dead code
> Prepare a test for #1479
> Avoid a print in tests
> Fix value/context mixup in mixed, fixes#1479
> Clarify that for Python 2 we will not fix bugs anymore
> Skip a test in Python 2
> Use is_compiled instead of isinstance checks
> Create CompiledModule to have a better differentiation between compiled modules and compiles values
> Make sure that the builtin docstring works again for infer calls
> Fix a bug with version_info, fixes#1477
> Make sure that del_stmt as a name can be handled, see #313
> Remove use_filesystem_cache and additional_dynamic_modules, it hasn't been implemented for a long time
> Removed settings.no_completion_duplicates
> Make sure to stop gathering buildout paths at a certain point, fixes#1325
> Rewrite some whitespace
> Reformat the changelog a bit
> Add follow_imports to Definition.goto, fixes#1474
> Raise a proper exception instead of assert in case only_stubs and prefer_stubs are given
> Make sure a certain test passes as well with tox
> Fix a Python 2 issue
> Rename Script.names to Script.get_names, fixes#1476
> find_signatures -> get_signatures, see #1476
> find_references -> get_references, see #1476
> Fix remaining tests
> Reduce limits of files to parse by quite a bit
> Make sure that CompiledObject doesn't have a file_io
> Use different limits for references and dynamic calls
> Explain why the references limits were chosen
> Start limiting opened files and parsed files for references
> Remove a few imports
> Make sure to not scan the same directory multiple times
> Don't search for usages when we are working with params
> Ignore .gitignore in get_references and therefore make get_references usable again
> Revisit reference finding, scan a lot of folders
> Added FolderIO.walk and FolderIO.get_base_name
> Move get_module_contexts_containing_name to the references module
> Apparently this one variable is needed
> Small changelog fix
> Forgot to use sudo for installing dependencies in travis
> Another few travis fixes
> Don't run some usage tests on Python 2
> Interpreter test fix for travis config
> Add a few more stub usage tests
> Disable flow analysis for finding usages
> Add another stub usage test
> Fix stub conversion
> Make sure to find both stubs and non-stubs with usages
> Use the proper fixture
> Implement interpreter test on travis
> Get interpreter environment tests working
> Make sure tests are proper packages, so that pytest doesn't do shenannigans with sys path
> Remove NestedImportModule, because it hasn't been used in years
> Refactor execute_operation a bit
> Finally get rid of call_of_leaf
> Refactor the isinstance checks a bit
> Makes sure examples are excluded from pytest
> speed test to examples
> Move not_in_sys_path tests to examples
> Move namespace tests to examples
> Moved zipped imports test files
> Move sample_venvs to examples
> Move the extension test to examples
> Move another test to examples
> Remove unused test code
> Move implicit namespace package code to example dir
> Move flask tests to examples folder
> Move the absolute import test files
> Avoid a bug that a compiler might have found, fixes#1469
> Make sure the latest changes work with Python 3.6/3.7
> Properly convert compiled values to generic classes
> Make Union/Optional works with compiled objects
> Make sure that CompiledValue can deal with string annotations
> Avoid a few warnings
> Skip a test that doesn't work in Python 2
> Remove a statement that didn't make sense
> Remove a method that was not necessary
> Fix a bug where parent_context was a value
> Fix some issues with converting names, see #1466
> Use one single way to convert stubs to Python, see #1466
> Make sure to use _stub_to_python_value_set for all conversions, see #1466
> Make sure fixture resolving works in conftest.py, see #791
> Fix signatures when used for Generic classes, fixes#1468
> Make sure inheritance works for fixtures, fixes#791
> Add a test to check if numpy tensorflow stuff is now cached, see #1116
> Make sure to check the module cache before loading a module (again)
> Add a completion cache for numpy/tensorflow, fixes#1116
> Ignore processing param names, fixes#520
> Avoid using params in tests and use get_signatures().params
> Make sure params have no name
> Make sure that kwargs are not repeated when they are inferred
> Undo finding signatures for everything and only do it for stubs and non-statements for when used in docstrings
> Fix an issue with the is_big_annoying_library function, see #520
> Make sure decorators are also not inferred for big annoying libraries, see #520
> Make sure the repr of compiled access isn't huge
> Disable some features for big annoying libraries like pandas, tensorflow, see #520
> Fix a debug message
> Fix a small issue
> Little refactoring
> Remove the old definition tests
> Remove get_signatures again from names
> Infer doctests and signatures uniformly, fixes#1466
> Fix a small pytest fixture bug
> Raise a deprecation warning on Definition.params
> Remove a weird assert
> Make sure coverage works again
> Use Python 3.7 for calculating test coverage
> Remoeve dead code
> Remove unnecessary code
> Separate getting docstrings and getting signatures for names, see discussion #1466
> LazyInstanceClassName -> Use NameWrapper
> Make sure that equals will only be added to keyword arguments and not just randomly
> Make some test code prettier
> Remove some skips around attribute docstrings
> Add some more dict completion tests with whitespace
> Change ModuleValue param order and add defaults
> Make sure to pass tests again on Python 3.4
> Make sure that include_signature always works, fixes#1466
> Add a test for #1465
> Skip dict key completion tests for Python 3.5, because it's just annoying with all the f-string stuff
> Revert "Use the root implementation for get_root_context"
> Added debug.warning to coveragerc, it's not relevant
> Remove unused code
> Use the root implementation for get_root_context
> Add github sponsor FUNDING.yml file
> Merge branch 'qa' of https://github.com/blueyed/jedi
> Get rid of most flake8 errors
> Get rid of a lot of flake8 errors
> Make sure to get completions for backticks in docstrings work, see #860
> Some more clarifications around docstrings, see #860
> Get doctest completions working, fixes#860
> Fix tests, skip more Python 2
> Make sure test prefixed functions are checked for pytest fixtures, see #791
> Make sure pytester is also used for fixtures, see #791
> Clarify a sentence around virtualenv security, see #1250
> Add the CHANGELOG entries for dict completions.
> Merge branch 'dict', fixes#951
> Make sure to mention that Jedi understands Pytest fixtures
> Fix namedtuple issues that were uncovered by the 'self' changes
> Self manipulations are now more correct, fixes#1392
> Skip pytest tests when environments is not the same one
> Make sure pytest stdlib fixtures are completable
> Make sure the monkeypatch fixture completion works
> Merge branch 'pytest', fixes parts of #791
> Make sure py__name__ and name are defined on all values
> Fix tests
> Keyword completion after ... should not work, fixesdavidhalter/jedi-vim#506
> Fix python 2 environment finalizing, fixes#1412
> Remove a function that did nothing anymore
> Make sure modules for dynamic searches are not checked twice
> Make sure classmethod param completion works better for the first param
> Make sure staticmethod params are (mostly) inferred correctly, fixes#735
> Python 2 test fixes
> Add the fix for #997 to the changelog
> Goto on a function/attribute in a class now goes to the definition in its super class, fixes#1175
> Fix a few tests
> Add support for completion even when __getattr__ is present, fixes#997
> Fix an error recovery goto issue, fixesdavidhalter/jedi-vim#962
> Add a test to make sure some renamings work always
> Refactor references: Matching more names that might be related
> Catch an additional case for get_context where the cursor is e.g. on the function name
> Add Script().get_context, fixes#253
> Fix some issues with Definition.parent()
> Test parents a bit better
> Refactor tests a bit
> Attribute docstrings work now, fixes#138
> Avoid some duplication of code
> Remove the _Help class completely
> Move the docstring checking code to the names
> Add the Script.help function, fixes#392
> Add big API changes to Changelog
> Merge branch 'api', fixes#1166
* pythonx/parso fb010f2...c864ca6 (14):
> Bump version to 0.6.0
> Make sure iter_funcdefs includes async functions with decorators, fixes#98
> Add a bit to the changelog
> del_stmt is now considered a name definition
> Forgot to increase the pickle version
> Revision on fstring issues (#100)
> Revision on assignment errors (#97)
> Add a Changelog note about dropping 2.6/3.3
> Fix a Python 2.7 issue
> Get rid of Python 3.3 artifacts
> Remove Python 2.6 grammar
> Merge branch 'rm-2.6' of https://github.com/hugovk/parso
> Make sure to limit the amount of cached files parso stores, fixesdavidhalter/jedi#1340
> Fix simple typo: utitilies -> utilities
* pythonx/jedi 005f6939...afbd8cad (208):
> Don't test Python 3.4 in tox anymore by default
> Add the release date to Changelog
> Jedi needs at least parso 0.5.2 now
> Fix a rb byte literal test
> Goto definition doesn't work on strings anymore, fixes microsoft/vscode#81520
> Make sure that sequence literals have the right generic classes, fixes#1416
> Improved Generic subclass matching
> Make sure overload checks work for TypeAlias, see #1416
> Avoid recursion in a specific example, see also #1458
> Fix simple typo: wheter -> whether (#1460)
> Completions.complete returns None for fuzzy completions #1409
> Cleanup fuzzy tests a bit, see #1409
> Add fuzzy completions to Changelog
> Merge branch 'feature_827_fuzzy_search' of https://github.com/jmfrank63/jedi
> Python 2 compatibility
> Make sure that the definition order in stubs are ignored
> Make sure that a goto on stubs even without the implementation is possible
> Fix a goto case with nested pyi files
> Catch some cases were _sqlite3.Connection was misidentified as sqlite3.Connection
> Rename goto_changes to options
> Fix: Stubs in typeshed weren't loaded properly sometimes, fixes#1434
> Files bigger than one MB (about 20kLOC) get cropped to avoid getting stuck completely
> Upgrade typeshed, fixes#1084
> Add a typeshed README
> Make sure param annotation completions work
> Make sure that you can select virtualenvs more precisely, fixes#1407
> Python 2 compatibility
> Make sure goto definitions works on with, fixes#1280
> Make sure that __getattr__ is always working with Interpreter
> qualified names can be None, so we need to handle it
> Cleanup some callbacks
> Make sure that Python 2 passes a test on more systems
> Make sure warnings are not shown if a property is executed, fixes#1383
> Fix RecursionError: global statements in modules should just be ignored, fixes#1457
> Make it possible to infer Callable TypeVars, fixes#1449
> Account for sys path potentially not being all unicode in typeshed, fixes#1456
> Make sure docstrings can always be inferred for builtins modules, fixes#1432
> Forgot to add some test files
> Get the context of a class name right, fixes#1396
> Make sure classmethod signatures don't include cls, fixes#1455
> Add a test to show that type var inference also works for tuples
> Don't use globals anymore
> Remove a duplicate method
> More docstrings
> A bit better documentation
> Refactor so typing uses BaseTypingValueWithGenerics
> Make some more classes private for inference.gradual.base
> Use _create_instance_with_generics
> The generics manager is now part of DefineGenericBase
> Remove get_index_and_execute and use something else
> _InstanceWrapper to _GenericInstanceWrapper
> Another rename for readability
> Merge GenericClass and _AbstractAnnotatedClass
> Use the generics manager for all the typing classes
> Start using generic managers, for #1413
> Start working on generic managers, see #1413
> Move iter_over_arguments to a separate file
> value_of_index -> context_of_index
> Try to prepare DefineGenericBase for a more general usage
> Make AbstractAnnotatedClass private
> Move parts of AbstractAnnotatedClass to the new class DefineGenericBase
> Formatting
> Rename two classes to make some things clearer
> Make some lines shorter
> Remove an unused InstanceArguments
> Move more stuff from gradual/typing.py to gradual/base.py
> Remove TypingName, it looks like it's not used
> Start splitting up gradual/typing.py
> Make sure Callable TypeVars are better identified, solves a part of #1413
> Remove a TODO that was implemented
> Pin colorama to a version that works for Python 3.4
> Find active conda environment and set it as default (if there is one) (#1440)
> Remove Python 3.4 test from appveyor
> Make sure overload signatures work, see #1417
> Shorten a line to < 100 chars
> Fix a tuple test
> Make sure that Tuple autocompletes properly, fixes#1389
> Avoid more Python 2
> Fix issues with interpreter completions on unittest.mock.
> Start writing the changelog
> Improve call signature detection by a lot
> Make sure we use the right context in case of goto with decorators, fixes#1427
> Python 2 still sucks.
> Make sure an assert no longer causes unnecessary trouble
> Make sure that decorator signature completion is working, see #1433
> Avoid wrong random call signature completion, fixes#1433
> Make sure py__iter__ has the right signature
> Remove probably dead code
> Python 2
> Make sure execute_function_slots and get_function_slots is defined for all instances
> Upgrade typeshed to the latest commits
> Make sure that the differences are calculated a bit more reliable, fixes#1429
> Unfortunately commited something that should not have been committed
> Fix Python 2 issues
> A CompiledInstance is not really compiled, it's an instance of a compiled class value
> Remove is_package on contexts
> Fix some more package issues
> Get rid of py__package__ from contexts
> Make is_package a function and call it consistently
> Make sure code_lines works on stubs, even if they are builtins
> Fix the MixedContext and also use MixedModuleContext
> Separate CompiledModuleContext from CompiledContext, fixes#1428
> Fix contextualizing of subscriptlist
> Fix a globals context issue, fixes#1435
> Remove Python 2 implicit relative imports feature
> A small rename of a value that is actually a context
> Merge pull request #1451 from pappasam/FIX_SHOW_SYSTEM_FOLDERS
> Add __ne__ to BaseValueSet. Might have caused issues in Python 2, see #1442
> Avoid finding submodules for compiled objects, because it's at least not implemented
> Make sure that goto on a subscript colon doesn't crash
> Fix getitem in compiled
> Make sure py__get__ is defined on all values
> Fix file name completions when file name is too long
> Fix an issue around completions in comments before strings
> Fix missing inference for typing.Type[typing.TypeVar] (#1448)
> Remove Python 3.9 dev build from travis, it's not needed
> Start to use Python 3.8 in the normal CI pipeline
> Fix the pow test for Python 3.8
> Fix inference from type comment for function parameter with dot
> Fix annotation string generated from wrong object
> Fix a small issue created in #1398
> fix static analysis test skips with latest pytest
> Fixed rST in changelog
> test: test_completion: Dynamically resolve current directory name.
> Skip Python 2 tests for some array issues
> Skip some param resolving tests in Python 2/3.4
> Add a few tests for a previous assertion failure
> Make sure a compiled instance is is_compiled
> Reenable a test
> Disable a test in Python2
> Remove a few unused imports
> Move eval_node to one place
> Avoid creating the same object twice
> Move inference_state.goto to the name and _follow_error_node_imports_if_possible away from inference_state
> Make some dynamic array variables private
> ContextualizedName -> TreeNameDefinition
> Use create_name instead of duplicated logic
> Replace obj with value
> Remove get_object, it's not needed anymore
> Implement properties properly
> Make arguments private for instance
> Refactor dynamic params a bit
> Revert "Refactor some dynamic function arguments things"
> Refactor some dynamic function arguments things
> Move some annotation inferring code to proper functions
> Remove an unneeded list cast
> SimpleParamName -> AnonymousParamName
> Make create_instance_context a lot more understandable (and shorter)
> Make sure a self variable is only defined in a function not outside
> get_first_non_keyword_argument_values is not really used anymore
> A simplification
> Fix recursion issues about dynamic param lookups and defaults work again
> Refactor search_param_names interface
> Use get_executed_param_names if get_executed_param_names_and_issues is not necessary
> Don't use get_executed_param_names_and_issues as an attribute on arguments
> Remvoe AnonymousArguments
> Refactor AnonymousInstance/TreeInstance, so that the anonymous instance doesn't have to use arguments
> Separate tree/compiled instances better
> Get rid of create_init_executions
> Implement super() properly
> Start using AnonymousMethodExecutionContext instead of the normal function execution context with arguments
> Prepare instance for AnonymousMethodExecutionContext
> Use the function execution filters with proper inheritance
> Deal with inheritance properly when dealing with function executions
> Separate FunctionExecution and AnonymousFunctionExecution
> Fix static analysis for params
> Add get_param_names to the function execution, which is needed to do some filtering
> Move the normal anonymous arguments case over to names
> Prefer annotations in SimpleParamName
> Add some dynamic inference checks for annotations
> Avoid using arguments.get_executed_param_names_and_issues
> Remove get_executed_param_names_and_issues from FunctionExecution
> Use function/arguments intead of execution
> Make FunctionExecutionContext.arguments private
> _ArrayInstance -> _DynamicArrayAdditions
> var_args -> arguments
> Fix the final issues about parameter arguments
> Fix infering of dynamic params
> Remove an unnecessary piece of code from goto
> Use Context.create_name instead of weird playing with params everywhere
> Use SimpleParamName everywhere it's needed
> Refactor params and what execution contexts need
> Fix a TODO
> Make the order of overloaded functions correct
> Get py__simple_getitem__ working on dicts that have a dict as a param, see #1385
> Move some code from SequenceLiteralValue to DictLiteralValue
> Fix usages in context of the new parso parameter include_setitem=True
> execution_allowed should be called with nodes
> Move a repr function
> Fix cases where dicts are passed to dicts and generics were not properly applied
> Small rename
> Change a test so it works with generics
> Get py__simple_getitem__ modifications working for list/dict instances
> Random objects should not be affected by list/dict modifications
> Add a way how dict setitem can be understood
> Move the dynamic module to dynamic_params
> Move the dynamic arrays code
> Add a comment about how _ArrayInstance is used
> check_array_additions -> _check_array_additions
> Remove methods that are not used
> Create separate classes for FakeSequence
> Remove _FakeArray, because it's no longer needed
> Enable a sys path test that is working now
> Reenable some tests
> Add a method implementation, that doesn't seem to be used, but it might one day be.
> Merge branch 'refactoring'
> Merge branch 'master' of github.com:davidhalter/jedi
> Bump version
* pythonx/parso c0ace63...fb010f2 (22):
> Add a release date to the Changelog
> Prepare the next release v0.5.2
> Merge pull request #93 from yangyangxcf/fstring_tokenize
> Add information about named expressions (#90) to the Changelog
> Use a lower pytest version so python3.4 is able to pass
> Use Python 3.8 on travis for testing
> Whitespace changes
> Fix: Add more cases for named expression errors, see #89, #90
> Add Python 3.8 to tested environments for tox
> Fix: Catch some additional cases named expr errors, see #89, #90
> Make sure that function executions are errors as well, see #90
> Some minor refactorings for #90
> Merge https://github.com/JarryShaw/parso into master
> Fix trailing comma error
> Remove debug print statement
> Fix typo (#84)
> Fix is_definition test
> Add include_setitem for get_defined_names, is_definition and get_definition
> Make is_definition work on setitem modifications, see #66
> ENH: update grammar for py39 (#78)
> Refactor the ambiguity tests a bit, see #70
> Make pgen2's grammar ambiguity detection handle more cases
* pythonx/jedi 35e5cf2c...005f6939 (5):
> Write the CHANGELOG for 0.15.1
> Remove forgotten debug/print in filename completion. (#1380)
> staticmethod and a few other cases might not have properly returned its signatures
> Remove reference to a file that doesn't exist anymore
> Bump version to make it clear that it's a different one than the current one
* pythonx/jedi 60415033...35e5cf2c (106):
> A small Changelog improvement
> Cleanup Changelog for the next release
> Make it possible to access functions that were inherited, see #1347
> Messed up a Windows test
> Again a small windows issue fixed.
> Need to escape the path backslash for windows slashes
> Make it possible to access properties again
> A small compatibility fix
> Fix an issue with type vars that might have been a problem for other things as well
> Don't add quotes after paths if they are already there
> A small speed optimization that helps a lot with sys.version_info >= (3, 0) patterns in typeshed
> Remove an unnecessary isinstance usage
> Make CompiledValue lazy
> A python 2 fix
> Fix an error that occured because of some refactorings
> Fix one more os.path.join issue
> Fix more issues with os.path path completion
> Make join detection much easier
> Fix some remaining issues with file path completions
> A few more tests for path completions (join)
> Fix quote completions for os.path.join path completions
> Fix first param argument of os.path.join file completions
> Fix arglist/trailer issues
> in os.path.join completions, directories should not end in a slash
> Make some file path completions in os.path.join work
> Fix os.path.join static value gathering
> os.path.sep should always have a clear value
> Fix string name completion for stuff like dirname and abspath
> Fix bytes issue with file path adding
> Python 2 compatibility
> Make __file__ return the correct value
> Fix string additions when used in certain ways
> Make string additions work for file path completion
> Fix interpeter project path
> Fix the REPL completer for file path completions
> Fix a few more file name completion cases
> A first iteration for file path completions
> Fix a test failure
> Add Definition.execute, fixes#1076
> Add tests to show that #516 is not working, yet
> List possible Definition.type in its docstring, fixes#1069.
> Start writing the Changelog for 0.15.0
> 3.8-dev should not be allowed to fail
> Skip more Python 2 tests
> Skipping of tests was done the wrong way again
> Skip the right tests
> Parameter.kind is not avaialble in Python 3.5
> Fix Python 2 test issues
> Add a few docstrings to make some things clearer
> Add ParamDefinition.kind, fixes#1361
> Make it possible to format a param to a string, fixes#1074
> Add another test for signature annotations
> Add Signature.to_string() with proper tests, fixes#779, fixes#780
> Test infer_default
> Use a helper to create definitions
> Add a execute_annotation option to infer_annotation
> Test infer_annotation
> Make it possible for users to infer annotations/defaults
> A first iteration of adding signatures to the API, fixes#1139
> Fix a caching issue
> Forgot the right resolve_stars parameters in one place
> Cache getting resolved param names
> Turn around resolve_stars, it shouldn't by default be resolved
> Remove a class that is not needed anymore
> Revert "Use __str__ instead of to_string"
> Fix the last remaining issues with function signature
> overloaded_functions should be private
> Use __str__ instead of to_string
> Better repr for CallSignature
> Fix some minor signature issues
> Remove the unused function signature_matches
> Fix signatures for __init__ calls when used with supers, fixes#1163
> Fix some issues for args resolving in method calls
> Remove a bit of code that i sprobably unused
> Move get_signatures from Function to FunctionMixin
> Merge the signature changes
> Remove an import
> Small refactoring
> Skipped Python 2 Interpreter tests the wrong way
> Remove a paragraph in docs that was arguing that stubs and generics (and other things) were not properly supported, fixes#1012
> Fix partial signatures for MixedObject
> Fix partial signatures, fixes#1371
> Make sure with a test that the staticmethod signature is also correct
> Fix classmethod signature, fixes#498
> Deal with annotation on *args and **kwargs correctly, fixes#980
> Use LazyContextWrapper more
> Very small refactoring
> Fix for failing assertion on native modules Issue #1354 (#1370)
> Skip numpydoc tests for Python 2
> Remove code that didn't mean anything
> Fix docstrings for wrapped functions, fixes#906
> Fix an isue about dict ordering in Python before 3.6.
> Make it clearer when get_param is used.
> A small dataclass refactoring
> Evaluate annotations for dataclasses when infer is called on param
> Get inheritance of dataclass right
> A first working iteration of dataclass signatures, fixes#1213
> Remove an unused function
> Dataclass progress
> Add enum changes to changelog
> Fix a recursion error about getting metaclasses
> One test about recursion issues only applied to Python 2
> Use class filters in instances differently so metaclass plugins work, fixes#1090
> First step in working with metaclasses in plugins, see #1090.
> Refactor the plugin registry
> Remove classes from plugins and use decorators instead
* pythonx/jedi 454447d4...60415033 (54):
> Prepare the v0.14.1 release
> Use fixture names everywhere
> Cleanup a test file
> Use parametrize instead of TestCase
> Remove a TestCase class usage
> Use pytest.mark.parametrize for something instad of a class
> Use the names fixture even more
> Use the names fixture more
> A small rename
> Refactor names tests
> A small refactoring
> Make sure __wrapped__ works properly when using an Interpreter, fixes#1353
> Small refactoring
> Make API param names appear without leading double underscores, fixes#1357 again
> Fix some call signature tests
> Make expected index work in Python 3
> Test both closing brackets and non-closing brackets for CallSignature.index
> Fix a case with errors
> Move some code around
> Fix some kwargs cases
> Fix some *args issues
> Deal better with some error nodes
> Delete unused code
> Fix some error node handling for call signatures
> Fix CallSignature index for a looot of cases, fixes #1364,#1363
> Use recursion rather than other stuff
> Even more refactorings
> small name refactoring
> Refactor some call details
> Disable a test for Python 2
> Fix pow() signature, fixes#1357
> Merge branch 'master' of github.com:davidhalter/jedi
> Fix signature to_string
> Fix an issue with keyword params, fixes#1356
> Simplify some test code for param defaults, see #1356
> Merge branch 'function_signature_in_interpreter' of https://github.com/linupi/jedi
> Improve a bit of dataclasses support, so at least the attributes can be seen
> Make sure in tests that pep 0526 variables are also able to be used when using self, see #933
> Get typing.NewType working (#1344)
> Cannot use pytest 5 yet
> Fix filters for classes and functions
> Fix some small things to get more tests passing
> A better repr for Definition
> Merge branch 'master' of github.com:davidhalter/jedi
> Fix ClassVar filter for instances
> Keyword completions are no longer possible directly after a number, fixes#1085
> Fix super call goto for multiple inheritance, fixes#1311
> goto should always goto definitions, fixes#1304
> Fix star imports checks, fixes#1235
> Actually import IsADirectoryError
> Add 3.8 to supported versions
> Fix project search if a directory is called manage.py, fixes#1314
> Return annotations for compiled objects now help to infer
> Update parso requirement
* pythonx/parso 59df3fa...c0ace63 (9):
> For Python 2.7 and 3.4 pytest 5 doesn't work anymore
> Prepare the 0.5.1 release
> Fix name tokenizing for Python 2
> Fix positioning for names that are interleaved with error tokens
> Merge branch 'master' of github.com:davidhalter/parso
> Fixed name tokenizing issues for tamil characters, fixes davidhalter/jedi#1368
> Document that the cache uses pickle files
> Fix changelog dates, fixes#77
> Try to use collections.abc.Mapping instead of collections.Mapping
Update submodule pythonx/parso a2b153e...f1ee761
* pythonx/parso a2b153e...f1ee761 (6):
> Release of 0.3.3
> Rename a test
> Remove some unused code
> Fix working with async functions in the diff parser, fixes#56
> Fix bytes/fstring mixing when using iter_errors, fixes#57.
> Improve a docstring, fixes#55.
Update submodule pythonx/parso 930ec08...a2b153e
* pythonx/parso 930ec08...a2b153e (91):
> Upgrade the Changelog
> Escape a backslash properly
> Bump the parso version to 0.3.2
> f-string parts are also PythonLeaf instances
> Fix get_start_pos_of_prefix
> Remove dead code
> Refactor f-string support
> Fix: End detection for strings was mostly wrong, fixes#51
> Fix f-string escapes, fixes#48
> Tokenizer: Simplify end of string regexes
> Add a comment from the Python3.7 code base
> Fix diff parser: The previous fix was a bit off
> Change fuzzer: Add ways to not always use correct parse input
> Fix diff parser: issue with opening brackets
> Async doesn't work in 3.4
> Skip some tests for Python2.6 and Python3.3
> Fix diff parser: DEDENT as error leaves should also be ignored and reparsed
> Fix diff parser: error dedents in between nodes should be ignored for now when copying
> Fix diff parser: Don't copy error nodes/leaves in the beginning, leads to strange issues in some cases
> Diff fuzzer: Create a check to see if the errors make sense.
> Fix diff parser: Don't copy DEDENT tokens at the beginning
> Fix diff parser: Skip last leaves for last line offset leaves
> Fix indentation error tokens
> Fix diff parser: Avoid side effects for prefix
> Fix tokenizer: backslashes sometimes led to newline token generation
> Make lines easier copyable in the fuzzer
> Fix tokenizer: Carriage returns after backslashes were not properly handled
> Fix in diff parser: prefix calculation was wrong when copying nodes
> Replace --print-diff with --print-code
> Fix diff parser generation for empty files
> Docopt should not be needed for tests
> Fix tokenizer: Form feeds and multiline docstrings didn't work together
> Fix diff parser: Need to calculate the prefix for the diff tokenizer better
> Remove copied nodes stuff, to simplify some things
> Fix newline endings and a few parser/copy counts
> Feature: The diff parser fuzzer is now able to use random Python fragments
> Fix: The Python 3.8 grammar did not include f-string support
> Fix tokenizer: Closing parentheses in the wrong place should not lead to strange behavior
> Fix end positions with error dedents
> Fix diff parser tests for Python 2
> Randomize the fuzzer a bit more with inserting characters
> Simplify the regexes
> Fix a regex clause that was totally wrong
> Fix another tokenizer issue
> Apply \r changes in syntax trees
> Actually make \r usable
> Fix split lines for Python code
> Add \r as a valid linebreak for splitlines
> Move split_lines tests around
> Start generating really random strings with the fuzzer
> Fix tokenizer: Dedents before error tokens are properly done, now.
> Fix diff parser: Also check async with
> Make the diff parser use a lot of different files by default
> Fix diff parser: Forgot that with statments are also flows
> Fix valid graph asserting for some dedents that are errors
> Fix diff parser: If funcs are not copied, errors shouldn't either
> Better documentation for the fuzz diff parser script
> Rename a test
> More verbose output for the diff fuzzer
> Fix diff parser: error dedent issues
> Fix diff parser: Avoid indentation issues
> Use the right diff order in debug output
> Fix diff parser: The prefix was wrong in some copy cases
> Make fuzz parser compatible with Python 2
> Make it possible to print all diffs in fuzzer
> Introduce a redo flag 'only_last' to narrow down issues
> Add docopt to testing dependencies
> Rewrite the fuzz diff parser to cache errors (so we can re-run those)
> Fix diff parser: Decorators were sometimes parsed without their functions
> Fix tokenizer: Dedents should only happen after newlines
> Usability for diff parser fuzzing
> Fix for diff parser: Rewrite prefix logic and don't mutate prematurely
> Fix diff parser: positioning of functions if decorators were removed
> Ignore ERROR_DEDENT in graph validation
> Add a small diff parser fuzzer
> Fix diff parser: get_last_line was sometimes wrong
> Fix error reporting order for diff issues
> Clean up the test diff parser file
> Avoid nasty side effects in creation of Node
> Make parso errors even more informative
> Fix diff parser get_last_line calculation
> Fix graph asserting for error indents
> Better checks for checking valid graphs
> Create better ways for debugging the diff parser
> Fix diff-parser: Copying parts of if else should not lead to the whole thing being copied
> Reenable diff parser parser counting in all tests
> Fix: Make the NodesStack to a NodesTree
> Fix for diff parser : Make sure that start_pos are growing always The problem was that functions/classes were sometimes not well positioned. Now all diff tests are ensuring that leaves always grow.
> Add a check to see if leaves have the right start positions
> Add parso version to an exception
> MNT: add grammar for python 3.8
* pythonx/jedi a79a1fb...bd1010b (91):
> Create a new 0.13.1 release
> Fixed completions of global vars and tensorflow slowness, fixes#1228, #1116
> The changelog date was wrong
> Release notes for 0.13.0
> Fix tensorflow issues with a few hacks (temporary), fixes#1195
> If the VIRTUAL_ENV variable changes, need to reload the default environment, fixes#1201, #1200
> Return SameEnvironment as a default, fixes#1226, #1196
> Document that using the REPL autocompletion is only available on Linux/Mac, fixes#1184
> Fix an issue with f-strings, fixes#1224
> Ignore some errors that are happening when the Python process ends and its subprocesses are cleaned up
> also remove crashes with pep 448 unpacking of lists and sets
> Fixed crash (and now recognises correctly) {**d, "b": "b"}["b"]
> Recognize {**d} as a dict instead of set
> Add extras_require=testing
> _get_virtual_env_from_var: use safe=False
> tests: venv_path: use session scope
> jedi/api/environment.py: minor flake8 fix
> jedi.api.environment._SUPPORTED_PYTHONS: add 3.7
> Removing a test from doctests, becaues it shouldn't be one
> Refactor argument clinic usage
> With the recent changes one performance optimization got lost
> FunctionExecutionContext should use the parent if possible
> Use anonymous instance arguments in a different way
> Move some anonymous instance function execution stuff
> Get rid of InstanceFunctionExecution, because it's really not needed
> Use the InstanceArguments for super as well
> Use InstanceArguments directly and not via InstanceFunctionExecution
> Remove old garbage code
> Don't use arguments that are not needed
> Also move the remaining get_params to get_executed_params
> get_params -> get_executed_params where possible
> Subprocess error reporting improvements
> Fix a recursion issue about compiled objects
> Use a CompiledInstanceNameFilter that wraps the class name as well
> Prefer Python 3 import over 2
> Now it's actually possible to specify a pytest environment for the same Python version
> Note that Python 3.3 support was dropped in Changelog
> Drop support for EOL Python 3.3 (#1019)
> Rewrite the pyc test
> Fix an issue with stderr debugging of subprocesses
> stderr of the child processes should be printed in debug output
> Use close_fds for posix.
> Remove some redundant code
> Use names of classes to infer names of instances
> Don't have execute and execute_evaluated on name
> Fix an issue where __ prefixed variables where not hidden when accessed from a class
> Bound methods are now working correctly in all Python versions. Therefore a test was wrong.
> Remove a print in tests
> BoundMethods now have access to the function that they are using
> Remove another usage of is_class where it's not needed
> FunctionContext should be created from a unified interface
> Don't create a FunctionExecutionContext if it's not used.
> Use TreeContext in a good way
> Fix broken link in documentation
> The implicit namespace package test from 4b276bae87a3170672f7ddb3e00f5851fe24d562 can only be used for Python 3.4+
> The import resolution for namespace packages was wrong
> Add a way to use the interpreter environment for tests
> MergedExecutedParams -> DynamicExecutedParams
> Fix a recursion error, fixes#1173
> Remove two recursion tests again that will belong into a commit at a point where it is not failing anymore
> Don't merge params if it's just one param
> Add a repr for AnonymousArguments
> Fix a doctest
> Some minor flake8 fixes
> Script.__repr__: include environment
> Add params to CallSignature.__repr__
> Environments are now always created on request
> Improve Environment
> Add max line length 100 to the config for flake8
> tests: use monkeypatch.setenv
> Don't test Python 3.3 on appveyor anymore, it's getting really hard to get all the right dependencies for it, because 3.3 is deprecated everywhere.
> Mention that it's ok to have a line length of 100 characters in our files.
> Don't use invalid escape sequences in regex, see https://github.com/davidhalter/jedi-vim/issues/843
> subprocess: listen: exit normally with EOFError
> Script: improve ValueError for column
> The sys path might be lazy or not in a venv
> The encoding parameter should be used again (includes test), fixes#1167
> Remove source_encoding from documentation (see #1167)
> Actually use the fast_parser setting again
> Use an import name list as long as possible
> Different _load_module API
> Directly load modules if possible, with this it's not necessary anymore to use dotted_from_fs_path, also fixes#1013.
> Don't mutate the sys.path. This is pretty nasty bug that fixes#1148
> Don't follow builtin imports anymore by default when follow_imports is on (goto)
> Add a changelog entry for the include_builtins change
> additional_module_paths in usages never actually worked
> Add include_builtins to usages, fixes#1131.
> Finally make it possible to use auto_import_modules for packages
> Test auto_import_modules in a very basic way
> Fix a bug about fstring completion
> Comprehensions are also possible arguments. Fixes 1146
The main motivation is to being able to use normal/other pytest based
tests.
Therefore this moves everything from conftest into test_integration
itself.
When being initialized via ftplugin `set shortmess+=F` (default in
Neovim by now) causes the additional information (the full traceback) to
be silenced.
This patch uses `:unsilent` to always display it.
Ref: https://github.com/neovim/neovim/issues/8675
I am seeing `p` twice for os.path.dirname, which seems to come from
Lib/posixpath.py and Lib/ntpath.py, as can be seen with `os.path.join`,
where I still see two with this patch:
```
(path, *paths)
import os (a, *p)
os.path.join()
```
It uses cterm=reverse, which is supported by current Neovim, and at
least for xterm and rxvt-unicode reverse and standout are the same.
See `infocmp`:
> smso=\E[7m / rev=\E[7m
a79a1fb Merge branch 'parso'
58141f1 Don't use requirements for now, and use the git version instead in tox
e0e2be3 Add a better comment about why people need to upgrade parso
1e7662c Prepare release of 0.12.1
68974ae Don't use internal parso APIs if possible
c208d37 Remove code that is no longer used, because parso was refactored.
3847406 Make jedi work with the next parso release
95f835a Force unicode when listing module names
282c6a2 Use highest possible pickle protocol
ea71ded Include stderr with "subprocess has crashed" exception (#1124)
106b11f Set stdout and stdin to binary mode on Python 2 and Windows
f9e90e8 Use system default buffering on Python 2
197aa22 Use cPickle on Python 2 if available
e96ebbe Add tests for DefaultDict support.
55941e5 Add support for DefaultDict on jedi_typing.py.
ff4a773 Parse correct AST attribute for version
70c2fce Replace distutils.spawn.find_executable with shutil.which
5dab97a Add an error message, see also #1139.
e2cd228 Dict comprehension items call should now work, fixes#1129
c1014e0 Fix flow analysis test
62a3f99 Fix a wrong branch check, fixes#1128
6ebe3f8 Drop 3.3 tests from travis
50812b5 A simple yield should not cause an error, fixes#1117
d10eff5 Travis: report coverage also to codecov.io
6748faa Fix _get_numpy_doc_string_cls: use cache
fc14aad Fix namespace autocompletion error
3c909a9 Travis: remove TOXENV=cov from allowed failures
b94b45c Environment._get_version: add msgs with exceptions
a95274d None/False/True are atom non-terminals in the syntax tree, fixes#1103
8d48e74 When searching submodules, use all of __path__, fixes#1105
9149956 Specially crafted docstrings sometimes lead to errors, fixes#1103
ba96c21 Follow up from the last async issue, fixes more related things about #1092.
8494164 Fix an async funcdef issue, fixes 1092.
4075c38 In some very rare cases it was possible to get an interpreter crash
because of this bug. Fixes#1087
0bcd170 Start using our own monkeypatch function for some things
ceb5509 Include function return type annotation in docstring if it is present
88243d2 Don't catch IndexError where we don't have to
5f37d08 Extend create_environment to accept an executable path
aa6857d check_fs: handle FileNotFoundError
* pythonx/parso 6f385bd...e6bc924 (44):
> Use a setuptools that still supports py33
> 3.1 release notes
> Use one simple functions to check for funcdefs in diff parser
> Make sure to treat async funcdefs the same way as normal funcdefs
> Fix an issue in the diff parser
> Asterisks in function definitions may be at the end of a func without a comma, fixes#44
> Remove some TODOs that were fixed
> Error recovery should not match the whole line in case of an invalid token, fixes#40
> docs: enable searchbox
> Merge branch 'pgen'
> Merge branch 'master' of github.com:davidhalter/parso
> Add docstring
> Add some comments
> Use a set instead of dict if it's not necessary
> Simplify some code
> Some refactorings for simplicity
> Use NFAArc as a class
> Some more renames
> Another rename
> Rename
> Move the grammar parsing to a separate module
> More renames
> Rename
> Separate generating dfas from parsing
> Cleanup some names
> Remove some code that is not used anymore
> Some cleanups and documentation
> Move out some more functions out of classes
> Some more refactorings for clarification
> Factor out start_symbol into a better position
> Refactor out dfas
> Document pgen grammars a bit better
> addarc -> add_arc
> Refactor some things in pgen
> Move some initializations
> Change license again. The year shouldn't matter
> Make first private
> Change copyright years
> Remove some code from the grammar
> Separate the grammar generation from the grammar parsing
> In pgen now everything is named grammar and not c
> Better recovery for online classes and functions
> Fix one-line error recovery for all things that are using a suite
> Add a debug function for first tokens
* jedi_vim.jedi_import_error: add location
This is useful for debugging. It contains e.g. the path to parso, if
importing failed from there.
Example:
> Error: jedi-vim failed to initialize Python: jedi#setup_python_imports: could not import jedi: cannot import name 'PythonTokenTypes' (in /…/jedi-vim/pythonx/jedi/jedi/api/completion.py:1). (in function jedi#init_python[3]..<SNR>44_init_python[27]..jedi#setup_python_imports, line 37)
* init: handle jedi_vim.jedi_import_error in Vim plugin
Using `set shortmess+=F` would suppress the `:echom` used in
`jedi_vim.no_jedi_warning` [1].
This patch makes `jedi#setup_python_imports` handle the error instead.
1: https://github.com/neovim/neovim/issues/8675
* Revisit error handling with loading jedi_vim
* jedi#debug_info: display parso submodule separately
* Fix jedi#reinit_python
* fixup! Revisit error handling with loading jedi_vim
* display_debug_info: handle exceptions with environment.get_sys_path
* fixup! Revisit error handling with loading jedi_vim
[ci skip]
It was removed previously, since it now refers to the environment being
used.
This brings it back for now.
There could be a separate var for it later, and/or it needs changes
later (or rather the environment needs a new variable name, since it
should refer to the executable really) - but this is good for now I think.
Fixes https://github.com/davidhalter/jedi-vim/issues/833
* pythonx/parso 579146b...6f385bd (7):
> Not testing Python 3.3 anymore on travis. It seems to be broken
> Add a changelog for 0.2.1
> Push the version
> The diff parser was slighly off with prefixes, fixes#1121
> A better repr for the endmarker
> Merge branch 'master' of github.com:davidhalter/parso
> Fix a prefix issue with error leafs.
- cache current environment
- s/jedi.get_python_environment/jedi.get_system_environment: the former
does not exist (renamed in https://github.com/davidhalter/jedi/commit/336087f)
- improve error display
This basically corrupts VIM's Python, because it might be having a differnet prefix (Python 3.6) than sys path. I don't know who came up with this - probably it's just how Python loads venvs - but it's definitely not good. Fortunately we can just reset it and be happy.
Unlisted buffers might come e.g. from `set viminfo+=%`, and Jedi would
crash on them.
This could additionally check for them to be readable, but this should
be handled in Jedi itself anyway.
* pythonx/jedi 02f238c...bd7c65d (413):
> Finally fix all the get_system_environment issues
> Fix a test
> There's really no bin/activate needed for an environment to work
> Document get_sys_path and change the signature of get_system_environment a bit
> Rework the time cache.
> Cache default environment
> Require parso 0.2.0 at least
> Set a release date
> Don't install the latest pip version anymore in appveyor
> Fix some references
> Mention Virtualenv support in readme and features
> Title case for Mänu :)
> Better overview over functions in the documentation
> A small docs correction
> Remove the old static analysis stuff. It was never really used
> Fix a lot of old docs code that doesn't exist anymore
> Remove the old parser documentation (that's now part of parso)
> Don't call it the plugin API anymore, that's confusing
> Reorder some functions
> Adda lot of environment documentation to sphinx
> find_python_environments -> find_system_environments
> A small refactoring
> Remove from_executable, were not really using it, yet.
> For the second time in a row it's called creationflags not creation_flags
> Refactor something small
> Remove an unnecessary else
> CREATE_NO_WINDOW was introduced in Python 3.7 and didn't exist before
> The SameEnvironment should not load by default if it's a portable
> Do binary comparisons to get virtualenvs working and not just venvs
> Use the correct parameter name for creation flags
> Add a way to generalize Popen
> Check the windows environments in a better way
> Fix some issues with the latest changes
> DefaultEnvironment -> SameEnvironment
> Better docs for Environment
> Some docstrings
> Make the Environment clearly non-public
> Make some names public
> Make sure Windows environments are safe
> Fix the module_name issue again
> Write a test for #1079 to avoid a regression in the future.
> Cleanup some of the module cache stuff
> Do not cache unimportable compiled module (#1079)
> Merge branch 'master' of github.com:davidhalter/jedi
> Multiple inheritance completion in Python 2 did not work
> Give the run.py script an environment parameter
> Fix an import
> Merge the environment changes for Windows
> Better tests for venvs
> Write a test for venvs
> Obviously Python 3 syntax cannot be used in Python 2
> Update the changelog about f-strings
> Use the latest parso version from master.
> Better goto definition for fstrings
> Fix an issue around the new grammar
> Fix some fstring issues for now
> Make a few modifications to always use the latest environment available.
> Better tests for fstrings
> With the changes in parso, f-strings are now completable
> With the changes in parso, f-strings are now completable
> Fix lambda dynamic param searches, fixes#1070
> Cleanup the namespace lookups so that it also works for Python 3.7
> Add a note about the fixed Windows tests in the changelog
> Add micbou to AUTHORS
> Add a badge for AppVeyor. Running tests for Windows
> Add comment explaining why test_versions is disabled on Windows
> Add JEDI_TEST_ENVIRONMENT_EXECUTABLE for AppVeyor
> Add AppVeyor configuration
> Improve venv_and_pths test
> Properly raise broken pipe exception
> Remove SourceLair from products, because it's a paid product
> Add a note to the readme.
> Fix another windows issue
> Merge branch 'master' of github.com:davidhalter/jedi
> Merge the windows fixes
> Some code cleanup
> Fix another error that surfaced in pandas
> Fix an issue with docstrings that contain errors
> Upgrade the wx widgets paths
> Somehow forgot about subscriptlist. Just ignore those for now.
> Fix the first issue in #1010
> colorama should always color, even if it's not a shell
> os.path.join completion speed test is sometimes slow, so give it a bit more of time
> Fix some code_lines issues
> Remove stdout/stderr from subprocesses (redirected to /dev/null)
> Add the module_path again
> Clean up the namedtuple test for #1060
> All modules now have a code_lines attribute, see #1062
> The release date for 0.12.0 should not be set, yet. See #1061.
> And now remove a pep0484 function that is no longer needed
> Some more code to a function
> More rewriting of the pep0484 logic
> Start moving some of the pep0484 comment code around
> Cleanup some instance stuff
> Better docs for funciton annotations
> Fix an error in param comments
> Catch parser errors instead of error recovery when splitting param comments
> Merge branch 'function_comment' of https://github.com/wilfred/jedi into mypy-comments
> Fix a few version issues in tests
> Use inspect.Parameter.kind for better differentiation between param types
> Fix REPL completion param name completion
> Add keyword argument test for #292
> Fix an interpreter test in Python 2
> Clean up zombie subprocesses, fixes#1048
> Fix an issue around __dir__ in the interpreter
> Fix doctest for replstartup
> Fix shell completion issues and documentation
> Merge branch 'master' of github.com:davidhalter/jedi
> Fix issues with default parameters in functions and classes
> Correct mistakes of lambda names
> eval_element -> eval_node
> Use the correct class for params when used in names. Fixes#1006
> Get rid of a regex warning, where escaping was not properly used in a normal string
> Get rid of the imp library import in Python3 to avoid warnings, fixes#1001
> Don't need the tests from #122
> Refactor the namespace package tests
> Add a failing test for nested PEP420 namespace packages
> Include __init__.py files in search for the project directory, fixes#773
> Add pytest cache to gitignore
> Improve some documentation/a failing doctest
> Fix the recursion error with globals
> Always pop nodes in recursion detector
> Globals should be looked up with the same priority as other defined nodes.
> fix set.append syntax error
> Add a few bits to the changelog
> Autocompletion in comments should at least not fail
> Fix the last async issue
> Fix a few numpydocs tests
> Fix some more await things
> Fix a yield from test
> AbstractIterableMixin -> IterableMixin
> Support generator returns when used with yield from.
> Show in a test that something doesn't work properly around async analysis
> A few documentation improvements
> Try to make the whole Builtin overwriting more abstract
> Fix a small remaining issue in Python 2
> Reorder tests to make the async stuff pass on all python versions
> Fix for all python versions
> Getting more edget cases work in 3.6 for async
> Fix an issue with async for
> Use the await method properly and just use it instead of some crazy things
> Refactor the way builtins can be overwritten by jedi's own contexts
> Move all the asynchronous contexts to a separate module
> Fix the merge issues. Now async stuff should at least partially work
> Merge the async branch
> Get rid of an unused import
> Add the name always to the script module
> Add the travere_parents function to a utility directory
> Fix Python 2 old-school relative imports
> Use the project path as a prefix, because many times it's used as a higher priority than other stuff
> Move the buildout_project stuff to a separate examples folder
> Rename of buildout stuff
> Small rename
> Use generators instead of complicated return of lists
> Some more sys path corrections.
> Generalize the use of smart import paths
> Better handling of smart sys path
> Some simplifications
> Add comments to implicit namespaces and fix some minor things.
> Merge the implicit namespace improvement (pkgutils.itermodules modification)
> Relative imports should be working again even when used in more special occasions. Fixes#973
> parent_module is not needed for loading modules
> Try to avoid CachedMetaClass for modules
> DOn't check the parser cache, that's parso's responsibility
> Better module loading
> Try to put all module loading in one place including namespace packages
> Unify load_module access
> Prepare a test to eventually solve a relative import problem
> Add a module cache that has a bit more capabilites
> print_to_stderr needs to be used with one argument
> Print errors that happen when importing certain objects
> Use unicode literals, to avoid potential issues
> Merge branch 'master' of github.com:davidhalter/jedi
> Fix some more dict.get/dict.values stuff
> Implememnt dict.values for FakeDict to avoid a recursion error. Fixes#1014.
> Add better error reporting
> Use a better string
> Add better debugging for an assert, see also #1010
> Make some things clearer about lazy contexts
> Finding the autocompletion stack is a bit more complicated than I initially thought
> Move the ahead of time tests to the pep0526 file
> Fix an issue where a default value was wrongly used
> Update the ahead of time tests
> Fix the last remaining issues with ahead of time annotations, see #982
> Fix a bug related to a wrong parametrization at one point
> Merge branch 'master' of https://github.com/johannesmik/jedi
> Force unicode on django paths
> Forgot to add the examples folder
> Fix a goto_assignments issue with a better internal API
> Merge remote-tracking branch 'origin/master' into virtualenv
> Disable predefined name analysis (if stmts) for all non-analysis tasks
> Remove something that obviously never happened
> Reorder some open flags for Python 2
> inspect.signature throws weird errors sometimes, just make it a bit simpler
> Fix the implicit namespace test
> Undo most of the namespace changes and use module again
> Differentiate between namespace and module as a type
> Use Scripts for virtualenvs instead of bin for windows
> Refactor django path support
> Fix project path finding
> Add a repr for Project
> Use the PathFinder, because the FileFinder doesn't work without suffixes
> Create the importer stuff Python2.7 and 3.3
> Trying to use the import machinery to import jedi/parso in python3.4+
> Better reporting of internal errors
> Better error handling for subprocesses
> Another merge with master
> Get rid of a cwd to tmpdir, because with the subprocess it doesn't behave the same depending on which tests you run first
> Hopefully the last merge with master
> Merge with master again
> Merge with master
> Get rid of some potential issues when using pandas interactively
> Clarity
> Remove the get_default_project caching
> Add a comment about why the project is None in the subprocess
> evaluate.project doesn't exist anymore. Eliminated code that used it
> At the moment, don't allow projects as an input to script
> Remove the old project
> Make the new project API fully work in tests
> Try to migrate to the new project API
> Refactor to make configuratios of sys paths easier
> Use the evaluate.project sys path stuff for api.project
> Cache the default project
> A possible introduction for projects
> Check for safe and unsafe environments when searching for them
> Sith still used NotFoundError which doesn't exist anymore in jedi
> Ignore pypy in travis for now
> A path to ignore in coveragerc was wrong
> Allowing the cov tests did not properly work. Trying again.
> Somehow previously removed the allowed failurs of TOXENV=cov
> Try to put env variables on one line
> Coverage was unfortunately excluded
> Merge with master
> Add a comment to refactoring that it's not in active development
> Change coveragerc a bit
> Use Python3.4 for coverage.
> Correct the issue about has_zlib
> Move the zip tests to the environment
> Only execute the zipimport tests fully if zlib is available for the environment Python.
> Some tests that involved jedi were actually a bit wrong and only worked in certain environments.
> Skip some 3.3 tests for travis
> Skipping was done wrong
> Try a bit more if modifying the PATH is now possible.
> A comparison was wrong
> A new version of the travis install script
> Try to use the virtual env that was defined in the VIRTUAL_ENV variable, if possible.
> If a subprocess gets killed by an OOM killer or whatever it should respawn and raise an InternalError
> Merge Subprocess and CompiledSubprocess
> Try to recover from errors that are happening in subprocesses
> Move the default environment around
> More travis trying
> Better debugging for travis
> Forgot to add the executable bit to the travis installer.
> Experiment with travis and installing packages differently
> The tags should be annotated if possible
> Remove 2.6 from travis
> Correct the travis file
> Some refactoring to finally get tests working with py27 and 3 environments
> Use the parser to check for certain namedtuple features
> Use unicode strings in test to pass some tests in Python 2
> Use Script everywhere where cwd_at is used, otherwise Python 2.7 is annoying
> The given sys_path gets converted to unicode now in py2
> Try to avoid the pth tests not working because of the created virtualenv in tox
> Test if virtualenvs and pth files work
> Move the module name searching to the subprocess
> Move the last test out of test_regressions and delete the file
> Move most of the regression tests into other test files
> Cleanup test_regression tests
> Move test_integration_keywrod to test_api/test_keyword
> Cleanup the docstring tests
> More test_evaluate Script fixtures
> stdlib fixture conversions
> More Script fixture conversions in test_evaluate
> Migrate parso integration to script fixture
> Use Script in more places
> Use the Script fixture more generally
> Differentiate between different Python versions in a specific test
> Start replacing Script calls with a fixture
> Monkeypatch the Unpickler in Python3.3
> Try to pass the environment variable for JEDI_TEST_ENVIRONMENT to pytest over tox
> Correct the two last unicode issues
> Some repr went crazy
> Make it more clear for debugging where dynamic search ended
> Fix pep0484 comments
> Better display of descriptors
> Remove some of the last py27 errors that were caused in combination with 3.6
> Forgot to cast a map to a list
> Fixing more py27 stuff
> More unicode literals
> Use force_unicode for all sys paths
> Add a print_to_stderr function in compatibility
> Some more small improvements for Python 2
> There were a few bugs in the previous commit
> Use unicode in way more cases
> Use unicode always for getting special objects
> Skip the correct tests
> Always use the parser of the environment
> Try to test on travis with different jedi test environment variables
> Try adding more automated tests to travis
> Fix potential issues with py2 analysis
> Check for bytes and unicode in dicts for Python 2
> Eliminate is_py3 usages
> Don't use python_version directly on evaluator anymore
> Use unicode always in getattr
> Fix an issue with the new behavior of special methods
> Fix some more py36 to py27 issues
> Refactor some things regarding Python 2 support
> Also don't cast do a string for other names
> Casting to str is not necessary
> Skip tests according to the current environment
> Remove unused keyword code
> Do more comparisons in the subprocess
> Use unicode in compiled access
> Remove unused code
> Refactor allowed_getattr_callback a bit to not raise random errors.
> The default of one function was not actually used
> The string_name of a Name should always be unicode
> Refactor a bit and force unicode in some places and use an appropriate function name for it
> Fix a test fail because of the unicode changes
> The environment selection had a bug
> More unicode conversions
> Don't use getattr, use the abstractions
> More unicode conversions
> Avoid execution of Jedi in test setup
> Start using a lot more unicode literals for Python 2
> Make it possible to explicitly state the version in pytest for different envs
> Use unicode sys paths always
> Don't cast bytes to strings when unpickling
> By default enable cross Python version tests in tox
> Skip more tests if it's necessary.
> Move the import logic to the subprocess
> Move even more import stuff to a separate function
> Move some import parts around to refactor it
> Make ImplicitNamespaceContext a bit cleaner
> Change two tests that were written in a strange way
> Add a way to skip typing tests in non default environments
> Use better error messages for import errors
> Start using the new virtualenv code
> Start using the correct parser for each environment
> Add a way to specify environments in tox
> Fix sys_path propagation for builtins load_module
> A small refactoring
> Merge branch 'master' into virtualenv
> evaluate/docstrings.py
> Don't import numpydoc in the beginning
> Ignore the build directory for pytest
> Refactor environments a bit
> Make it possible to get the right version parser for a certain environment
> api.virtualenv -> api.environment
> DefaultEnvironment -> get_default_environment
> Better support for searching python environments
> Add an optimization around compiled dir()
> Fix a potential issue in evaluate/stdlib
> Cache the subprocess results
> Remove unit test class from speed tests
> Make get_repr static in access.
> There was a bug in creating modules in a subprocess
> Fix the Python 2.7 tests
> Don't try to pickle ellipsis
> Ignore __main__ modules
> Use builtins_module instead of BUILTINS
> Don't use the pickler modification anymore. That doesn't work in other python versions and was in general a bit hard to do
> Remove the last test failures.
> Get the subprocess mostly working
> Don't use the create function anymore in compiled
> Use even more subprocess accesses
> Use the subprocess access to create acceses
> Use access handles everywhere
> Create a way of accessing access objects through a subprocess
> Use a class instead of a dict in get_special_objects
> Move get_special_object
> Move load_module a bit around
> Move another usage of create to builtin_from_name
> Use sys.modules instead of __import__
> Remove unused code
> Start creating access objects in a different way
> Use create_simple_object for a lot of use cases
> Move some of the compiled.create calls to compiled.builtin_from_name
> Move some stuff from compiled to context
> Remove stuff from CompiledObject that didn't belong there and wasn't used
> create_from_access -> _create_from_access
> Refactor a bit more and remove the parent_context parameter from create_from_access
> Isolate fake stuff a bit more
> Simplify an if
> Rewrite bases
> Use create only for non access objects
> Some simplification of _create_from_access
> Remove the old now unused fake code
> Fake context python code is now not the base for a lot of things anymore. It just gets executed.
> Rename SelfNameFilter to SelfAttributeFilter
> Simplify one if statement
> Refactor LazyInstanceName -> SelfName
> Fix all tests except fake docstring stuff
> Fix a slice test with a better helper function
> Fix a few more tests
> Creating objects works now a bit better but is a huge mess.
> Get a lot of tests passing
> Make params work with access
> Add an access abstraction (only array tests work, yet)
> The check if we should add type completions is now a bit more obvious
> Eliminate CompiledObject.type
> Make some faked things private
> Cleanup some compiled stuff.
> Progress in executing builtin stuff in submodules.
> Exceptions now also work over the subprocess.
> Finally got compiled_objects and the access to them working
> Progress when working with evaluators
> Make the subprocesses work and return the right sys paths for the different versions
> Make it possible to connect to a subprocess to get the sys path
> Subprocess progress
> Use types instead of special objects (see also #988)
> Some subprocess progress
> virtualenv progress
* pythonx/parso 2ca629a...579146b (22):
> Don't test python 2.6 in tox by default, because the newer pip versions don't support it anymore
> Set a release date
> Fix whitespace issues with prefixes
> Use proper leafs for fstring start/end
> Fix a Python 2 related issue.
> Forgot to delete another print. WTF I'm tired
> Unfortunately forgot to delete a print
> Start to write a changelog for 0.2.0
> Merge branch 'fstrings'
> A comparison was slightly off
> Ignore the pytest cache
> Fix an error message change in Python 3.7
> Fix an issue in the diff parser about endmarker newlines
> Remove the copyright for good
> Remove copyright years from documentation. (Closes: #25)
> Fix typo: containes→contains
> tox: use older pytest only for py26/py33
> Merge branch 'master' of github.com:davidhalter/parso
> Don't use a newer pytest version
> The tags should be annotated if possible
> Bump version for the next release
> Use 3.7-dev not 3.7 for travis
Squashed:
* fix: CheckHealth doesn't work in neovim
CheckHealth calls display_debug_info().
However, error occured when loading jedi_vim.jedi.Script('')._evaluator.sys_path,
because Evaluator.sys_path is unavailable variable.
* handle the both the old and new API
Evaluator.project.sys_path is temporary fix, because Evaluator.sys_path is correct API.
So, we handle the both the old and new API in preparation for merging virtualenv branch.
This allows vimlparser to parse the vim file again, and it is more
convenient to have it in a real Python file anyway.
Small refactoring and minor text changes included.
Move jedi_vim.py and the jedi submodule into pythonx, which gets added
to Vim's internal sys.path. While jedi cannot be imported directly from
there, it still makes sense for consistency.
This is relevant for when using `:JediDebuginfo` in a broken setup,
where the call to `jedi#init_python()` in the autoload file throws an
error already (but `jedi#debug_info` is defined/loaded before already).
Syntax highlighting on chars in g:jedi#call_signature_escape may
break the funciton parameter hint from jedi-vim. Change its default
value to non python syntax symbols to reduce chance of collision
This was meant as a hack to make the jumplist work (7520290), but it
works without it. Maybe the underlying code to go to the definition has
been changed/improved.
Tested both in Vim (7.4.2143) and Neovim.
I got this error after I updated my local copy of jedi-vim to
the commit 'e8b88f0161ef2adabda97d4e083585a3becdcb53':
```
Error detected while processing function jedi#clear_call_signatures:
line 6:
E492: Not an editor command: jedi_vim.clear_call_signatures()
``
`
The error below happened because 'PythonJedi' was missing before
jedi_vim.clear_call_signatures() in autoload/jedi.vim.
I put back 'PythonJedi' and it solved the issue.
This makes it clearer/more strict altogether and is future-proof in the
sense that `vim.eval` might return integers for numbers in Vim after all.
See #553 for a very noisy discussion of what led to this.
Although it might be a regression in neovim-python after all it is good
to be on the safe side.
The bonus here is that we get "notified" in case of a option that is
expected to be an integer is a string instead.
Fixes https://github.com/davidhalter/jedi-vim/issues/553.
The cursor would change to something else once in the insert mode. If
the new name was longer than the old one, the Parser couldn't get
the stmt at the wrong cursor position, thus rename failed.
Now save the cursor position before entering the insert mode. Then pass
the saved position to the 2nd call to the rename().
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
