-
Notifications
You must be signed in to change notification settings - Fork 564
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Remove duplicate type stub #979
Remove duplicate type stub #979
Conversation
`mypy` exits in an error state if the `pyodbc.pyi` file is included in a type checking session, because of a duplicate definition of `SQL_CONVERT_VARCHAR`. This can be verified by installing `mypy` and running: mypy src/pyodbc.pyi This patch corrects the issue and allows the `pyodbc` usage to be type-checked through at least use of these methods: * `pyodbc.drivers()` * `pyodbc.connect(...)` * `pyodbc.connect(...).cursor()` * `pyodbc.connect(...).cursor().__enter__()` (using a `with` context) * `pyodbc.connect(...).cursor().__exit__()` * `pyodbc.connect(...).cursor().execute()` * `pyodbc.connect(...).cursor().fetchone()` With a note for future work: `mypy --strict` against the same stub file raises several errors pertaining to missing parameter types. Signed-off-by: Alex Nelson <[email protected]>
It seems since PR 864, It seems this is the kind of error that would be caught by including this call somewhere in the CI commands:
Is there an appropriate place to include such a call? I'd be happy to update this PR (or for you to add a follow-on patch) adding this command, so the type stubs can be confirmed to pass static type checking themselves. Also, I hope you'll pardon the boilerplate, but I should include this disclaimer in this conversation:
|
Ah, dear, the CI job died on Python 3.5. However, Python 3.5 is past end of life. Does this project still target 3.5? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Good catch @ajnelson-nist , thank you.
* Add support for Python 3.10, drop EOL 3.5 (mkleehammer#952) * Remove duplicate entry in pyi stub (mkleehammer#979) * Replace deprecated SafeConfigParser with ConfigParser (mkleehammer#953) * Designate connection string as optional (mkleehammer#987) * Fix spelling typos (mkleehammer#985) Co-authored-by: Gord Thompson <[email protected]> * Fix for DSN Names with non-ASCII chars (mkleehammer#951) * Fix for DSN Names with non-ASCII chars Fixes: mkleehammer#948 Co-authored-by: bamboo <[email protected]> Co-authored-by: Gord Thompson <[email protected]> * Added InterfaceError to pyodbc.pyi. (mkleehammer#1013) Co-authored-by: Benjamin Holder <[email protected]> * Upgrade deprecated unicode encoding calls (mkleehammer#792) * Do not include .pyc artifacts in source tarball mkleehammer#742 * Build wheels with cibuildwheels on GitHub Actions Fixes mkleehammer#175 Ref mkleehammer#688 Closes mkleehammer#668 Closes mkleehammer#685 Fixes mkleehammer#441 and pretty much most issues that mention ` sql.h: No such file or directory` This also need to setup some PyPI keys for automated uploads. * Install unixodbc-dev for Linux wheels * Enable GitHub Actions for pull requests * Use Debian based `manylinux_2_24` image * `apt-get` update before installing in wheel build * Use PEP 440 version name required for wheels * Skip building 32-bit wheels * 4.0.dev0 for default version, because test_version() wants 3 parts here Checked this won't shadow released minor version (credit goes to @hugovk) >>> from packaging.version import Version >>> Version("4.0.dev0") > Version("4.0.24") False * Had to use Debian image for PyPy too * Disable PyPy wheels https://cibuildwheel.readthedocs.io/en/stable/options/#build-selection PyPy is missing some C functions that `pyodbc` needs. * Update README.md * Avoid error when testing with DSN= connection Fixes: mkleehammer#1000 * Disable setencoding/setdecoding in tests3/pgtests.py Fixes: mkleehammer#1004 * Adjust test_columns() in tests3/pgtests.py for newer driver versions Fixes: mkleehammer#1003 * Move driver version check out of function * Add comment to _get_column_size() * Fix memory leak with decimal parameters Fixes: mkleehammer#1026 * Create codeql-analysis.yml * Bugfix/sql param data memory leak (mkleehammer#703) * Updated .gitignore * * Created a test file for the specific scenario * * Updated doc of test file for the specific SQLParamData scenario * * Fixed the test file for the specific SQLParamData scenario by Py_XDECREF the PyObject with 1 reference. * * Improved the test to close the cursor and set it to None, then forcing the gc * * Changed the fix of the memory leak and updated the test. * * Removed redundant empty line * * Converted tabs to spaces * * Moved variable out of conn's scope * Update gitignore, remove duplicated * Replace deprecated PyUnicode_FromUnicode(NULL, size) calls (mkleehammer#998) Current versions of Python write a deprecation warning message to stderr, which breaks CGI scripts running under web servers which fold stderr into stdout. Likely breaks other software. This change replaces the deprecated calls with PyUnicode_New(size, maxchar). The accompanying code to populate the new objects has also been rewritten to use the new PyUnicode APIs. * Making pyodbc compatible with PostgreSQL infinity dates, returning MINYEAR and MAXYEAR to python, instead of values out of python's limits * Removing autoformat from code * Removing autoformat from code * Add odbc_config support on mac and m1 homebrew dir * Note EOL of 2.7 support in README (mkleehammer#945) * Fix version of CI generated wheels The CI system is checking out exact tags like "git checkout 4.0.33", which results in a detached HEAD. The version calculation was adding the commit hash. * Fix for mkleehammer#1082 libraries in Linux wheels (mkleehammer#1084) * use argparse instead of optparse (mkleehammer#1089) Co-authored-by: Hugo van Kemenade <[email protected]> Co-authored-by: Alex Nelson <[email protected]> Co-authored-by: Kian Meng, Ang <[email protected]> Co-authored-by: Gord Thompson <[email protected]> Co-authored-by: bamboo <[email protected]> Co-authored-by: Gord Thompson <[email protected]> Co-authored-by: bdholder <[email protected]> Co-authored-by: Benjamin Holder <[email protected]> Co-authored-by: Inada Naoki <[email protected]> Co-authored-by: Michael Fladischer <[email protected]> Co-authored-by: Anatoli Babenia <[email protected]> Co-authored-by: Francisco Morales <[email protected]> Co-authored-by: Gord Thompson <[email protected]> Co-authored-by: Michael Kleehammer <[email protected]> Co-authored-by: Gilad Leifman <[email protected]> Co-authored-by: Bob Kline <[email protected]> Co-authored-by: Leandro Scott <[email protected]> Co-authored-by: Jordan Mendelson <[email protected]> Co-authored-by: Keith Erskine <[email protected]>
mypy
exits in an error state if thepyodbc.pyi
file is included in atype checking session, because of a duplicate definition of
SQL_CONVERT_VARCHAR
. This can be verified by installingmypy
andrunning:
This patch corrects the issue and allows the
pyodbc
usage to betype-checked through at least use of these methods:
pyodbc.drivers()
pyodbc.connect(...)
pyodbc.connect(...).cursor()
pyodbc.connect(...).cursor().__enter__()
(using awith
context)pyodbc.connect(...).cursor().__exit__()
pyodbc.connect(...).cursor().execute()
pyodbc.connect(...).cursor().fetchone()
With a note for future work:
mypy --strict
against the same stub fileraises several errors pertaining to missing parameter types.
Signed-off-by: Alex Nelson [email protected]