Datasets:

---

akumoth

/

peewee-issues

like 0

[]

[]

[ "Maybe related `netAlias.select().get()` fails too with the same error, although `netAlias.get()` works.", "Fixed in d7f0350, thanks so much for reporting." ]

[]

[]

[ "Is there a very compelling reason for using `raw()` in your case? The above could simply be written as `Article.select()`.\r\n\r\nRaw queries *do* support iterator in 3.x:\r\n\r\n```python\r\n\r\nquery = Article.raw('select * from article')\r\nfor article in query.iterator():\r\n # do something with article.\r\n```", "In 3.x raw queries support `iterator()`.", "Thank you." ]

[]

[]

[ "I'm not sure about the answer to 1, but I'm pretty sure that before the only approach that worked was to substitute the circular instance of the `ForeignKeyField` with an `IntegerField` (or whatever), e.g.:\r\n\r\n```python\r\n\r\nclass User(Model):\r\n favorite_tweet_id = IntegerField()\r\n\r\nclass Tweet(Model):\r\n user = ForeignKeyField(User)\r\n```\r\n\r\nIn the above instance no constraint is created for the \"favorite_tweet_id\" column.\r\n\r\nThe approach described in the docs for 2.10 allows you to declare the foreign-key *and* create the constraint:\r\n\r\n```python\r\n\r\n# Create a reference object to stand in for our as-yet-undefined Tweet model.\r\nDeferredTweet = DeferredRelation()\r\n\r\nclass User(Model):\r\n username = CharField()\r\n # Tweet has not been defined yet so use the deferred reference.\r\n favorite_tweet = ForeignKeyField(DeferredTweet, null=True)\r\n\r\nclass Tweet(Model):\r\n message = TextField()\r\n user = ForeignKeyField(User, related_name='tweets')\r\n\r\n# Now that Tweet is defined, we can initialize the reference.\r\nDeferredTweet.set_model(Tweet)\r\n\r\n# Foreign key constraint from User -> Tweet will NOT be created because the\r\n# Tweet table does not exist yet. `favorite_tweet` will just be a regular\r\n# integer field:\r\nUser.create_table()\r\n\r\n# Foreign key constraint from Tweet -> User will be created normally.\r\nTweet.create_table()\r\n\r\n# Now that both tables exist, we can create the foreign key from User -> Tweet:\r\n# NOTE: this will not work in SQLite!\r\ndb.create_foreign_key(User, User.favorite_tweet)\r\n```\r\n\r\nIn 3.0 it has gotten somewhat easier:\r\n\r\n```python\r\n\r\nclass User(Model):\r\n username = CharField()\r\n # Tweet has not been defined yet so use the deferred reference.\r\n favorite_tweet = DeferredForeignKey('Tweet', null=True)\r\n\r\nclass Tweet(Model):\r\n message = TextField()\r\n user = ForeignKeyField(User, backref='tweets')\r\n\r\ndb.create_tables([User, Tweet])\r\nUser._schema.create_foreign_key(User.favorite_tweet)\r\n```", "I think I'll reopen this to add the \"create_foreign_key()\" API back.", "See commit c0c5a0af9ee985bbcb1c4350d992925c1d5b7799 -- updated above comments to reflect changes.", "Documentation for 3.x updated:\r\n\r\n* http://docs.peewee-orm.com/en/latest/peewee/models.html#circular-foreign-key-dependencies\r\n* http://docs.peewee-orm.com/en/latest/peewee/api.html#SchemaManager.create_foreign_key", "My goodness you're fantastic. Thanks for the quick turnaround and thoughts. Any thoughts as to why it was just working before? \r\n\r\nI was using the `test_database` context manager / decorator from `playhouse` which in turn uses `create_model_tables`. It hadn't had any problems or complaints until I upgraded from `2.8.0` (woah, just realized how far behind I'd been; that's like 2 years!) just recently.", "By \"working before\" do you mean that the constraints were all set-up in the database schema? I'm not sure that that's the case. I'm also not sure what specific problems/complaints you ran into after upgrading?", "Sorry, to clarify:\n\n2.8.0: using `test_database` decorator had no problem building the entire graph of tables (circular FK relationships included) from scratch in my test suite, for each test.\n\n>2.8.0: some tables just fail to create because their depended-upon tables don't yet exist.\n\n> On Feb 26, 2018, at 4:03 PM, Charles Leifer <[email protected]> wrote:\n> \n> By \"working before\" do you mean that the constraints were all set-up in the database schema? I'm not sure that that's the case. I'm also not sure what specific problems/complaints you ran into after upgrading?\n> \n> —\n> You are receiving this because you authored the thread.\n> Reply to this email directly, view it on GitHub <https://github.com/coleifer/peewee/issues/1516#issuecomment-368696079>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABA3s8ahuaZcJIQ7DsBbxAi1rm-DjhYwks5tY0ZEgaJpZM4SShgX>.\n> \n\n", "You're positive that *both* foreign key constraints are being created? Hmm... and what version are you using now? It'd be great if you could show me the versions you're using and the exact issues you're experiencing with each.", "I'll have to pull together an example for you without proprietary info. I'm not certain the actual FK constraint is being generated, frankly, but at least the tables get successfully created. I am certain that `2.8.0` it \"works\" on -- insofar that at least the tables don't fail to be created.\r\n\r\nIf you're okay with waiting a bit, I'll try to come up with a simplified example and find the version breakpoint.", "Thanks, that'd be great. I'm willing to bet that the constraint isn't being created in the 2.8 example, so it appears to work, but if you wanted you would be able to break referential integrity.\r\n\r\nYou could test this with your existing code:\r\n\r\n```python\r\nclass User(Model):\r\n favorite_tweet = DeferredForeignKey('Tweet')\r\n username = TextField()\r\n\r\nclass Tweet(Model):\r\n user = ForeignKeyField(User, backref='tweets')\r\n content = TextField()\r\n\r\ndb.create_tables([User, Tweet])\r\n# If the constraint exists, the following will raise an error since the\r\n# tweet with the corresponding ID (12345 in the example) wouldn't exist:\r\nu = User.create(favorite_tweet=12345, username='foo')\r\n```", "Re-reading the issue, I'm a little unclear what exact issue you're encountering? Is there a specific error you're running into after upgrading? Is this error related to `create_tables()` or does it only happen when using some test setup?\r\n\r\nFor what it's worth, peewee 3.x removes `test_database`. There are a few alternatives described in #1440 and the peewee tests themselves.", "Okay, after more monkeying around, I've narrowed it down to `2.10.x`, here's the difference of what occurs.\r\n\r\nWhen using `create_model_tables` (in my case the usage of it embedded within `test_database`), I have determined that, given models `Foo` and `Bar` with a circular FK dependency:\r\n\r\n* `< 2.10.x`: tables create okay, but as you suspect, `Foo`'s FK to `Bar` doesn't actually have the constraint (but at least the table gets created with the right type on the column for `bar_id`).\r\n\r\n* `>= 2.10.x`: tables fail to create right away; creation of `Foo` complains that relation `bar` doesn't yet exist.\r\n\r\nI'll poke around the 2.10.x changelog and commits to see what might have happened, but I'd guess we're zeroing in on some change that more strictly enforces the creation of the constraint straightaway.", "Hmm... it looks like changes to the way deferred foreign-keys were resolved, although there are quite a few changes so I'm not positive.\r\n\r\nAt any rate, I don't think there's a whole lot I can offer, as Peewee is now at 3.x and any fixes would pertain to how the code functions in 3.x (which is somewhat different from 2.10). Sorry not to be of more help.", "I appreciate your help with the investigation. I'm looking to get to 3.x eventually, but wanted to understand what was going wrong here before I opened that can of worms." ]

[]

[]

[ "I don't plan on adding that API, you can very easily use the `get_columns()`, as you've indicated, to achieve equivalent results." ]

[]

[]

[ "The feature was dropped due to it's perceived smelliness... If you want to achieve something similar in 3.x, you can always:\r\n\r\n```python\r\nclass User(Model):\r\n username = TextField()\r\n\r\n @classmethod\r\n def ordered(cls, *args):\r\n return cls.select(*args).order_by(cls.username)\r\n\r\nunordered = User.select().where(...)\r\nordered = User.ordered().where(...)\r\n```", "I miss this feature too, though I do agree that it's a bit smelly. Nevertheless, it can be very handy, especially when you're trying to get a quick and dirty web app put together. One notable example is in flask-admin, where you want the dropdown menus for automatically scaffolded foreign key fields to be in a sane order.\r\n\r\nJust wanted to share my quick re-implementation of this feature for anyone else that might be looking for it:\r\n\r\n```python\r\nclass BaseModel(ModelBase):\r\n\r\n class Meta:\r\n order_by = [] # override this value in subclasses, e.g. `order_by = ['name']`\r\n\r\n # Simple re-implementation of default ordering that was removed in peewee 3.x\r\n # Ideally, ordering should always be explicitly set in query, but this can be\r\n # very handy, esp. for auto-scaffolding frameworks like flask-admin\r\n @classmethod\r\n def select(cls, *fields):\r\n order_by = []\r\n\r\n for field_name in cls._meta.order_by:\r\n if field_name.startswith('-'):\r\n order_by.append(getattr(cls, field_name[1:]).desc())\r\n else:\r\n order_by.append(getattr(cls, field_name))\r\n\r\n return super(BaseModel, cls).select(*fields).order_by(*order_by)\r\n```", "I would think `flask-admin` doesn't provide a way to define or override the query of related objects, but *shrug*?" ]

[]

[]

[ "+1 This error is often due to dependency of GCC, however for me it still fails on WSL (GCC installed via `sudo apt-get install build-essentials`)", "I don't understand that traceback. What file is missing? Is this a recent regression or common to all 3.x?", "Do you have a C compiler and Cython installed? Have you tried installing different versions?", "https://www.microsoft.com/en-us/download/details.aspx?id=44266", "My guess is that you have Cython installed, but perhaps it was installed as a binary package, and that you don't actually have a C compiler? The setup.py for peewee checks if you have Cython, and only if you have Cython, will it include instructions for building the C extensions... So my guess is that Cython is there, but somehow the build_ext command is failing when attempting to compile the actual extension?", "@coleifer \r\n\r\nI am running WSL (Windows subsystem linux) which in my case using Ubuntu distro so no, there is no need for Cython.\r\n\r\nThe build essential should be enough for a C and C++ compiler\r\n```bash\r\nsudo apt-get install build-essential\r\n```\r\n\r\n## Linux (WSL) Error\r\nThe WSL is similar to @Mastergalen's error interestingly.\r\n\r\n```\r\npip install peewee\r\nCollecting peewee\r\n Using cached peewee-3.1.0.tar.gz\r\nBuilding wheels for collected packages: peewee\r\n Running setup.py bdist_wheel for peewee ... error\r\n Complete output from command /home/keto/anaconda3/bin/python -u -c \"import setuptools, tokenize;__file__='/tmp/pip-build-t_5tqb83/peewee/setup.py';f=getattr(tokenize, 'open', open)(__file__);code=f.read().replace('\\r\\n', '\\n');f.close();exec(compile(code, __file__, 'exec'))\" bdist_wheel -d /tmp/tmproueelbcpip-wheel- --python-tag cp36:\r\n running bdist_wheel\r\n running build\r\n running build_py\r\n creating build\r\n creating build/lib.linux-x86_64-3.6\r\n copying peewee.py -> build/lib.linux-x86_64-3.6\r\n copying pwiz.py -> build/lib.linux-x86_64-3.6\r\n creating build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/__init__.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/apsw_ext.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/dataset.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/db_url.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/fields.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/flask_utils.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/hybrid.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/migrate.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/pool.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/postgres_ext.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/reflection.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/shortcuts.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/signals.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/sqlcipher_ext.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/sqlite_ext.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/sqlite_udf.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/sqliteq.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/test_utils.py -> build/lib.linux-x86_64-3.6/playhouse\r\n running build_ext\r\n cythoning playhouse/_speedups.pyx to playhouse/_speedups.c\r\n building 'playhouse._speedups' extension\r\n creating build/temp.linux-x86_64-3.6\r\n creating build/temp.linux-x86_64-3.6/playhouse\r\n gcc -pthread -B /home/keto/anaconda3/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -I/home/keto/anaconda3/include/python3.6m -c playhouse/_speedups.c -o build/temp.linux-x86_64-3.6/playhouse/_speedups.o\r\n gcc -pthread -shared -B /home/keto/anaconda3/compiler_compat -L/home/keto/anaconda3/lib -Wl,-rpath=/home/keto/anaconda3/lib -Wl,--no-as-needed -Wl,--sysroot=/ build/temp.linux-x86_64-3.6/playhouse/_speedups.o -L/home/keto/anaconda3/lib -lpython3.6m -o build/lib.linux-x86_64-3.6/playhouse/_speedups.cpython-36m-x86_64-linux-gnu.so\r\n cythoning playhouse/_sqlite_udf.pyx to playhouse/_sqlite_udf.c\r\n building 'playhouse._sqlite_udf' extension\r\n gcc -pthread -B /home/keto/anaconda3/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -I/home/keto/anaconda3/include/python3.6m -c playhouse/_sqlite_udf.c -o build/temp.linux-x86_64-3.6/playhouse/_sqlite_udf.o\r\n gcc -pthread -shared -B /home/keto/anaconda3/compiler_compat -L/home/keto/anaconda3/lib -Wl,-rpath=/home/keto/anaconda3/lib -Wl,--no-as-needed -Wl,--sysroot=/ build/temp.linux-x86_64-3.6/playhouse/_sqlite_udf.o -L/home/keto/anaconda3/lib -lpython3.6m -o build/lib.linux-x86_64-3.6/playhouse/_sqlite_udf.cpython-36m-x86_64-linux-gnu.so\r\n cythoning playhouse/_sqlite_ext.pyx to playhouse/_sqlite_ext.c\r\n warning: playhouse/_sqlite_ext.pyx:313:57: local variable 'pNew' referenced before assignment\r\n warning: playhouse/_sqlite_ext.pyx:341:55: local variable 'pCur' referenced before assignment\r\n warning: playhouse/_sqlite_ext.pyx:1017:48: local variable 'bf' referenced before assignment\r\n building 'playhouse._sqlite_ext' extension\r\n gcc -pthread -B /home/keto/anaconda3/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -I/home/keto/anaconda3/include/python3.6m -c playhouse/_sqlite_ext.c -o build/temp.linux-x86_64-3.6/playhouse/_sqlite_ext.o\r\n playhouse/_sqlite_ext.c:531:21: fatal error: sqlite3.h: No such file or directory\r\n compilation terminated.\r\n error: command 'gcc' failed with exit status 1\r\n\r\n ----------------------------------------\r\n Failed building wheel for peewee\r\n Running setup.py clean for peewee\r\nFailed to build peewee\r\nInstalling collected packages: peewee\r\n Running setup.py install for peewee ... error\r\n Complete output from command /home/keto/anaconda3/bin/python -u -c \"import setuptools, tokenize;__file__='/tmp/pip-build-t_5tqb83/peewee/setup.py';f=getattr(tokenize, 'open', open)(__file__);code=f.read().replace('\\r\\n', '\\n');f.close();exec(compile(code, __file__, 'exec'))\" install --record /tmp/pip-jszvpzby-record/install-record.txt --single-version-externally-managed --compile:\r\n running install\r\n running build\r\n running build_py\r\n creating build\r\n creating build/lib.linux-x86_64-3.6\r\n copying peewee.py -> build/lib.linux-x86_64-3.6\r\n copying pwiz.py -> build/lib.linux-x86_64-3.6\r\n creating build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/__init__.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/apsw_ext.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/dataset.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/db_url.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/fields.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/flask_utils.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/hybrid.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/migrate.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/pool.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/postgres_ext.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/reflection.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/shortcuts.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/signals.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/sqlcipher_ext.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/sqlite_ext.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/sqlite_udf.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/sqliteq.py -> build/lib.linux-x86_64-3.6/playhouse\r\n copying playhouse/test_utils.py -> build/lib.linux-x86_64-3.6/playhouse\r\n running build_ext\r\n skipping 'playhouse/_speedups.c' Cython extension (up-to-date)\r\n building 'playhouse._speedups' extension\r\n creating build/temp.linux-x86_64-3.6\r\n creating build/temp.linux-x86_64-3.6/playhouse\r\n gcc -pthread -B /home/keto/anaconda3/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -I/home/keto/anaconda3/include/python3.6m -c playhouse/_speedups.c -o build/temp.linux-x86_64-3.6/playhouse/_speedups.o\r\n gcc -pthread -shared -B /home/keto/anaconda3/compiler_compat -L/home/keto/anaconda3/lib -Wl,-rpath=/home/keto/anaconda3/lib -Wl,--no-as-needed -Wl,--sysroot=/ build/temp.linux-x86_64-3.6/playhouse/_speedups.o -L/home/keto/anaconda3/lib -lpython3.6m -o build/lib.linux-x86_64-3.6/playhouse/_speedups.cpython-36m-x86_64-linux-gnu.so\r\n skipping 'playhouse/_sqlite_udf.c' Cython extension (up-to-date)\r\n building 'playhouse._sqlite_udf' extension\r\n gcc -pthread -B /home/keto/anaconda3/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -I/home/keto/anaconda3/include/python3.6m -c playhouse/_sqlite_udf.c -o build/temp.linux-x86_64-3.6/playhouse/_sqlite_udf.o\r\n gcc -pthread -shared -B /home/keto/anaconda3/compiler_compat -L/home/keto/anaconda3/lib -Wl,-rpath=/home/keto/anaconda3/lib -Wl,--no-as-needed -Wl,--sysroot=/ build/temp.linux-x86_64-3.6/playhouse/_sqlite_udf.o -L/home/keto/anaconda3/lib -lpython3.6m -o build/lib.linux-x86_64-3.6/playhouse/_sqlite_udf.cpython-36m-x86_64-linux-gnu.so\r\n skipping 'playhouse/_sqlite_ext.c' Cython extension (up-to-date)\r\n building 'playhouse._sqlite_ext' extension\r\n gcc -pthread -B /home/keto/anaconda3/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -I/home/keto/anaconda3/include/python3.6m -c playhouse/_sqlite_ext.c -o build/temp.linux-x86_64-3.6/playhouse/_sqlite_ext.o\r\n playhouse/_sqlite_ext.c:531:21: fatal error: sqlite3.h: No such file or directory\r\n compilation terminated.\r\n error: command 'gcc' failed with exit status 1\r\n\r\n ----------------------------------------\r\nCommand \"/home/keto/anaconda3/bin/python -u -c \"import setuptools, tokenize;__file__='/tmp/pip-build-t_5tqb83/peewee/setup.py';f=getattr(tokenize, 'open', open)(__file__);code=f.read().replace('\\r\\n', '\\n');f.close();exec(compile(code, __file__, 'exec'))\" install --record /tmp/pip-jszvpzby-record/install-record.txt --single-version-externally-managed --compile\" failed with error code 1 in /tmp/pip-build-t_5tqb83/peewee/\r\n```\r\n", "Cython is installed\r\n```\r\n$ pip show cython\r\nName: Cython\r\nVersion: 0.27.3\r\nLocation: c:\\programdata\\anaconda3\\lib\\site-packages\r\n```\r\n\r\nInstalling the latest Visual Studio Build Tools produces the same error as before.\r\n\r\nI'm temporarily working around this by just setting `cython_installed = False` in `setup.py`", "Thanks for sharing the traceback, this is the problem:\r\n\r\n```\r\nplayhouse/_sqlite_ext.c:531:21: fatal error: sqlite3.h: No such file or directory\r\n```\r\n\r\nIt looks like you've got Cython and a C compiler installed, but are missing the SQLite3 headers. You should be able to:\r\n\r\n```\r\nsudo apt-get install libsqlite3-dev\r\n```\r\n\r\nI wonder if there's a way, in setup.py, to test for the existence of the appropriate headers?", "Regarding checking for the existence of the appropriate headers, just look at the lengths one has to go to:\r\n\r\nhttps://github.com/PyTables/PyTables/blob/develop/setup.py\r\n\r\nUnfortunately, I don't think this is one I'm going to be able to fix... If you have a compiler and Cython, you'll need libsqlite3 as well.\r\n\r\nI'm open to fixing if there's a clean way of doing it... I'm not very keen on setup.py so someone may have good ideas.", "Got this error again and forgot `sudo apt-get install libsqlite3-dev`, would it be possible to print a reminder note on pip installs?", "I'm not sure how one would know to display a warning.", "> `sudo apt-get install libsqlite3-dev`\r\n\r\nThanks!\r\nMaybe it should be specified in the [installation docs](http://docs.peewee-orm.com/en/latest/peewee/installation.html#installing-and-testing)?" ]

[]

[]

[ "Thanks for reporting. Fucking stupid-ass postgres with it's janky raggedy-ass overly-complicated upsert and rules for using aliases in insert queries...grumble grumble.\r\n\r\nWorking on a fix.", "Thanks again for the very helpful report -- added tests and have it working: e642ab9e35603a565d89240b62e0e93c532e569e", "Wow, thanks for the quick fix as always. Very excited to start using it, can't wait to port my raw upsert queries to peewee 3.0. And yeah, fuck their stupid rules for using aliases in insert queries.", "Cool, keep me posted if you run into issues. Pushed 3.0.20 which contains this latest fix.", "Hello, I have the same issue with insert_from method.\r\nDo you know how to fix this?\r\n\r\n```\r\nTraceback (most recent call last):\r\n File \"...\\venv\\lib\\site-packages\\peewee.py\", line 2484, in execute_sql\r\n cursor.execute(sql, params or ())\r\npsycopg2.ProgrammingError: la referencia a la columna «source_cd» es ambigua\r\nLINE 1: ..._name\", \"source_cd\" = EXCLUDED.\"source_cd\" WHERE (\"source_cd...\r\n```", "Can you provide an example?", "I need update rows only when source_cd is different than 1\r\n\r\n*Code*\r\n```\r\n person_alias = Person.alias('x')\r\n inst = person_alias.insert_from(\r\n person_raw_ext.select(person_raw_ext.ext_person_cd,\r\n person_raw_ext.ext_person_formatted_cd,\r\n person_raw_ext.ext_person_name,\r\n person_raw_ext.ext_source_cd),\r\n [person_alias.code, person_alias.formatted,\r\n person_alias.name, person_alias.source_cd]) \\\r\n .on_conflict(conflict_target=Person.code,\r\n preserve=[Person.name, Person.source_cd],\r\n where=(person_alias.source_cd != 1))\r\n\r\n print(inst.sql())\r\n inst.execute()\r\n```\r\n\r\n*Output*\r\n```\r\n('INSERT INTO \"public\".\"person\" (\"person_cd\", \"person_formatted_cd\", \"person_name\", \"source_cd\") SELECT \"t1\".\"ext_person_cd\", \"t1\".\"ext_person_formatted_cd\", \"t1\".\"ext_person_name\", \"t1\".\"ext_source_cd\" FROM \"lnd_sii\".\"vw_person_ext\" AS \"t1\" ON CONFLICT (\"person_cd\") DO UPDATE SET \"person_name\" = EXCLUDED.\"person_name\", \"source_cd\" = EXCLUDED.\"source_cd\" WHERE (\"source_cd\" != %s) RETURNING \"person_cd\"', [1])\r\nTraceback (most recent call last):\r\n File \"C:\\Users\\ah250065\\Documents\\Personal\\bidding\\bidding-etl\\venv\\lib\\site-packages\\peewee.py\", line 2484, in execute_sql\r\n cursor.execute(sql, params or ())\r\npsycopg2.ProgrammingError: la referencia a la columna «source_cd» es ambigua\r\nLINE 1: ..._name\", \"source_cd\" = EXCLUDED.\"source_cd\" WHERE (\"source_cd...\r\n ^\r\n```\r\n", "muchas gracias, I'll put together a failing test and see if I can work out a solution.", "You might try a couple things:\r\n\r\n* upgrade to the latest peewee if you haven't already\r\n* looks like you're using `person_alias` in some places and `Person` in others -- try just using `Person` throughout as the alias doesn't seem necessary in the example given\r\n\r\nAttempted to replicate the query here: 7946037c40030ea5dd7539b054562bf19f57c317 -- the test is passing, though the sql looks a bit different than in your example -- namely, the where clause column reference is fully-qualified." ]

[]

[]

[ "Thanks for reporting, I must've missed this during the rewrite. Fixed in master now.", "Looks like there are still issues with the `filter()` API...looking into it." ]

[]

[]

[ "Out of curiosity, why are you closing the database inside a transaction? That seems incorrect.\r\n\r\nFYI: you ought to change that `db_column=\"document_uuid\"` to `column_name=\"document_uuid\"`.", "> Out of curiosity, why are you closing the database inside a transaction? That seems incorrect.\r\n\r\nWell, this logic was implemented a long time ago and I'm pretty sure, that some reasons exist :) Basically, it depends on `cleanup()`/`on_finish()`/`after_request()` methods inside of the application, and in some cases transaction closes after DB connection has already gone.\r\nPlease, suggest if peewee 3.X can handle this case (2.X version could).\r\n\r\n> FYI: you ought to change that db_column=\"document_uuid\" to column_name=\"document_uuid\".\r\n\r\nthanks for the suggestion, will do\r\n\r\n", "I'm not quite sure how to handle this... I kind of think that Peewee *should* report an error if you close the DB while a transaction is open (although if that's the case, the error message should indicate the problem).", "ok, thanks for suggestions", "f40ec035d2b6f13f4bb5f517ff71e08cf66a1ea1", "FYI `ReconnectMixin` can attempt to close and reopen a DB connection while being inside a transaction, I've had this situation:\r\n\r\n```python\r\nTraceback (most recent call last):\r\n …\r\n File \"/usr/local/lib/python3.9/site-packages/peewee.py\", line 6884, in get\r\n return clone.execute(database)[0]\r\n File \"/usr/local/lib/python3.9/site-packages/peewee.py\", line 1907, in inner\r\n return method(self, database, *args, **kwargs)\r\n File \"/usr/local/lib/python3.9/site-packages/peewee.py\", line 1978, in execute\r\n return self._execute(database)\r\n File \"/usr/local/lib/python3.9/site-packages/peewee.py\", line 2150, in _execute\r\n cursor = database.execute(self)\r\n File \"/usr/local/lib/python3.9/site-packages/peewee.py\", line 3157, in execute\r\n return self.execute_sql(sql, params, commit=commit)\r\n File \"/usr/local/lib/python3.9/site-packages/playhouse/shortcuts.py\", line 226, in execute_sql\r\n self.close()\r\n File \"/usr/local/lib/python3.9/site-packages/peewee.py\", line 3098, in close\r\n raise OperationalError('Attempting to close database while '\r\npeewee.OperationalError: Attempting to close database while transaction is open.\r\n```\r\n\r\nThe funny thing is that it then tries to rollback, but fails because the connection is not working:\r\n\r\n```python\r\nTraceback (most recent call last):\r\n File \"/usr/local/lib/python3.9/site-packages/peewee.py\", line 3301, in rollback\r\n return self._state.conn.rollback()\r\nMySQLdb._exceptions.OperationalError: (2006, 'MySQL server has gone away')\r\n```\r\n\r\nI don't know how to properly handle mid-transaction connection errors. I will probably have to reconnect to the SQL server for each request (even if it seems inefficient to not re-use a connection).", ">I don't know how to properly handle mid-transaction connection errors. I will probably have to reconnect to the SQL server for each request (even if it seems inefficient to not re-use a connection).\r\n\r\nYou have to reconnect and rerun all your queries. If the connection is lost, so is any TX you had open." ]

[]

[]

[ "The resulting query looks like this, which doesn't quite make sense:\r\n\r\n```sql\r\nSELECT COUNT(1) FROM (\r\n SELECT \"t1\".\"id\" FROM \"whatever\" AS \"t1\"\r\n INTERSECT \r\n SELECT \"t2\".\"id\" FROM \"whatever\" AS \"t2\"\r\n) LIMIT 3 AS \"_wrapped\"\r\n```\r\n\r\nWhy are you applying a LIMIT and also using COUNT?", "Looks like this query works, however:\r\n\r\n```sql\r\nSELECT COUNT(1) FROM (\r\n SELECT \"t1\".\"id\" FROM \"whatever\" AS \"t1\"\r\n INTERSECT \r\n SELECT \"t2\".\"id\" FROM \"whatever\" AS \"t2\"\r\n LIMIT 3\r\n) AS \"_wrapped\"\r\n```", "Fixed by 42d9642b54a8242bc1ab86497d943177bf95dff2" ]

[]

[]

[ "You can use `Model._meta.add_field(field_name, field)`.", "> You can use `Model._meta.add_field(field_name, field)`.\r\n\r\nNot able to add field in database using this. How can I do it??", "> Not able to add field in database using this. How can I do it??\r\n\r\nProbably you need to use a schema migration? http://docs.peewee-orm.com/en/latest/peewee/playhouse.html#schema-migrations", "Is any way to get only fields of class ??\r\nEx - \r\nclass Test1(BaseModel):\r\n name = CharField(default = 'Test')\r\n new_column = peewee.IntegerField(default = 0)\r\n new_column1 = IntegerField(default = 0)\r\n new_column2 = IntegerField(default = 0)\r\n new_column3 = IntegerField(default = 0)\r\n new_column4 = IntegerField(default = 0)\r\n\r\noutput expected - \r\n['name', 'new_column', 'new_column1', 'new_column2', 'new_column3', 'new_column4 ']\r\n--\r\n\r\n\r\n", "> > Not able to add field in database using this. How can I do it??\r\n> \r\n> Probably you need to use a schema migration? http://docs.peewee-orm.com/en/latest/peewee/playhouse.html#schema-migrations\r\n\r\nThanks coleifer, It worked...", "@jayashreesathe http://lmgtfy.com/?q=Get+fields+on+model+Peewee+python\r\n" ]

[]

[]

[ "The APIs you're referring to were not ever public APIs, hence the changes to them are not documented. I'll update the \"Hacks\" page to correct the reference to `Clause` (which is now `NodeList`).\r\n\r\n`Model.prepared` has been removed -- depends on what you're trying to do. You can always override the `__init__()` method.", "a309831f50e1a090204024f2a738c3c6f333b6fc", "Adding or removing a field can now be done using `Model._meta.add_field` or `Model._meta.remove_field`.", "I'm using it in a bunch of places, mostly for adding some extra properties to model instances that depend on the data from the db.\r\n\r\nMaybe I should just make these things `@property`s…" ]

[]

[]

[ "Thank you!", "You're welcome. Great ORM, btw." ]

[]

[]

[ "Peewee provides such nice facilities for transaction management (and rollback) ... It's a shame you're not using them.\r\n\r\nIf you want to open a \"manual\" block without using the context manager, just:\r\n\r\n```python\r\nmanual = db.manual_commit()\r\nmanual.__enter__()\r\n```\r\n\r\nTo close:\r\n\r\n```python\r\nmanual.__exit__(None, None, None)\r\n```", "Thank you, this should work. I feel ashamed I missed this yesterday", "No worries!" ]

[]

[]

[ "You need to pass `register_hstore=True` when instantiating your `PostgresqlExtDatabase`.\r\n\r\nhttp://docs.peewee-orm.com/en/latest/peewee/playhouse.html#postgres-ext-api-notes" ]

[]

[]

[ "I'm not sure I understand what your question is. gevent+flask+peewee is my preferred setup for all my projects...the thing I like about gevent is that you don't really need to change your code.\r\n\r\nIf you're using gevent and Postgresql, you can register a wait callback. Here is the code from the [psycogreen](https://bitbucket.org/dvarrazzo/psycogreen) project (see project for license and details):\r\n\r\n```python\r\n\"\"\"A wait callback to allow psycopg2 cooperation with gevent.\r\n\r\nUse `patch_psycopg()` to enable gevent support in Psycopg.\r\n\"\"\"\r\n\r\n# Copyright (C) 2010-2012 Daniele Varrazzo <[email protected]>\r\n# All rights reserved. See COPYING file for details.\r\n\r\n\r\nfrom __future__ import absolute_import\r\n\r\nimport psycopg2\r\nfrom psycopg2 import extensions\r\n\r\nfrom gevent.socket import wait_read, wait_write\r\n\r\ndef patch_psycopg():\r\n \"\"\"Configure Psycopg to be used with gevent in non-blocking way.\"\"\"\r\n if not hasattr(extensions, 'set_wait_callback'):\r\n raise ImportError(\r\n \"support for coroutines not available in this Psycopg version (%s)\"\r\n % psycopg2.__version__)\r\n\r\n extensions.set_wait_callback(gevent_wait_callback)\r\n\r\ndef gevent_wait_callback(conn, timeout=None):\r\n \"\"\"A wait callback useful to allow gevent to work with Psycopg.\"\"\"\r\n while 1:\r\n state = conn.poll()\r\n if state == extensions.POLL_OK:\r\n break\r\n elif state == extensions.POLL_READ:\r\n wait_read(conn.fileno(), timeout=timeout)\r\n elif state == extensions.POLL_WRITE:\r\n wait_write(conn.fileno(), timeout=timeout)\r\n else:\r\n raise psycopg2.OperationalError(\r\n \"Bad result from poll: %r\" % state)\r\n```\r\n\r\ngevent monkey-patches socket (and threading) so pure python database drivers like pymysql should automatically work fine with gevent.", "I am sorry, my English is not best and sometime my writings are not clear, i will try to be more verbose.\r\n\r\nConsider, that i am new to `gevent` and relative new to `peewee` and i wan to to know, if are peewee's DB adapters safe to use (nonblocking). It is not explicitly declared (at least i not found this) in docs and i have basically two options: go into sources or try it. But i am not sure, if i will able to distinguish it in sources. I did some tests, but again, i am not sure if i choose the right test case for it.\r\n\r\nYou provided example with Postgresql, and mentioned `pymysql`, but what about SQLite, or APSW, or ...? Somewhere in your docs i read about `SqliteQueueDatabase` and `gevent`, then have i to choose this if i want use `gevent` with SQLIte? Etc, etc.\r\n\r\nThe `flask-peewee` doc (linked from peewee's doc) has dedicated page for `gevent` usage. But `flask-peewee` is no more developed, then one cannot know, if the information there is still valid and if it is valid for `peewee` itself.\r\n\r\nYou (the author) know these nuances best, then i ask to collect all related information at one place, where the new users can read it (and do not bother you with questions or waste time on internet search, where is near nothing about this) and then they can choose right tools for it needing.", "So `gevent` uses an event loop, meaning when a socket wants to do some blocking operation, it registers with the event loop, which will wake it up when data is ready for it to read/write. Currently gevent does not work with disk i/o, only socket i/o, so you will only need to be concerned about your peewee database when you're talking to a db over a socket. For this reason, sqlite will always be blocking, since it runs in the same address space as your application. If you're talking to a remote postgres database (or mysql), you can use gevent to make the db sockets cooperative. This will happen automatically if the driver is python, but in the case of psycopg2, you are best off registering a wait callback, as shown in the above code.\r\n\r\nRegarding SqliteQueueDatabase, this module exists to work around the fact that SQLite locks the database for writes. You can read [more about the problem here](http://charlesleifer.com/blog/multi-threaded-sqlite-without-the-operationalerrors/). If you are writing a multi-threaded (or multi-green-thread) application, you want to avoid locking the db from multiple threads. The SqliteQueueDatabase works by using a single dedicated worker thread for processing all writes, ensuring that there's never more than one connection writing to the db. The SqliteQueueDatabase accepts a parameter, `use_gevent`, which uses gevent threading primitives instead of the standard lib threading primitives, so you can easily use the library in an application that uses gevent. If you're using gevent and monkey-patching everything already, then technically you don't need to specify use_gevent, because the standard lib threading implementation will have been monkeypatched already. The bottom line, however, is that you should only use SqliteQueueDatabase if you thoroughly understand the issue, the implementation and the trade-offs, as it's designed for a rather narrow use-case and carries a significant trade-off -- no transactional support.\r\n\r\nTo make a long story short, if you're writing a web app with flask, peewee and gevent, and want to use sqlite, you probably don't need to do anything special. Just use `peewee.SqliteDatabase` (or `playhouse.sqlite_ext.SqliteExtDatabase`).", "Very thanks, that i exactly what i want to know now, and slightly more than i was asking to exactly mention in dedicated doc's `gevent` section. ;-)" ]

[]

[]

[ "Never mind. It works after updating from 2.8.5 to 3.0.18." ]

[]

[]

[ "I'm not familiar with mysql-connector-python, but Peewee should work with any DB-API 2.0-compatible driver. If the mysql-connector-python library is fully DB-API 2.0 compatible, you should be able to simply subclass `MySQLDatabase` and override the `_connect()` method so it uses the mysql connector.\r\n\r\nYou can refer to this document for more information: http://docs.peewee-orm.com/en/latest/peewee/database.html#adding-a-new-database-driver", "You can always use PyMySQL", "Peewee supports mysql-connector via `playhouse.mysql_ext.MySQLConnectorDatabase`." ]

[]

[]

[ "I think you might be confusing triggers with the `ON DELETE` behavior, which I don't believe falls under the same rules as triggers (recursive or otherwise). To avoid the issue, it should be sufficient to remove either the \"on delete cascade\" bit, or to disable the foreign_keys pragma.\r\n\r\nI was able to replicate the behavior you've described using just the sqlite shell, so this definitely isn't a peewee issue. Whether it's a bug in sqlite is probably worth bringing up on the mailing list.", "Would there be a way to replace rows without deleting them first or is it necessary to disable `ON DELETE`/the `foreign_keys` pragma to avoid this problem?", "I'm not sure, that's a good question for the SQLite mailing list. It does seem like the combination of:\r\n\r\n1. Pragma foreign_keys = on\r\n2. On delete cascade\r\n3. Insert or Replace\r\n\r\nCan lead to rows in the related table being deleted.\r\n\r\nAn alternative, if you don't want to change the first 2, would be to execute a normal INSERT and then, if that fails with an IntegrityError, to do an update instead." ]

[]

[]

[ "Yes, it was a bad commit, the test is fixed in master now." ]

[]

[]

[ "For sqlite the autofield IS an INTEGER, so this solves nothing. Opened issue #1499" ]

[]

[]

[ "I've edited your comment to include code formatting...please take the time to format your issues. It takes two seconds.", "Wontfix, peewee won't magically create directories for you.", "Thank you for fixing my post I will put more attention into the formatting in the future 😊" ]

[]

[]

[ "I am going to pass on this, but thanks for the patch." ]

[]

[]

[ "Fix pushed. It's not necessary to check both Mapping and MutableMapping, since MutableMappings are Mappings." ]

[]

[]

[ "Made a quick fix, the method was out-of-date. This commit fixes the issue: 0416786d56f770c6cec3b9c7493fae7ffd6a75ec", "Pushed a new version, 3.0.18, which includes the fix.", "@coleifer thank you for the quick response!", "No problems, you should have no issues vendoring the peewee.py module, as that's one of the goals for the project -- to make it easy to drop in where you need it." ]

[]

[]

[ "Cheers, had those lines indented by an extra 2 spaces on accident. Should be fixed." ]

[]

[]

[ "Can you share the model definition with me? Is the primary key of `AdwordsClient` a foreign-key to some other model?\r\n\r\nThe recursion error is occurring during the save operation when you're creating the new `client` instance. It's attempting to access the model class' `_pk` property, which obtains the value of the primary key. It seems that this is triggering a foreign-key lookup which is self-referential.", "@coleifer Yeah, that was the issue. The foreign relation was configured incorrectly. Thanks a lot for your help." ]

[]

[]

[ "Solved by replacing db_table with table_name \r\ndb_table is from an older version (2.x.x). Probably it also caused the foreignKeyConstraint Error" ]

[]

[]

[ "Final I get the code passed, but I need to change something.\r\n\r\n```\r\n--- a/test/package/peewee.py\r\n+++ b/test/package/peewee.py\r\n@@ -315,7 +315,7 @@ OP = attrdict(\r\n BIN_AND='&',\r\n BIN_OR='|',\r\n XOR='^',\r\n- MOD='%',\r\n+ MOD='%%',\r\n EQ='=',\r\n LT='<',\r\n LTE='<=',\r\n@@ -1557,7 +1557,7 @@ class QueryCompiler(object):\r\n OP.XOR: '#',\r\n OP.AND: 'AND',\r\n OP.OR: 'OR',\r\n- OP.MOD: '%',\r\n+ OP.MOD: '%%',\r\n OP.REGEXP: 'REGEXP',\r\n OP.CONCAT: '||',\r\n }\r\n\r\n--- a/test/library/peewee_ext.py\r\n+++ b/test/library/peewee_ext.py\r\n@@ -48,3 +48,6 @@ def result_to_dict(obj):\r\n ele[key] = value\r\n return ele\r\n\r\n+def mod(lhs, rhs):\r\n+ return peewee.Expression(lhs, '%%', rhs)\r\n```", "It appears that you're using Peewee 2.x?", "I've added a test-case in ba8b72a which seems to me to indicate that the code is correct on 3.x" ]

[]

[]

[ "From: https://www.postgresql.org/docs/current/static/datatype-datetime.html\r\n\r\n> For timestamp with time zone, the internally stored value is always in UTC (Universal Coordinated Time, traditionally known as Greenwich Mean Time, GMT). An input value that has an explicit time zone specified is converted to UTC using the appropriate offset for that time zone. If no time zone is stated in the input string, then it is assumed to be in the time zone indicated by the system's TimeZone parameter, and is converted to UTC using the offset for the timezone zone.\r\n>\r\n> When a timestamp with time zone value is output, **it is always converted from UTC to the current timezone zone, and displayed as local time in that zone. To see the time in another time zone, either change timezone or use the AT TIME ZONE construct (see Section 9.9.3).**", "Thank you for the link to the documentation.\r\nIs there a way to tell peewee to output a DateTimeTZField in a desired tz, as suggested in [1]?\r\nThis would be very useful.\r\n\r\nDani\r\n\r\n[1](https://www.postgresql.org/docs/current/static/runtime-config-client.html#GUC-TIMEZONE)", "If you bothered to read the linked documentation, you would have found:\r\n\r\n> The TimeZone configuration parameter can be set in the file postgresql.conf, or in any of the other standard ways described in Chapter 19. There are also some special ways to set it:\r\n>\r\n> The SQL command `SET TIME ZONE` sets the time zone for the session. This is an alternative spelling of `SET TIMEZONE TO` with a more SQL-spec-compatible syntax.\r\n>\r\n> The `PGTZ` environment variable is used by libpq clients to send a `SET TIME ZONE` command to the server upon connection.\r\n\r\nSo, you can run: `db.execute_sql('SET TIME ZONE ...')`." ]

[]

[]

[ "The way prefetch is implemented, it does not use joins. Instead, it nests dependent clauses in subqueries. So your example would:\r\n\r\n```sql\r\n-- Get branches corresponding to shops in the outer query.\r\nSELECT \"t1\".\"id\", \"t1\".\"shop_id\" \r\nFROM \"branch\" AS \"t1\" \r\nWHERE (\"t1\".\"shop_id\" IN (SELECT \"t2\".\"id\" FROM \"shop\" AS \"t2\" LIMIT 10));\r\n\r\n-- Get the shops themselves.\r\nSELECT \"t1\".\"id\", \"t1\".\"name\" FROM \"shop\" AS \"t1\" LIMIT 10;\r\n```\r\n\r\nIt should be possible to write it instead as:\r\n\r\n```sql\r\nSELECT \"t1\".\"id\", \"t1\".\"shop_id\"\r\nFROM \"branch\" AS \"t1\"\r\nINNER JOIN \"shop\" AS \"t2\" ON (\"t1\".\"shop_id\" = \"t2\".\"id\")\r\n-- but what do we do with the limit?\r\n-- if there were tons of branches we could end up reading way more data than we need\r\n-- given the LIMIT on the outer query.\r\n```\r\n\r\nSo instead, maybe something like:\r\n\r\n```sql\r\nSELECT \"t1\".\"id\", \"t1\".\"shop_id\"\r\nFROM \"branch\" AS \"t1\"\r\nINNER JOIN (SELECT \"t2\".\"id\" FROM \"shop\" AS \"t2\" LIMIT 10) as subq ON (subq.id = t1.shop_id)\r\n```\r\n\r\nBut I'm thinking if MySQL doesn't support the first form, it probably doesn't support this form either?", "```sql\r\nSELECT \"t1\".\"id\", \"t1\".\"shop_id\"\r\nFROM \"branch\" AS \"t1\"\r\nINNER JOIN (SELECT \"t2\".\"id\" FROM \"shop\" AS \"t2\" LIMIT 10) as subq ON (subq.id = t1.shop_id)\r\n```\r\n\r\nThis form IS supported and works as expected. \r\n\r\nIt is equivalent to the `IN subq`, but sadly MySQL query planner is really lame and doesn't get it... Just MySQL things." ]

[]

[]

[ "Install psycopg2 yo.", "Thanks bro." ]

[]

[]

[ "You need to install a MySQL python driver like pymysql.\r\n\r\n pip install pymysql", "It worked! Thanks :grinning:" ]

[]

[]

[ "Fixed and pushed new version, 3.0.17. Added more tests to hopefully prevent this (or similar regressions) from recurring." ]

[]

[]

[ "I'm unsure that this can be done cleanly and correctly, so I'm not inclined to implement or merge an implementation.", "Briefly, what are your concerns?", "Where would you install the signal hooks? What would they do? How would you make them work on SQLite and MySQL, which do not have `returning` clause support?", "They can be installed the same way they are now. They could function the same way, by mapping the signal handler over the iterable or providing a hook for passing the collection. \r\n\r\nThey wouldn't work on SQLite or MySQL, just like `returning` doesn't work on them.", "Going to close this as I'm not interested in the feature, but feel free to implement it in a third-party module." ]

[]

[]

[ "For now, can you try this change for 3.x?\r\n\r\n```python\r\n@contextmanager\r\ndef context(schema_name):\r\n with DB.atomic() as transaction:\r\n del DocumentsModel._meta.table # Don't worry, this won't delete any data.\r\n DocumentsModel._meta.schema = schema_name\r\n yield transaction\r\n DocumentsModel._meta.schema = None\r\n```", "yes, this works for peewee==3.0.15 and fails on 2.X, thanks for suggestion", "Pushed 3.0.16, which contains fix for this issue.", "Thanks!" ]

[]

[]

[ "What am I looking at here? What is the output of this on your computer? What's going wrong?", "Possibly related commit: 6bdcd7dc3800654ba034d03b3e7dc01971388e34\r\n\r\n-------------------\r\n\r\nNote: you're joining on several models unnecessarily...or, you've forgotten to add them to the `select()`:\r\n\r\n```python\r\n# e.g.\r\nquery = OAuth2Account.select(OAuth2Account, UserAccount, OAuth2Client).join(...)\r\n```", "I've taken your code and modified it slightly w/some test data to run locally. I see the problem -- when you include the join on `Tenant` then the `UserAccount` is empty.\r\n\r\nFailing test case\r\n\r\n```python\r\n @requires_models(A, B, C)\r\n def test_join_issue_1482(self):\r\n a1 = A.create(a='a1')\r\n b1 = B.create(a=a1, b='b1')\r\n c1 = C.create(b=b1, c='c1')\r\n\r\n query = (C\r\n .select()\r\n .join(B)\r\n .join(A)\r\n .where(A.a == 'a1'))\r\n with self.assertQueryCount(3):\r\n accum = [(c.c, c.b.b, c.b.a.a) for c in query]\r\n\r\n self.assertEqual(accum, [('c1', 'b1', 'a1')])\r\n```\r\n\r\nThe thing I'm confused about in your queries is why, if you're joining on those models and wish to access joined instances, you aren't also selecting the columns from those models? In other words, do you realize you're doing O(n) queries?", "Fixed by c76f07b", "Upgrade to peewee==3.0.16, the issue is fixed.\r\n\r\nThanks!", "Actually, I want to find a `OAuth2Account` by `UserAccount.id` and `Tenant.id`, so I tried to join `Tenant` into the query.\r\n``` python\r\nquery = OAuth2Account.select().join(OAuth2Client).switch(OAuth2Account).join(UserAccount).join(Tenant).where(UserAccount.id == 1).where(Tenant.id == 1)\r\n```" ]

[]

[]

[ "Thanks for reporting.", "Fixed, pushed new version to PyPI - 3.0.15. Thanks again for letting me know.", "Thanks a ton! Confirmed the issue is fixed with 3.0.15." ]

[]

[]

[ "Ewww, thank you.", "thx", "3.0.14 on PyPI." ]

[]

[]

[ "There isn't really a comparable API anymore.", "I was about to submit a new issue but found this one, which seems to be the same, so I'll post mine here: \r\n\r\nI had implemented a decently functional change-tracking system to my peewee v2 integration, which allows me to log/record any changes made to models that this feature is enabled for. In order to do this, I leveraged the `post_init` signal (though for reasons I can't remember, I basically merged the `playhouse.signals` code into my base model to call instance methods (e.g. `self._post_init()`) rather than signals).\r\n\r\nI recently started migrating to peewee v3 and I found that this signal was removed, and I can't find a convenient way to replicate it. \r\n\r\nThis signal is important because I use it to store the initial data of the model, by copying `self._data` (or `self.__data__` in v3) to a separate dict. Then when the model is saved, I compare these old/original values against the current `self._data`/`self.__data__` to determine the differences, generating any relevant \"change history\" records.\r\n\r\nWould it be possible to restore this signal or similar functionality (e.g. `self.prepared()`)? Or do you have any suggestions on how I could build this out? This is the only thing holding me back from migrating to v3 (I'm a big fan of all the other changes/updates). ", "Well, as I wrote, there isn't really a comparable API anymore. You should probably isolate the places in your codebase where changes can occur and do the tracking there. Implementing a service layer is one option.", "Ok bummer. Definitely not a fan of separate service layers. I like having only one way to CRUD (e.g. [wicked domain models](https://vimeo.com/43598193)), which is why hooking into these signals made change tracking so easy/transparent. I will think of something...", "I can't remember how exactly I did it since I left the company that I used it for. But the way I did it is I override __setattr__ (again not 100% sure I haven't used peewee & python in a while) with my BaseAuditModel that inherits ndb.Model. Then when a set happens, I check if __originalValue__ is None then copy __data__ to __originalValue__ then call parent set.", "> I can't remember how exactly I did it since I left the company that I used it for. But the way I did it is I override **setattr** (again not 100% sure I haven't used peewee & python in a while) with my BaseAuditModel that inherits ndb.Model. Then when a set happens, I check if **originalValue** is None then copy **data** to **originalValue** then call parent set.\r\n\r\nAh that's a good approach. Thanks!" ]

[]

[]

[ "> I added recursive resolution without allowing two resolutions of the same scheme.\r\n\r\nI'm not sure I understand the need for this. So your overall db-url would be something like `env://sqlite:///:memory:`?", "The idea is to support the chaining of resolution schemes. Because of the scheme loop constraint, this would only work if at least two resolution schemes are available.\r\n\r\nSo, imagine we have added support for two or more resolution schemes, say, the `env://` resolution scheme as above (which replaces an url of the form `env://#ENVVAR` with the url contained _inside_ the \"$ENVVAR\" environment variable) and also an `rds://` resolution scheme as in the PR #1359 (which provides the endpoint, port, login, db engine, etc. from an AWS RDS cluster id provided as `rds://rds-db-cluster-id`).\r\n\r\nWithout recursive resolution, one would not be able to use an environment variable and the `env://` scheme to store an `rds://` db-url to obtain the real db connection information. One would only be able to store urls that are already supported by playhouse. With recursive resolution, instead, a chain like this is possible:\r\n\r\n`env://#ENVVAR` -> (reads the content of `$ENVVAR`) -> `rds://rds-db-cluster-id` -> (uses boto3 to get endpoint connection information for `rds-db-cluster-id`) -> `mysql://[email protected]:1234/dbname` -> (resolution is finished since `mysql://` is not a resolution scheme) -> the corresponding `MysqlDatabase` is instantiated.\r\n\r\nOne could imagine further resolution schemes, such as parsing a config file or obtaining database connection information from a credential management service. The `env://` scheme is just the simplest possible example I could think of to demonstrate the feature.", "I think this conflates the idea of url scheme handling with configuration.\r\n\r\nTo my thinking, the proper way to handle environment variables is:\r\n\r\n```python\r\ndef configure_database():\r\n connection_string = os.environ['DATABASE']\r\n return db_url.connect(connection_string)\r\n```", "I'm not 100% sure if by \"this\" you mean (1) env:// as a proposed scheme, (2) the chaining of url resolutions or (3) the entire idea of url resolution.\r\n\r\n(1) Handling environment variables was just an illustration here, it's not part of what I propose for playhouse. I agree with you that what you suggest is a simpler way to handle this.\r\n\r\n(2) It's true that the idea of chaining came to me while considering the env:// case, but I also see the generality behind it. You might have a credential/configuration manager providing you an url that you have to resolve again (e.g. the rds:// example) to get the actual endpoint.\r\n\r\n(3) To be brief, my team has a couple of use cases at work where we retrieve database connection information via an API by providing an identifier string. It is only after the API request that we know, what is the database engine type (and thus the right peewee Database subclass to use), the username, the password, etc. This \"dynamic\" mapping from id string to Database subclass and other connection information seemed to fit the whole db-url spirit quite well, but we need more than just urlparse to deduce the class and also a way to inject the login/password obtained from the API response into the `connect_kwargs` dict. The simple idea of generating a new (parsed) url from another url seemed like a reasonably elegant solution. We're open to suggestions, though.", "Having given this some additional thought, I'm going to pass on the proposed change." ]

[]

[]

[ "@cLupus this would limit the versions of Python which peewee can run on ...\r\nThis should probably be added after 3.5 and 3.4 are EOL.", "Does that include [snubs (mypy)](http://mypy.readthedocs.io/en/stable/basics.html#library-stubs-and-the-typeshed-repo) as well, @oz123 ?", "I'm of the opinion that type annotation/hinting is absolutely un-pythonic and just adds noise to what would otherwise be very clean/readable code. Python's a dynamic language. If you want Python + types, there's other, statically-typed languages available...but I'm not interested in them.", "Type annotations do not make the code statically typed, and they are absolutely not noise. For example they let me save hours every day at work by virtue of letting me check what to give to any parameter without leaving the editor instead of having to switch to the browser to read the documentation (because docstrings are sadly also missing), and type checkers have also spared me hours of debugging by letting me find mistakes well before they could become bugs. Type annotations help massively with auto-completion too.\r\n\r\nThe situation since 2018 has changed dramatically and annotations are a tried and true, useful tool at this point, and pretty much considered an expected luxury by active, popular libraries. I think Peewee's API is nice enough that it deserves to be (more) popular, and type annotations can be a deal maker for many.\r\n\r\nOn the topic of compatibility, there is the possibility of publishing type annotations through a separate package (e.g. `peewee-stubs`) by following [PEP 561](https://peps.python.org/pep-0561/), if for some reason placing them directly in the code isn't deemed feasible.\r\n\r\nI propose reopening this issue and I'm willing to help with the implementation myself." ]

[]

[]

[ "I don't see it in the documentation, it's been removed.", "My old code run on peewee 3.1.5 raise ImportError: cannot import name RetryOperationalError\r\nhow to use RetryOperationalError on 3.1.5, the docs can't find anything about it", "It was removed in 3.0." ]

[]

[]

[ "I'm not sure why `flask-admin` was importing `BaseModel`...it's changed in 3.0 to `ModelBase`. So the following should work:\r\n\r\n```python\r\nfrom peewee import (CharField, DateTimeField, DateField, TimeField,\r\n PrimaryKeyField, ForeignKeyField)\r\ntry:\r\n from peewee import BaseModel\r\nexcept ImportError:\r\n from peewee import ModelBase as BaseModel\r\n```", "`peewee.BaseModel` (and latterly `peewee.ModelBase`) is the metaclass that is used to create `Model` classes. It's an internal implementation thing, so I'm not sure why it's being imported by flask-admin.", "They are checking something with \"is_instance()\" against the BaseModel\n\nOn Tue, Feb 6, 2018 at 10:56 AM, Charles Leifer <[email protected]>\nwrote:\n\n> peewee.BaseModel (and latterly peewee.ModelBase) is the metaclass that is\n> used to create Model classes. It's an internal implementation thing, so\n> I'm not sure why it's being imported by flask-admin.\n>\n> —\n> You are receiving this because you authored the thread.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/coleifer/peewee/issues/1475#issuecomment-363467201>,\n> or mute the thread\n> <https://github.com/notifications/unsubscribe-auth/ADyqBIioDqlqVzFdl_10Q4_EguxmbPiDks5tSHYjgaJpZM4R7QH4>\n> .\n>\n\n\n\n-- \nJeff Muday\nInstructional Technology Analyst\nWake Forest University\n1834 Wake Forest Rd.\nWinston Hall Rm. 27\nWinston Salem, NC 27106\nmudayja @ wfu . edu\np 336.758.6171\nf 336.758.6004\n" ]

[]

[]

[ "Apologies. My mistake.\r\nChanging `AutoFloatField` to the following fixed it.\r\n\r\n```Python\r\nclass AutoFloatField(AutoField):\r\n\r\n def python_value(self, value):\r\n return value/10\r\n```\r\nBut with `PostgresqlExtDatabase` both `python_value` and `db_value` are required. :/" ]

[]

[]

[ "Wrote a failing test, will work up a fix.", "Pushed a new release 3.0.13." ]

[]

[]

[ "The code should work with a few modifications...the signature for `execute_sql()` is slightly different. `exception_wrapper` is now `__exception_wrapper__`. `get_cursor()` is just `cursor()`...\r\n\r\nYou'll have to experiment, but look at the way `execute_sql()` is implemented and I'm sure you can sort it out.", "Can you tell us why RetryOperationalError is removed? @coleifer ", "Removed as I have no intention of continuing to maintain it...also, it's my opinion that it's a gross hack.", "This should work if you have legacy code.\r\n\r\n```python\r\nfrom peewee import MySQLDatabase\r\nfrom peewee import OperationalError\r\n\r\n\r\nclass RetryOperationalError(object):\r\n\r\n def execute_sql(self, sql, params=None, commit=True):\r\n try:\r\n cursor = super(RetryOperationalError, self).execute_sql(\r\n sql, params, commit)\r\n except OperationalError:\r\n if not self.is_closed():\r\n self.close()\r\n with __exception_wrapper__:\r\n cursor = self.cursor()\r\n cursor.execute(sql, params or ())\r\n if commit and not self.in_transaction():\r\n self.commit()\r\n return cursor\r\n\r\n\r\nclass MyRetryDB(RetryOperationalError, MySQLDatabase):\r\n pass\r\n\r\n```", "Small note:\r\n\r\n``__exception_wrapper__`` should not have parentheses afterwards -- you don't need to call it. I went ahead and edited your code snippet to make the change.", "The work around seems to have broken. Was `self.__exception_wrapper__` renamed to something else, or removed entirely?\r\n", "Sorry, ``__exception_wrapper__`` is just at module scope.\r\n\r\nSo,\r\n\r\n```python\r\nimport peewee\r\n\r\nwith peewee.__exception_wrapper__:\r\n ...\r\n```" ]

[]

[]

[ "Danke" ]

[]

[]

[ "This is a known limitation of the introspection tools. I'm not sure how to go from the OID returned by the cursor description to a suitable definition of an ArrayField object.", "then leave this issue open. =)\r\n\r\nside note: each array type has its own OID, you just need to map them:\r\nhere is an example list of OIDs pulled from https://github.com/jackc/pgx/blob/master/pgtype/pgtype.go,\r\nthere is a query you can run to pull the list from the DB as well.\r\n```\r\n\tBoolOID = 16\r\n\tByteaOID = 17\r\n\tCharOID = 18\r\n\tNameOID = 19\r\n\tInt8OID = 20\r\n\tInt2OID = 21\r\n\tInt4OID = 23\r\n\tTextOID = 25\r\n\tOIDOID = 26\r\n\tTIDOID = 27\r\n\tXIDOID = 28\r\n\tCIDOID = 29\r\n\tJSONOID = 114\r\n\tCIDROID = 650\r\n\tCIDRArrayOID = 651\r\n\tFloat4OID = 700\r\n\tFloat8OID = 701\r\n\tUnknownOID = 705\r\n\tInetOID = 869\r\n\tBoolArrayOID = 1000\r\n\tInt2ArrayOID = 1005\r\n\tInt4ArrayOID = 1007\r\n\tTextArrayOID = 1009\r\n\tByteaArrayOID = 1001\r\n\tBPCharArrayOID = 1014\r\n VarcharArrayOID = 1015\r\n```", "also if the problem is figuring out the OID of the table columns there is a query for that as well.", ":+1: ", "Support for arrayfield added in: c6b17b59cb491470b975c83ab23fd6e28311a4d2" ]

[]

[]

[ "Committed 8131847467a7be0f44e22f5ceb64470bb8586f93, which fixes the issue. Thank you for reporting, I've also pushed version 3.0.12 to PyPI, which includes the fix.", "Fantastic! many thanks." ]

[]

[]

[ "Ahhh...I believe you're right. Let me look.", "Yeah, that's no longer provided. Updated the docs and pushed a commit." ]

[]

[]

[ "It looks like you now pass in the `field_types` keyword argument when first creating the database. Is that right? ", "Trying the above might work, but now I'm seeing the following when trying to import my models.\r\n\r\n # OUT: Traceback (most recent call last):\r\n # OUT: File \"<input>\", line 1, in <module>\r\n # OUT: from devtest.db import models\r\n # OUT: File \"/home/keith/src/devtest/devtest/db/models.py\", line 81, in <module>\r\n # OUT: class EquipmentModel(BaseModel):\r\n # OUT: File \"/usr/local/lib/python3.6/site-packages/peewee.py\", line 4888, in __new__\r\n # OUT: cls._meta.add_field(name, field)\r\n # OUT: File \"/usr/local/lib/python3.6/site-packages/peewee.py\", line 4689, in add_field\r\n # OUT: field.bind(self.model, field_name, set_attribute)\r\n # OUT: File \"/usr/local/lib/python3.6/site-packages/peewee.py\", line 3750, in bind\r\n # OUT: self._constructor = model._meta.database.get_binary_type()\r\n # OUT: File \"/usr/local/lib/python3.6/site-packages/peewee.py\", line 388, in __getattr__\r\n # OUT: raise AttributeError('Cannot use uninitialized Proxy.')\r\n # OUT: AttributeError: Cannot use uninitialized Proxy.\r\n", "Thanks for reporting. What kind of special field-types are you trying to register?\r\n\r\nIt looks like that traceback you posted is a result of a `BlobField` trying to interrogate the database object for it's \"binary_type\". Since the db isn't initialized, the lookup fails. Let me see about a fix.\r\n\r\nRegarding custom field types, yep, you pass a dict mapping field type to db-specific data-type.", "You might try this commit: 32d9ca7787e8634aba63f216bb7993160f34fc51", "I've pushed a new release, 3.0.11, if you're interested.", "Thanks! I have custom fields for Postgres, some map to `inet`, `cidr`, and `macaddr` special fields. ", "In case you're interested in the custom fields:\r\n\r\nhttps://github.com/kdart/devtest/blob/master/devtest/db/fields.py\r\n\r\n", "Kool, thanks for sharing!", "Hi! \r\n\r\nCould you please share how-to declare a `cidr` field in a peewee model ?\r\n\r\nBTW, the link above handles only IPv4 networks if I am not mistaken.", "Probably depends on your database driver. Postgres has a native data-type for CIDR network. You could probably use an integer column and the Python 3 `ipaddress` module:\r\n\r\n```python\r\nimport ipaddress\r\nnet = ipaddress.ip_network('192.168.1.0/28')\r\nnetmask = net.netmask\r\nfour_bytes = netmask.packed\r\nto_int = struct.unpack('!I', four_bytes)[0]\r\n```", "actually I was wondering if there is sth similar to `JSONField`, `ArrayField`, etc. that I could use in the case of type `CIDR`:\r\n\r\n```\r\nfrom peewee import *\r\nfrom playhouse.postgres_ext import PostgresqlExtDatabase, ArrayField, JSONField\r\nimport psycopg2\r\nimport psycopg2.extras\r\n\r\nclass Interconnect(BaseModel):\r\n as_set = ArrayField(field_class=TextField)\r\n pulse = JSONField(null=True)\r\n prefix = CIDRField()\r\n ...\r\n```\r\n\r\nwhere `prefix` in PostgreSQL is defined already as:\r\n```\r\n prefix CIDR NOT NULL,\r\n```", "@coleifer I just upgraded to peewee 3 and don't see documentation for the missing `register_fields` and `register_ops`. Could some documentation be added on how to proceed? I believe, as kdart mentioned, that there is no classmethod anymore, but we can use `field_types` and `register_function` on the instance. Is that right?", "You don't need to register anything anymore.\r\n\r\n```python\r\nclass MyMagicalField(Field):\r\n field_type = 'column_data_type'\r\n\r\n def special_field_expression(self, value):\r\n return Expression(self, 'operator', value)\r\n```" ]

[]

[]

[ "This is python, a dynamic language. The linter is, quite simply, wrong. Not fixing because there's no bug here. Just silly tools." ]

[]

[]

[ "Postgresql treats `IN ()` (and incidentally `NOT IN ()`) as invalid SQL expressions. Because the semantics of `IN ()` are equivalent to false, Peewee replaces `IN ()` with `1 = 0`. I suppose it should do similar rewriting for `NOT IN ()`, converting it to `1 = 1` since it is always true.\r\n\r\nThat being said, it seems like the proposed change may be correct.", "Replaced your patch with bc61d3fa515c462cb86714e00ffdf728121a90f7 -- which handles the IN/NOT IN special-case correctly. I **did**, however, choose to keep your changes to `NodeList.__sql__` as I think they're semantically correct." ]

[]

[]

[ "It's `Database.cursor()` now. Updated docs to reflect this." ]

[]

[]

[ "I've restricted the kwargs to just \"coerce\" and merged a similar patch. Thank you for pointing this out: 51cb6bdc25f62916fbc1f9bc471f1a1bccc1a7da" ]

[]

[]

[ "Specify ``coerce=False`` in the call to the function.\r\n\r\n```python\r\nfn.ARRAY_AGG(Some.thing, coerce=False)\r\n```\r\n\r\nhttp://docs.peewee-orm.com/en/latest/peewee/api.html#Function" ]

[]

[]

[ "> Isn't it possible for peewee to detect a PostgreSQL ArrayField is involved, and thus the operands should not be unpacked?\r\n\r\nWhat behavior are you seeing in relation to ArrayField?", "I get:\r\n``` python\r\nIn [1]: Network.target_names.__class__\r\nOut[1]: playhouse.postgres_ext.ArrayField\r\n\r\nIn [2]: Network.select().where(Network.target_names == ['efeITG_GB']).sql()\r\nOut[2]:\r\n('SELECT \"t1\".\"id\", \"t1\".\"filter_id\", \"t1\".\"train_script_id\", \"t1\".\"feature_prescale_bias\", \"t1\".\"feature_prescale_factor\", \"t1\".\"target_prescale_bias\", \"t1\".\"target_prescale_factor\", \"t1\".\"feature_names\", \"t1\".\"feature_min\", \"t1\".\"feature_max\", \"t1\".\"target_names\", \"t1\".\"target_min\", \"t1\".\"target_max\", \"t1\".\"timestamp\" FROM \"develop\".\"network\" AS \"t1\" WHERE (\"t1\".\"target_names\" = (%s))',\r\n [['e', 'f', 'e', 'I', 'T', 'G', '_', 'G', 'B']])\r\n```\r\nWhere I would expect the SQL as generated by the mentioned workaround:\r\n``` python\r\nIn [4]: Network.select().where(Network.target_names == AsIs(['efeITG_GB'])).sql()\r\nOut[4]:\r\n('SELECT \"t1\".\"id\", \"t1\".\"filter_id\", \"t1\".\"train_script_id\", \"t1\".\"feature_prescale_bias\", \"t1\".\"feature_prescale_factor\", \"t1\".\"target_prescale_bias\", \"t1\".\"target_prescale_factor\", \"t1\".\"feature_names\", \"t1\".\"feature_min\", \"t1\".\"feature_max\", \"t1\".\"target_names\", \"t1\".\"target_min\", \"t1\".\"target_max\", \"t1\".\"timestamp\" FROM \"develop\".\"network\" AS \"t1\" WHERE (\"t1\".\"target_names\" = %s)',\r\n [['efeITG_GB']])\r\n```\r\nI'm not sure if this should be natively handled by peewee, but I think equality check with an ArrayField is a common thing to do.", "Postgresql array equality is fixed in 4a14174 (and inequality/gt/lt/ge/le).", "More commonly, though, folks use the containment operators when working with arrays. So this hasn't really come up before.", "Thanks! I did not notice this in the howto's/guides, but to be honest it has been a while since I read them thoroughly." ]

[]

[]

[ "Ahhh, interesting. Will get this fixed.", "Fixed and added a test-case." ]

[]

[]

[ "Very nice! Thank you and apologies for the oversight." ]