Datasets:

---

akumoth

/

peewee-issues

like 0

[]

[]

[ "Aha, thanks! Should be fixed by f68ff7fb30a149fe9df8050b8001a1b4d3636bec", "This fix didn't work -- the error still happens because an AttributeError is the problem, the fix catches an ImportError.\r\n\r\nIf you look at the file `sqlite_ext.py`, line 54 `from peewee import sqlite3`\r\nThen looking at lines 155-165 in `peewee.py` `sqlite3` gets set to `None` if I don't have pysq3 or sqlite3, which I apparently don't have. :)\r\n\r\nAgain super appreciative of all the help and quick fixes. Thanks again.\r\n\r\n-s\r\n\r\n", "Is the traceback the same?\r\n\r\n```\r\nFTS_VER = sqlite3.sqlite_version_info[:3] >= (3, 7, 4) and 'FTS4' or 'FTS3'\r\nAttributeError: 'NoneType' object has no attribute 'sqlite_version_info'\r\n```", "Yes, I believe so, here's the full traceback (using a fresh pip install of 2.9.1):\r\n\r\n```\r\nTraceback (most recent call last):\r\n File \"/Applications/GoogleAppEngineLauncher.app/Contents/Resources/GoogleAppEngine-default.bundle/Contents/Resources/google_appengine/google/appengine/runtime/wsgi.py\", line 240, in Handle\r\n handler = _config_handle.add_wsgi_middleware(self._LoadHandler())\r\n File \"/Applications/GoogleAppEngineLauncher.app/Contents/Resources/GoogleAppEngine-default.bundle/Contents/Resources/google_appengine/google/appengine/runtime/wsgi.py\", line 299, in _LoadHandler\r\n handler, path, err = LoadObject(self._handler)\r\n File \"/Applications/GoogleAppEngineLauncher.app/Contents/Resources/GoogleAppEngine-default.bundle/Contents/Resources/google_appengine/google/appengine/runtime/wsgi.py\", line 85, in LoadObject\r\n obj = __import__(path[0])\r\n File \"XXXX/main.py\", line 20, in <module>\r\n from controllers.XXXX import XXXX\r\n File \"XXXX/controllers/XXXX.py\", line 19, in <module>\r\n from playhouse.flask_utils import PaginatedQuery\r\n File \"XXXX/lib/playhouse/flask_utils.py\", line 12, in <module>\r\n from playhouse.db_url import connect as db_url_connect\r\n File \"XXXX/lib/playhouse/db_url.py\", line 7, in <module>\r\n from playhouse.pool import PooledMySQLDatabase\r\n File \"XXXX/lib/playhouse/pool.py\", line 279, in <module>\r\n from playhouse.sqlite_ext import SqliteExtDatabase\r\n File \"XXXX/lib/playhouse/sqlite_ext.py\", line 70, in <module>\r\n FTS_VER = sqlite3.sqlite_version_info[:3] >= (3, 7, 4) and 'FTS4' or 'FTS3'\r\nAttributeError: 'NoneType' object has no attribute 'sqlite_version_info'\r\n```", "Could you try the latest commit, fc84b65ff3a02eba05ccf63aead2de38412ac9c3 and let me know how it works? Sorry for the bad fix!", "Hi -- just checking in. This fix hasn't made it in to the latest version yet, afaict, on install from pip. I do see it is in the master repo, though.", "It should be in 2.9.2.", "got it! thanks!\n\nOn Thu, Apr 13, 2017 at 8:34 PM, Charles Leifer <[email protected]>\nwrote:\n\n> It should be in 2.9.2.\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/1218#issuecomment-294055750>,\n> or mute the thread\n> <https://github.com/notifications/unsubscribe-auth/AAXqHY27cuB3eDVczU2JPKgVRlJ7niueks5rvr8mgaJpZM4MaWl5>\n> .\n>\n\n\n\n-- \nSavraj Singh\nFounder, Wattvision.com\n609.540.2204 • twitter/gtalk/skype: savraj\[email protected]\n", "Still facing the sqlite3 issue, even though we are using peewee==2.10.1. Please advice at the earliest. Thank you.\r\n\r\nError Stack Trace -\r\n'NoneType' object has no attribute 'sqlite_version_info': AttributeError\r\nTraceback (most recent call last):\r\nFile \"/var/task/handler.py\", line 511, in lambda_handler\r\nreturn LambdaHandler.lambda_handler(event, context)\r\nFile \"/var/task/handler.py\", line 242, in lambda_handler\r\nreturn handler.handler(event, context)\r\nFile \"/var/task/handler.py\", line 335, in handler\r\napp_function = self.import_module_and_get_function(whole_function)\r\nFile \"/var/task/handler.py\", line 233, in import_module_and_get_function\r\napp_module = importlib.import_module(module)\r\nFile \"/var/lang/lib/python3.6/importlib/init.py\", line 126, in import_module\r\nreturn _bootstrap._gcd_import(name[level:], package, level)\r\nFile \"\", line 978, in _gcd_import\r\nFile \"\", line 961, in _find_and_load\r\nFile \"\", line 950, in _find_and_load_unlocked\r\nFile \"\", line 655, in _load_unlocked\r\nFile \"\", line 678, in exec_module\r\nFile \"\", line 205, in _call_with_frames_removed\r\nFile \"/var/task/schedular.py\", line 16, in \r\nimport models_postgres as models\r\nFile \"/var/task/models_postgres.py\", line 4, in \r\nfrom playhouse.db_url import connect\r\nFile \"/var/task/playhouse/db_url.py\", line 118, in \r\nfrom playhouse.berkeleydb import BerkeleyDatabase\r\nFile \"/var/task/playhouse/berkeleydb.py\", line 10, in \r\nsqlite3_lib_version = sqlite3.sqlite_version_info\r\nAttributeError: 'NoneType' object has no attribute 'sqlite_version_info'", "8ba62478f7ce4fc3890fad0d1c42e851a382bdd4", "Thanks for the traceback, that helped me narrow it down." ]

[]

[]

[]

[]

[]

[ "Yeah, it's fixed." ]

[]

[]

[]

[]

[]

[ "There is a mechanism to inspect fields, MyModelClass._meta.fields. Check it out -- should be what you need." ]

[]

[]

[]

[]

[]

[ "https://github.com/coleifer/peewee/blob/master/CHANGELOG.md", "Thank you!!" ]

[]

[]

[ "Welllll... based on your issue description, there's not really anything I can go off of besides that it didn't work for your tests. Can you provide me with something I can reproduce? Some code that exemplifies the issue? What \"not working\" actually means?", "I wrote a small gist with a small flask app using FlaskDB (30 lines) and a test:\r\nhttps://gist.github.com/gkrnours/f2fa9609617c27dd7dc621eff8eb898e\r\nThe test try to follow http://flask.pocoo.org/docs/0.12/testing/#testing\r\n\r\nThe app run as excepted but the test fail when using the test_client with peewee.OperationalError: Connection already open.\r\nMaybe I'm doing something wrong.\r\n\r\n```\r\nTraceback (most recent call last):\r\n File \"test.py\", line 22, in test_client\r\n self.app.get(\"/\")\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/werkzeug/test.py\", line 778, in get\r\n return self.open(*args, **kw)\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/flask/testing.py\", line 127, in open\r\n follow_redirects=follow_redirects)\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/werkzeug/test.py\", line 751, in open\r\n response = self.run_wsgi_app(environ, buffered=buffered)\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/werkzeug/test.py\", line 668, in run_wsgi_app\r\n rv = run_wsgi_app(self.application, environ, buffered=buffered)\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/werkzeug/test.py\", line 871, in run_wsgi_app\r\n app_rv = app(environ, start_response)\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/flask/app.py\", line 1994, in __call__\r\n return self.wsgi_app(environ, start_response)\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/flask/app.py\", line 1985, in wsgi_app\r\n response = self.handle_exception(e)\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/flask/app.py\", line 1540, in handle_exception\r\n reraise(exc_type, exc_value, tb)\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/flask/_compat.py\", line 33, in reraise\r\n raise value\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/flask/app.py\", line 1982, in wsgi_app\r\n response = self.full_dispatch_request()\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/flask/app.py\", line 1614, in full_dispatch_request\r\n rv = self.handle_user_exception(e)\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/flask/app.py\", line 1517, in handle_user_exception\r\n reraise(exc_type, exc_value, tb)\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/flask/_compat.py\", line 33, in reraise\r\n raise value\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/flask/app.py\", line 1610, in full_dispatch_request\r\n rv = self.preprocess_request()\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/flask/app.py\", line 1831, in preprocess_request\r\n rv = func()\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/playhouse/flask_utils.py\", line 171, in connect_db\r\n self.database.connect()\r\n File \"/home/gkr/work/web/peewee-flask/nv/lib/python3.4/site-packages/peewee.py\", line 3652, in connect\r\n raise OperationalError('Connection already open')\r\npeewee.OperationalError: Connection already open\r\n```\r\n", "The error seems pretty clear to me. Are you opening a database connection (even implicitly, to create tables, for instance) anywhere in your tests? If so, be sure to close it. The FlaskDB helper binds handlers to the request setup/teardown cycle in which it connects/closes, respectively. If there's already a connection open in the given thread in which the request is run, then you'll see the above error. It's not a bug, it's a sign to you that you've left a connection dangling somewhere.", "I do not leave open connection. FlaskDB do when used with the Flask test_client.\r\nThe only time I open the database is for creating the table and I close it 2 line after opening it.\r\n\r\nThat's the whole point of the issue, I had to choose between FlaskDB and testing my application.", "Sorry, reopening to look into this more.", "I see the issue. The problem is that you're initializing the database once for every test-case. Because there are two test-cases, the database gets \"init_app()\" called twice. Inside `db.init_app()`, handlers are bound to the app's `before_request()` and `request_teardown()` hooks. Inside the `before_request()` hook, the conn is opened.\r\n\r\nThe problem can be resolved by only calling `app.init_db()` *once*. Your app is essentially a singleton object so it maintains its state from run to run.\r\n\r\nNot a bug in peewee, but in the way you're setting up your tests. Hope that makes sense.", "The explanation make sense yes." ]

[]

[]

[ "It looks like the `id` field is not unique. Since I don't have your model definition, I'm not sure how you've got `id` defined, but that seems to be the issue. If it's the primary key, it needs to be unique.\r\n\r\nThe reason the calls to save() don't work, I'm guessing, is because you're using a non-auto-incrementing primary key. When this is the case, you must call save() with force_insert=True in order to create a new record.", "Here are the models:\r\n```\r\nclass BaseModel(Model):\r\n class Meta:\r\n database = get_database()\r\n\r\n\r\nclass Feature(BaseModel):\r\n class Meta:\r\n db_table = \"feature\"\r\n identifier = CharField(max_length=MAX_IDENTIFIER_LENGTH, null=False)\r\n json_data = JSONField(default={})\r\n filter = ForeignKeyField(Filter)\r\n icon_group = ForeignKeyField(IconGroup)\r\n icon_state = ForeignKeyField(IconState)\r\n tile_set = ForeignKeyField(TileSet)\r\n longitude = DoubleField()\r\n lattitude = DoubleField()\r\n zone_name = CharField(max_length=MAX_ZONE_NAME_LENGTH, null=True)\r\n battery_percentage = IntegerField(null=True)\r\n last_updated = DateTimeField()\r\n```\r\n\r\nThe table is created in code by:\r\n```\r\n model_classes = [getattr(sys.modules[__name__], model_name) for model_name in MODEL_NAMES]\r\n database.drop_tables(model_classes, safe=True, cascade=True)\r\n database.create_tables(model_classes)\r\n```\r\nand after it is created the id field is:\r\n```\r\nid INTEGER default nextval('feature_id_seq'::regclass)\r\n```", "I thought we were talking about the `Filter` model? could you include that models' definition?", "Oops, sorry. I totally included the wrong model..\r\n\r\n```\r\nclass Filter(BaseModel):\r\n class Meta:\r\n db_table = \"filter\"\r\n\r\n gui_label = CharField(max_length=MAX_FILTER_NAME_LENGTH, null=False)\r\n supress_alarms = BooleanField(default=False)\r\n json_data_format = JSONField(null=True, default={})\r\n icon_group = ForeignKeyField(IconGroup, null=False)\r\n```\r\n\r\nThe id field is identical:\r\n```\r\nid INTEGER default nextval('filter_id_seq'::regclass)\r\n```\r\n", "Thank you! Awesome, so you're using the default \"id\" field definition, which is an auto-incrementing primary key. Yet in your code examples, you're setting the \"id\" to a particular value. Is there a reason for that?\r\n\r\n```python\r\nFilter.insert(\r\n id=DEFAULT_FILTER_ALARMS_ID,\r\n```\r\n\r\nCheck out the docs in this and the following related sections for more info: http://docs.peewee-orm.com/en/latest/peewee/models.html#non-integer-primary-keys-composite-keys-and-other-tricks", "I need to ensure particular id values, because there are constants in the code that will reference it.\r\nThis code is all extracted from a database init script. \r\n\r\nIs including the id field causing the issue?\r\n\r\nI guess I could throw an exception if the value returned from `Model.insert().execute()` doesn't match the expected id.", "Eww. Why not add another column that contains a human-readable, non opaque value and key off of that?", "The linked documentation (thanks, it is great) seems to mention `Model.save()` only. My failed examples were not using `Model.save()` and the id numbers being specified were matching those that autoincrement would have provided. \r\n\r\nIs `Filter._meta.auto_increment = False` required if I am trying to use `Model.insert()` with a value specified for an autoincrementing field set to a value that would match the value returned?\r\n\r\n```\r\nFilter.insert(id=4, ...).execute()\r\nFilter.insert(id=24, ...).execute()\r\nFilter.insert(id=64, ...).execute()\r\n```\r\nworks without error without using `Filter._meta.auto_increment = False` first\r\n\r\nIt would seem to me something isn't quite right with `Model.insert()` because it seems to bypass the `Model._meta.auto_increment` value.", "FYI\r\n\r\n```\r\n if DEFAULT_FILTER_ALARMS_ID != Filter.insert(\r\n gui_label=DEFAULT_FILTER_ALARMS_LABEL,\r\n supress_alarms=False,\r\n icon_group=IconGroup.Value.default,\r\n ).execute():\r\n raise ValueError(\"Filter id for {} is incorrect\".format(DEFAULT_FILTER_ALARMS_LABEL))\r\n```\r\n\r\nworks without causing exception in subsequent code.", "If you truly wish to manually specify the IDs, then they shouldn't be auto-incrementing. You can change this by specifying\r\n\r\nid = IntegerField (primary_key=True)\r\n\r\nHonestly I think you're going down the wrong path by specifying the id like this. Just add a column that contains a human-readable key and use that as your constant value.", "I will take your advice and change the field type to: `id = IntegerField (primary_key=True)`\r\n\r\nThank you so much for the prompt responses!" ]

[]

[]

[]

[]

[]

[ "If I'm understanding your \"example\" code, then the following is equivalent:\r\n\r\n```pythondb = SqliteDatabase(':memory:')\r\n\r\nclass Register(Model):\r\n value = IntegerField()\r\n class Meta:\r\n database = db\r\n\r\ndb.create_tables([Register])\r\n\r\nwith db.transaction() as txn:\r\n Register.create(value=1)\r\n Register.create(value=2)\r\n txn.rollback()\r\n txn.commit()\r\n\r\nprint Register.select().count()\r\n```\r\n\r\nWhen I run that script, the output is `0` as I'd expect.", "Did you test with PostgresqlExtDatabase (Postgres 9.6)? With Python 3.6? With peewee 2.8.7?\r\n\r\nFor me the output is `2`.\r\n\r\n```\r\n from playhouse.postgres_ext import PostgresqlExtDatabase\r\n db = PostgresqlExtDatabase()\r\n db.connect()\r\n with db.transaction() as txn:\r\n # User(email='[email protected]').save()\r\n # User(email='[email protected]').save()\r\n User.create(email='[email protected]')\r\n User.create(email='[email protected]')\r\n txn.rollback()\r\n txn.commit()\r\n print(User.select().count())\r\n db.close()\r\n\r\n```\r\n\r\nAny ideas?", "I did not test with Python 3. Maybe you can test with Python 2 and see if that works for you?\r\n\r\nI re-ran the script and I'm still seeing 0 as the output:\r\n\r\n```python\r\nfrom peewee import *\r\nfrom playhouse.postgres_ext import PostgresqlExtDatabase\r\n\r\n\r\n#db = SqliteDatabase(':memory:')\r\ndb = PostgresqlExtDatabase('peewee_test')\r\n\r\nclass Register(Model):\r\n value = IntegerField()\r\n class Meta:\r\n database = db\r\n\r\ndb.create_tables([Register])\r\n\r\nwith db.transaction() as txn:\r\n Register.create(value=1)\r\n Register.create(value=2)\r\n txn.rollback()\r\n txn.commit()\r\n\r\nprint(Register.select().count())\r\ndb.drop_tables([Register])\r\n```\r\n\r\nAre you sure you don't have users in the db you're counting? Maybe drop then re-create the table?", "I tested here a clean example with Py2 and Py3 and it is working. Well, no idea what is happening in my current project. Thanks for the help.", "@paulocheque -- glad to hear it's working for you. One thing to also consider is the possibility you had nested `transaction()` blocks...Because the semantics of nested transactions may be confusing, I'd recommend using `atomic()` instead. By using `atomic()`, peewee will automatically use a transaction or savepoint where appropriate.", "I understand now my mistake. I was using two different `PostgresqlExtDatabase` intances. One for the User model declaration and another one for the queries. Now I am using a single `db` variable and it is working. Thanks for the help again." ]

[]

[]

[ "Peewee runs in autocommit mode unless you're within the scope of a transaction block, so any COMMIT issued as part of a SELECT query is occurring because a transaction is *not* active. If you're managing transactions outside of Peewee, then you may want to turn autocommit off (by passing `autocommit=False` in to the database constructor). I'd recommend against doing that, however.\r\n\r\nAssuming you're running transactions using either the `atomic()` or `transaction()` helpers, then there's nothing to worry about.", "thanks a lot :)" ]

[]

[]

[ "What Python data-type is used by psycopg2 to represent an interval in Postgres? I'd assume a `datetime.timedelta`? I'm good with merging the changes (though please rebase into a single commit and remove the PEP8 changes).", "Yep uses `timedelta` internally. Should be all handled transparently by `psycopg2` and the operations by the operators on `Node` from my initial tests.\r\n\r\nRebased as suggested and removed formatting.", "Looks like the test model differs from the model used in the test-case itself, which is causing the builds to fail. You mind fixing that, and maybe adding an integration test that actually creates the table and verifies the data and data-types being stored/retrieved from the DB?", "It was so close! Just needed to fix the tests up a bit. Any reason you closed this?", "Unfortunately the project I needed this for was cancelled and I'm unlikely to get around to fixing this up anytime soon so didn't want to clutter things up. Feel free to take it on if you think it's a good addition :)", "Thanks @alexlatchford -- I may do that when I've got a little free time. I appreciate the effort!", "Committed f4ebd78b990efc6f0fc57b9a4d81f0cd6ecdf276" ]

[]

[]

[ "Hmm... I received another pull-request with similar changes a while back and passed on it due to the complexity it introduced. My current thinking is that it'd be better to just remove `create_or_get` since it seems to cause confusion and the code one would need to write to work around it is very concise:\r\n\r\n```python\r\ntry:\r\n return MyObj.create(whatever='foo')\r\nexcept IntegrityError:\r\n return MyObj.get(MyObj.whatever == 'foo')\r\n```\r\n\r\nTo me the result does not justify the means (and the subtlety involved). So I think in the next minor release I'll just remove the method instead." ]

[]

[]

[ "Is the create not raising an IntegrityError in the create_or_get case?", "The problem was a combined unique index. My PR should fix it.", "Thanks for your work on the PR. As I said in my comment on the PR, the code one needs to write to do a create-or-get is only a few lines long and, in my opinion, does not justify the complexity of the corresponding helper method. In the next minor release, I'll be removing `create_or_get()` since it seems to cause more confusion that actually help.", "That is a very good idea. I wasn't sure if I could suggest this.\r\n\r\nIn both the old and my improved state, there exist configurations (e.g. multiple unique indexes) where the wrong entry would be returned. Heck, it might not even be possible to say what the right entry would be: If you tried to insert an entry violating _two_ unique indexes at the same time, the method could return the entry for either of those.", "Exactly. For only the most trivial use-cases would that method be very helpful, and the resulting code to achieve the equivalent functionality is only a couple extra lines... Thanks for your work on this PR, I hope you don't feel it was time wasted." ]

[]

[]

[ "Cool, I may make some small tweaks here and there, but looks great, thank you!", "Actually, I've gone ahead and reverted this. It didn't seem to work when I started writing tests for it -- possibly I'm misunderstanding something?", "Could you give me an example of what was not working as expected?", "Ach, I take it back! I think I must've introduced a bug when I was editing the code to clean up some of the names.", "Changes merged, I swapped the `id_column` parameter for one named `referencing_key` instead and made them `VirtualIntegerFields`. Other than that, changes are intact.", "https://github.com/coleifer/peewee/blob/master/playhouse/sqlite_ext.py#L670-L758", "Thanks. The fields `idcolumn` and `parentcolumn` are still strings, though. See the docs from closure.c:\r\n```sql\r\nSELECT element.* FROM element, ct1\r\nWHERE element.groupid=ct1.id\r\nAND ct1.root=?1\r\nAND ct1.idcolumn='parentId'\r\nAND ct1.parentcolumn='groupId';\r\n```", "Oh shit. Thanks! Will re-fix your fix.", "04a3e022144ebf3db6a078ecd83f51dbbf766034" ]

[]

[]

[ "Nice, thanks!" ]

[]

[]

[ "I bumped into this too. The commit that causes this is 9e76c999868197a4911fa2fa457c50ab42431ff1, which came about because of #1108.\r\n\r\nThe pull request explains why the commit is necessary and how you can solve your problem, but in short use `our_db.get_conn()` instead of `our_db.connect()`.\r\n\r\nIf this is a permanent solution then the docs should probably be updated (especially the [Adding Request Hooks](http://docs.peewee-orm.com/en/latest/peewee/database.html#adding-request-hooks) section), but I'll leave that up to @coleifer to decide.", "Thanks for the quick reply Matt, much appreciated!", "Also you can use [`FlaskDB`](https://github.com/coleifer/peewee/blob/master/playhouse/flask_utils.py#L78) from `playhouse.flask_utils` module", "Related: Where are the release notes for 2.8.5 through 2.8.8? :)", "> In our boilerplate Flask connection open/close code (copied from the docs), and it happens only on the first time our server comes up running on the google app engine development server.\r\n\r\nAre you doing anything with the database at module-scope? For instance, creating tables, or anything that might unintentionally cause a connection to be opened?", "@flother -- thanks, by the way, for providing the details on the commit and a workaround.\r\n\r\nThe change prevents connections from being opened and then subsequently not closed because the reference to them was overwritten with a 2nd connection.\r\n\r\nThe OperationalError is clue-ing you in to a spot in your code where you've already got an open connection for the given thread and are trying to open a new one on top of it. Like I said -- check for database accesses at module scope or during application startup, those are likely to be the culprits.\r\n\r\nAs for using `get_conn()` in the hooks, I still think it is advisable to use `connect()`, because in most applications an exception would only be raised because the developer had forgotten to close one somewhere...which seems like a bug." ]

[]

[]

[ "I think that we shouldn't break the convention that accessing the foreign key will, if necessary, load the related object. It makes the code simpler, too.\r\n\r\nJust compare the IDs if that's all you need:\r\n\r\n```python\r\nif tweet.user_id:\r\n```\r\n\r\nOr\r\n\r\n```python\r\nif tweet.user_id == someUser.id:\r\n```" ]

[]

[]

[ "Thanks for the report, pushed 2.8.8." ]

[]

[]

[ "Ummm...that's an interesting question, I guess... I think that you shouldn't need to even check the return value. An exception will be raised if a constraint is violated, for instance. So calls to `save()` should really be assumed to work unless an exception is raised.", "Thanks for replying much faster. " ]

[]

[]

[ "Possibly related to https://github.com/ghaering/pysqlite/issues/110", "If you're iterating over a large result set, peewee caches rows in-memory (to make subsequent iterations/indexing operations fast). In the case of a single-shot iteration over a lot of rows, though, this can OOM you.\r\n\r\nTo fix, just make this change to the problematic line of code:\r\n\r\n```python\r\nwhile True:\r\n # Note we add \".iterator()\"\r\n for row in FileModel.select().iterator():\r\n ...\r\n```\r\n\r\nPlease read:\r\n\r\nhttp://docs.peewee-orm.com/en/latest/peewee/querying.html#iterating-over-lots-of-rows", "This happens for an empty database too.", "then I highly doubt the issue is in Peewee." ]

[]

[]

[ "It seems that Travis still builds from HEAD of origin/master with the PR commit placed on top, so this also fails.", "Reverting \"Move cdefs back into extern from block\" and \"Bring back tests\" makes the tests work again.\r\n\r\nAgain, this branch probably should not be merged. But at least my commit is not breaking anything." ]

[]

[]

[ "Related to https://github.com/coleifer/peewee/issues/1192", "This grosses up the code quite a bit. They already have the classmethod decorator.", "I agree that it grosses up the code a bit. But it's very useful.\r\n\r\nOh well, it's your repo.\r\n ", "The grossness isn't a reflection of your coding ability, I should make that clear -- it's just one of those problems that requires a kinda gnarly fix. Thanks for the patch, I'm sorry if it was wasted time on your part." ]

[]

[]

[]

[]

[]

[ "This is definitely not the ideal scenario, but it fixed my errors so I can continue development... \r\n\r\ncc @coleifer ", "Actually this breaks other functionality... so disregard. \r\n\r\nHad to do this instead:\r\n\r\n```python\r\ndb.execute_sql(\"\"\"\r\n INSERT INTO account_board (account_id, board_id)\r\n VALUES (%s, %s)\r\n\"\"\", (session[\"id\"], b.id))\r\n```", "You're calling Board.create().execute() which is weird...", "Sorry, I deleted my previous comment, because it's actually not technically necessary to use the \"_id\" accessor when passing in the plain old integer foreign key ID (instead of a related model instance).\r\n\r\nI'm having trouble with your example -- it doesn't make sense, is that from your actual code or is it something you made up for the ticket?\r\n\r\n`Board.create(**kwargs).execute()` doesn't make any sense.", "I see that part of the issue is related to having PRIMARY KEY=False. Why not make a composite key on both fields instead?\r\n\r\n```\r\nprimary_key = CompositeKey('account', 'board')\r\n```\r\n\r\nOr does that just cause more issues for you?", "Let me try making a composite key. I'm using Postgres and it doesn't\r\nsupport composite keys but it does support adding a unique index with two\r\nor more columns.\r\n\r\nI didn't mean to put `Board.create`, I meant to put `Board.insert`. Also\r\nthe code is copy/pasted. I'll try your suggestions and get back to you.\r\nThanks for your help.", "@coleifer i modified my code in my OP to reflect your changes. I'm now getting this error:\r\n\r\n```\r\npeewee.ProgrammingError: column \"id\" does not exist\r\nLINE 1: ...rd\" (\"account_id\", \"board_id\") VALUES (2, 79) RETURNING \"id\"\r\n```\r\n\r\n", "Disregard, I got it to work :) thanks @coleifer, I just needed to know about the `CompositeKey` class" ]

[]

[]

[ "And the correct syntax in this case seems to be\r\n\r\n for row in rows:\r\n <do things>\r\n row.job_status = NEW_STATUS\r\n row.save()", "`update` is a classmethod, and is document as such. An update without a where clause is going to cause problems, obviously, as you found." ]

[]

[]

[ "`.join(User, Bot)` is not valid. The docs are pretty clear on the method: http://docs.peewee-orm.com/en/latest/peewee/api.html#Query.join\r\n\r\nInterestingly, since you're joining the same model twice, you'll need to work around Peewee's concept of the query context. Here's what you're probably going for:\r\n\r\n```python\r\n return (Message.select(User, Message, Bot)\r\n .join(User)\r\n .switch(Message) # Query context was User, since we want to join message->bot, we switch back.\r\n .join(Bot)\r\n .where(User.id == user_id,\r\n Bot.id == bot_id,\r\n Message.pub_datetime.between(\r\n start_datetime,\r\n end_datetime)))\r\n```", "Oh my bad, I missed this point in the doc. Thank you for the code fix.\r\nI think however that it would be great if the error message were more user-friendly." ]

[]

[]

[ "No, `exc` is set in the `else` clause, which runs if no Exception is raised...", "I was mistaken! In Python 3, the expression: `except <Exception> as <var>` does not keep `<var>` in scope beyond the exception block. So you were correct in raising this issue. It is fixed in `master` now.", "Great! My bad, I didn't specify python version (it was 3.5, for the record)" ]

[]

[]

[ "What exception is underneath the `UnboundLocalError` I wonder? I would've thought that catching Exception would suffice but it looks like whatever error is being raised extends `BaseException` instead?", "I couldn't replicate. I added a test-case in which invalid SQL is executed via the AsyncCursor and it all works as I'd expect. I've pushed a new release (2.8.6) in case the bug existed in the last release, however." ]

[]

[]

[ "I wrote a blog post on this subject, http://charlesleifer.com/blog/multi-threaded-sqlite-without-the-operationalerrors/\r\n\r\nBut, tl;dr is that you probably are running up against timeouts while waiting for the write lock...possibly?\r\n\r\nYou can try increasing the timeout, by passing `timeout=10` or more to the initialization method of `SqliteDatabase`. Or you can try out `SqliteQueueDatabase` which works as a drop-in replacement for `SqliteDatabase`. Note that you cannot use transactions if you're using `SqliteQueueDatabase`, however.\r\n\r\nI'd suggest taking a look at the post. I'd also suggest that, if you're planning to use SQLite in a multi-threaded application, you read the docs on how locking works. The SQLite docs are really good.", "@coleifer Thanks for that, will look at it. In order to diagnose I ran it single threaded, and I can see the problem really clearly. Most writes happen in .02-.04 seconds. However every once in a while I will get writes that take 1-7 seconds, even up to 49 seconds!\r\n\r\nObviously even a timeout of 10 won't work, so hopefully the post will help out, but is there any way to diagnose why these writes are taking so long comparatively?", "Posting here because it is the last issue related to the database locked error. I've been dealing with this issue the last few months on my application, and here it is how I solved it. It was actually quite simple.\r\n\r\nReading on the web about the problem, most people seem to be solving it by using `timeout=10`. Specifically my app runs on very old/cheap computers and this was simply not cutting. I needed to pass an absurdly large value to `timeout` to make it work, `timeout=100000`. \r\n\r\nIt did raise me the question, is timeout in seconds or milliseconds? I have a couple writes that take more than a second, and thought that setting it to `10` would solve the problem(becaus I thought it was 10 seconds), but it was only solved after it being larger than `1000`.\r\n\r\nKeep reading if you are using Qt.\r\n\r\nThis problem was terribly inconsistent for me, because somehow the way Qt handles its threads would prevent it from happening most of the time. And it was impossible to debug it, because I couldn't reproduce it whenever I wanted. The solution was also quite simple, create some python threads (not Qt threads), and run a bunch of concurrent writes, tweak the timeout number until it stops locking.\r\n\r\nHope this helps some other adventurous developers dealing with incredibly bad computers. ;)\r\n\r\n", "Thanks for bringing this to my attention. I've updated the interfaces and docs so that milliseconds are used everywhere: 171af17f5cced96c15d5e32e832474aa1618ab8c", "Sorry for necro but I think it's missed in a few places, unless that SqliteExtDatabase interface really does use seconds?\r\nhttps://docs.peewee-orm.com/en/latest/peewee/database.html\r\n\r\nEDIT: Also, this doesn't seem to be merged into master? Sorry if i'm asking something stupid here.\r\nhttps://github.com/coleifer/peewee/blob/master/peewee.py#L3416\r\n\r\nEDIT2: I see actually sqlite lists that timeout as seconds, so maybe it's correct now? Rather confusing", "That comment is outdated - seconds are used everywhere for sqlite timeouts." ]

[]

[]

[ "In-memory databases only exist for the connection that opened them. Peewee, by default, uses a connection-per-thread, so each thread has it's own connection (and hence its own in-memory database). You can disable this by passing `threadlocals=False` to the db when initializing it. You might also need to pass `check_same_thread=False`, which is required to use pysqlite when you're sharing a connection across threads.", "thanks for the clarification!", "If i run this:\r\n```\r\ndb = SqliteDatabase(':memory:', threadlocals=False, check_same_thread=False)\r\n```\r\n\r\nI get:\r\n```\r\nTypeError: 'threadlocals' is an invalid keyword argument for this function\r\n```\r\n\r\nCan someone help me how to get it work?", "If you're using peewee 3.x, you do not need to pass the threadlocals keyword argument anymore. The error indicates this.", "But i have the same problem without passing it.", "You can use `thread_safe=False`. Sorry, I should have read the original issue description.\r\n\r\nIn 3.x, `thread_safe=True` means all threads have their own connection. You can pass `thread_safe=False` to make all your threads share the same connection to the underlying memory database." ]

[]

[]

[ "`autocommit` refers to the behavior of Peewee, not what the underlying driver may or may not be doing." ]

[]

[]

[ "Make a for loop. Jesus.", "Wauw" ]

[]

[]

[ "Begin is a no-op on Postgres and MySQL because transaction handling is baked into the Python driver. Don't ask me why this is the case, it just is part of the DB-API 2.0. As you have found out, the way to manually perform a transaction is to turn off autocommit, then execute your queries and call commit manually when done (or rollback as the case may be). The atomic decorator/context manager does this as well.\r\n\r\nYou should be able to just:\r\n\r\n```\r\nwith db.atomic():\r\n db.execute_sql('UPDATE whatever')\r\n```" ]

[]

[]

[ "These flags are used to indicate features supported by the db. They are used, for example, test_queries.py line 1867." ]

[]

[]

[ "Would this also happen if i running peewee in a multi-threaded web-server? I'm not familiar with concurrent programming.", "@chenbo515 I found this case only exists in below scenarios\r\n1) concurrent access to peewee model class. AND\r\n2) wanna use one peewee model class to access multiple databases which have same schema.", "You'll need to implement your own locking, unfortunately. Otherwise the `Using` blocks would need to add a lock. Something like this might work:\r\n\r\n```python\r\nfrom contextlib import contextmanager\r\n\r\ndb_lock = threading.Lock()\r\n\r\n@contextmanager\r\ndef ConcurrentUsing(*args):\r\n with db_lock:\r\n with Using(*args):\r\n yield\r\n```", "@coleifer thank you." ]

[]

[]

[ "Try this:\r\n\r\n```python\r\nfn.GROUP_CONCAT(Clause(\r\n SQL('Distinct'),\r\n MyModel.test_score,\r\n SQL('ORDER BY'),\r\n MyModel.test_score,\r\n SQL('DESC SEPARATOR \\' \\'')))\r\n```\r\n\r\nNot very clean but that should hopefully help you get the SQL you need." ]

[]

[]

[ "Added in 7f6f578f6f06e865ea5f3823c163d39aa6ff2a96\r\n\r\n```python\r\nSomeModel.select().with_lock('SHARE')\r\n```" ]

[]

[]

[ "This is not a feature of peewee." ]

[]

[]

[ "Are you dead set on using `SQL('*')`? The \"peewee\" way would be to write:\r\n\r\n```python\r\nsub = sub.alias('sub')\r\nquery = A.select(sub.c.a, sub.c.b).from_(sub)\r\n```", "OK, you might be right. But there is still a discrepency in functionality between the Cython speedups and non-speedup code.\r\n\r\nI close this issue, because we will use the columns from now on." ]

[]

[]

[ "It's Python...", "@coleifer, Please elaborate. \r\n\r\nI'm using peewee in cases where I am required to query 1 million or more rows.\r\n```python\r\nquery = table.select()\r\ndata = list(query.dicts())\r\n```\r\n\r\nThis process takes over 30s and when there are where clauses, it may take up to 1~2min.\r\n\r\nAre you saying there are no way to increase the performance of peewee or do you believe other libraries say sqlalchemy will also take the same amount of time?", "That's a shitcomment I made, must have been in a hurry or something, so apologies.\r\n\r\nThere's a short section in the docs about iterating over lots of rows: http://docs.peewee-orm.com/en/latest/peewee/querying.html#iterating-over-lots-of-rows\r\n\r\nThe key points are:\r\n\r\n* If you are doing a single-shot iteration and do not need to cache rows in memory, tack on a `.iterator()` to your query. This helps keep memory usage down for large result sets.\r\n* The fastest way to get lots of rows is to avoid creating expensive Python objects. Requesting that rows be returned as dicts, tuples or namedtuples is going to be fastest. If you have joins and are selecting from multiple models, you can also try using \".objects()\", which skips traversing the model graph. You can find examples here: http://docs.peewee-orm.com/en/latest/peewee/relationships.html#selecting-from-multiple-sources\r\n\r\nYou can compare the runtime of the query by itself versus the overhead of Peewee by using \"database.execute_sql(...)\", passing it a SQL string and a list of params. Alternatively, if you have a query object, you can use \"database.execute(query)\". These methods return a DB-API cursor which you can iterate over. This should give you a good baseline for how the query performs in Python.", "@coleifer thank you that you finally admitted that was shit comment. I was quite frustrated that you closed and dismissed my question as like I was stupid. Unfortunately, it's too late for me to use peewee again ... and at that time we decided to use PHP and now I'm working for another company :-)\r\n" ]

[]

[]

[]

[]

[]

[ "Yes, you can specify arbitrary stuff in the `constraints` field parameter:\r\n\r\n```python\r\ntimestamp = DateTimeField(constraints=[SQL('DEFAULT CURRENT_TIMESTAMP')])\r\n```\r\n\r\nhttp://docs.peewee-orm.com/en/latest/peewee/models.html?#default-field-values" ]

[]

[]

[ "I'd accept a PR with docs.", "Maybe something like this:\r\n\r\n```\r\[email protected]_middleware()\r\ndef preProcessing(request, response):\r\n db.connect()\r\n\r\[email protected]_middleware()\r\ndef afterProcessing(request, response, resource):\r\n db.close()\r\n```" ]

[]

[]

[ "I fixed this using slightly more compact code, hope it's more readable now." ]

[]

[]

[ "It's only included in the git repo.", "Actually `pskel` is declared usign the `script=[]` parameter of `setup.py` installation script, which it fails on many occasions to properly install things.\r\n\r\nIt would work out of the box in all python environments if it was converted into console entry-point:\r\nhttp://python-packaging.readthedocs.io/en/latest/command-line-scripts.html#the-console-scripts-entry-point\r\nSetuptools will generate an `.exe` on Windows, and guess correctly the path if python-interpreter on POSIX.\r\n\r\nI can provide a pull-request, but it would require some minor changes:\r\n- create new `main()` functions on `/pwiz.py` and `/playhouse/pskel`, \r\n- rename it as `pskel.py`.\r\n\r\nWould that be ok?\r\n" ]

[]

[]

[]

[]

[]

[ "I do not have any plans for supporting this." ]

[]

[]

[ "It can be validated in the application layer -- *your* application!", "Heh, we already do that! Was wondering if there was any interest upstream.\r\n\r\nThanks!" ]

[]

[]

[ "Fixed in d339c03. Thanks for the patch!" ]

[]

[]

[ "Unfortunately, prefetch cannot be lazy by definition. I'd suggest just writing the queries you need explicitly." ]

[]

[]

[]

[]

[]

[ "Anyone? I've been digging on this for a while myself now, but there is so much abstraction it's hard to figure out how to properly fix this.", "Looking at this now, sorry for the delays!", "Fixed -- Peewee was trying to be too clever by resolving the JOIN-ed closure table to a model instance and patching it to the place it was joined (the source table's primary key)." ]

[]

[]

[ "Going to pass on this." ]

[]

[]

[ "How would it be automatically resolved? What would happen? How would Peewee pick one?\r\n\r\nThe answer you've come up with -- specifying an alias -- is the right approach.", "I'm not sure why you couldn't automatically prepend the name of the model infront of the column name and return both in cases where there is a clash.\n\nI've seen this in other ORMs I've used. The current behavior is dangerous because you don't know which column you are getting back and there are no warnings emitted. It fails silently. \n\nWhat do you think? \n\nOn Mon, Jan 09, 2017 at 4:25 PM Charles Leifer\n\n<\nmailto:Charles Leifer <[email protected]>\n> wrote:\n\na, pre, code, a:link, body { word-wrap: break-word !important; }\n\nClosed\nhttps://github.com/coleifer/peewee/issues/1163\n.\n\n—\n\nYou are receiving this because you authored the thread.\n\nReply to this email directly,\nhttps://github.com/coleifer/peewee/issues/1163#event-916600137\n, or\nhttps://github.com/notifications/unsubscribe-auth/AAznRLh5vOf5cpz7cudDbBGElQciGeUEks5rQqXAgaJpZM4Lb76I\n." ]

[]

[]

[ "Looks like sqlite by default has foreign key support turned off:\r\n\r\nhttp://stackoverflow.com/a/13641365/724251\r\n\r\nNeed to add something like:\r\n\r\n```python\r\nTweet.raw(\"PRAGMA foreign_keys=ON\").execute()\r\n```\r\n\r\nMight be a nicer way included somewhere in the peewee docs.", "@alexlatchford \r\nthanks for pointing it out. \r\nSeems to be mentioned in the documentation, in http://docs.peewee-orm.com/en/latest/peewee/database.html?#additional-connection-initialization\r\n\r\nAlthough imho it is a little bit scary, that referential integrity is turned off by default.\r\n", "@aellwein -- don't use SQLite if you don't know how it works. Simple as that." ]

[]

[]

[ "Added in df65b09722bb14b884f89235e09709c4f398a7aa - thanks for your patience. Language is now an optional extra parameter to the `match` and `Match` functions.\r\n\r\n```python\r\nEntity.select().where(Match(Entity.display_name, prefix, 'english'))\r\n# or\r\nEntity.select().where(Entity.display_name.match(prefix, 'english'))\r\n```", "Excellent! Thanks @coleifer 🥇 " ]

[]

[]

[]

[]

[]

[ "Looks like the 'get' part in 'create_or_get' uses 'AND' because at that point you would be trying to get a specific record, not a record that looks like the supplied parameters.\r\n\r\nhttp://docs.peewee-orm.com/en/latest/peewee/querying.html#create-or-get\r\n\r\nThe same is true for 'get_or_create', you want that specific record, not something that looks like it, and if it couldn't be found, create it.\r\n\r\nThat said, imagine if it used 'OR' instead and I was, for example, looking for a certain resident belonging to a certain address.\r\n\r\nResident.get((Resident.address == '123 Fake Street') | (Resident.name == 'John'))", "These methods use `and` because the Django APIs they are intended to mirror use that behavior. This isn't mongo." ]