but I'll try anyway to confirm

Do you mean the problem is not that the object appeared mysteriously before model(**x) is called, but that model.get(...) could not find it? Do you pass the correct primary key to model.get(...)?

Yes exactly. Yes I even tried hardcoding it. And model.select() is empty

One thing that might be of interest, is that I actually have two databases, I import data from one and copy it to model. Maybe when I call commit() and rollback() it's only doing it on the first one?

No, it should work with both databases

Is there anything I can log to help?

maybe model._database_._get_cache() ?

You can print model._database_._get_cache().indexes

or cache = model._database_._get_cache() print(cache.indexes[model._pk_attrs_])

defaultdict(<class 'dict'>, {})

cache.indexes[model._pk_attrs_] this is also empty

The same checks are happens inside _get_from_identity_map_ when new object is created: cache = entity._database_._get_cache() pk_attrs = entity._pk_attrs_ cache_index = cache.indexes[pk_attrs] And they found the created objects. That means one of the following: * You check the cache for a different database * The object was created later, after the print that you added Are your model belongs to a correct database?

As you are working with two databases, is it possible that you have two different model classes belonging to a different databases and use the wrong model during the model.get(...) call?

hmm

I'm using model on two consecutive lines, one to get and one to write, so I dont think so. Also it works when running it outside of the tests.

I will triple check

this makes no sense, something very cursed is happening. I'm printing the cache and the next line in model(**x) it's just checking it and it has changed somewhere in between

I will dig further and let you know if I figure it out

Is there any way the cache gets updated within the model(**x) call?

Creating the model updates cache at the end. But it should happen later

Is it possible that the overrided __init__ method of the model has some side effects?

It's not overriden

This is so weird, I print model._database_._get_cache().indexes[model._pk_attrs_] on the line right before model(**x) and it's empty

ok so actually I removed the first write of the model in the DB and it still happens

and a raw SQL query confirms that it does not exist in the DB

Did you try this? If you can reproduce the failed test on your developer machine, you can find the line 4426 in pony.orm.core: if obj_to_init is None: obj = object.__new__(entity) And add the next line: if entity.__name__ == 'X' and pkval == (1,): assert False

I'll check if I can do that, currently its within a container thats rebuilt for each test run, so I would need to change the pony install within it

maybe I can monkeypatch the pony function instead haha

You can do something like: from pony.orm.core import EntityMeta original = EntityMeta._get_from_identity_map_ def patched(entity, pkval, status, *args, **kwargs): obj = original(entity, pkval, status, *args, **kwargs) if entity == X and pkval == (1,): assert False return obj EntityMeta._get_from_identity_map_ = patched

Hello, is there any info about migration tool, release, and async support?

Currently the second best option for doing Updates with ponyorm is the pony_up library.

hey guys there isn't a reflection tool to autoload table definitions from an existing database right?

hey guys there isn't a reflection tool to autoload table definitions from an existing database right?

Not with pony. There is one for sqlalchemy that you can execute and adapt the results afterwards

hey guys there isn't a reflection tool to autoload table definitions from an existing database right?

There are Pony inspect

There are Pony inspect

Is pony inspect the same similar to reflection?

Not with pony. There is one for sqlalchemy that you can execute and adapt the results afterwards

By adapt you mean create pony definitions from sqlalchemy?

By adapt you mean create pony definitions from sqlalchemy?

Yes yes

https://github.com/abetkin/pony-inspect You might also try this one. I didn't know about this project. Has some limitations but can be a good start

https://github.com/abetkin/pony-inspect You might also try this one. I didn't know about this project. Has some limitations but can be a good start

Permanent link to the abetkin/pony-inspect project you mentioned. (?)

Hello Please I have a Cloud postgres DataBase How to Backup it to db file i want make some changes such as convert a field type from TEXT to Json

Can pony orm handle system versioning?

Data versioning for the db?

the below is what i meant with regards to system versioning https://docs.microsoft.com/en-us/sql/relational-databases/tables/temporal-tables?view=sql-server-ver15

i don't think pony supports this. neither postgre does AFAIK

ty

Hey all - it doesn't look like Pony supports traditional migrations. If that's the case how do I update the model?

Sorry just searched the chat channel and it looks like migrations are coming. And people are using dbmate to do thier migrations currently

Hey. Tbh they aren't coming in near future. It is really hard for us to find time to work on such things as migrations. In current state migrations do work with PostgreSQL, but some bugs are known. I actually built one project using Pony Migrations with Flyway as migrations manager. If someone in community could help finishing migrations we probably will help with information needed. Even issues with migrations would be really helpful.

Sorry just searched the chat channel and it looks like migrations are coming. And people are using dbmate to do thier migrations currently

That and pony_up

Hey guys, has someone tried to use pony with gevent and postgres? Apparently we are having a ProgrammingError caused by a select call when an async query is underway. The async is set on by gevent with the set_wait_callback of psycopg but I'm not sure if pony code designed to work that way

Pony is not adapted for async at all. The only way to use Pony is - never interrupt db_session with any async call.

https://docs.ponyorm.org/integration_with_fastapi.html#async-and-db-session

I mean async queries not python. Async calls

Pony cannot do async anything

Hi Mario! You can try this approach: https://stackoverflow.com/questions/34812942/psycopg2-copy-from-raises-asynchronous-programmingerror-with-sync-connection-in/37606536#37606536

Nice hint! Thank you much @metaprogrammer

https://github.com/ponyorm/pony/issues/564

@metaprogrammer what about this approach, i have tried and seems to be working. Substituting the pg pool by the class using the threaded connection pool so far is working fine. Any concerns about it?

hey guys does ponyORM support MSSQL? - looking at this link it seems like no? https://docs.ponyorm.org/database.html#binding-the-database-object-to-a-specific-database

hey guys does ponyORM support MSSQL? - looking at this link it seems like no? https://docs.ponyorm.org/database.html#binding-the-database-object-to-a-specific-database

I've never been able to get it working

Hello, I've just installed Pony ORM and I'm testing it a bit. But I get an error when calling db.bind(provider='sqlite', filename='students.db'): "File "/home/massive/.local/lib/python3.9/site-packages/pony/orm/dbproviders/sqlite.py", line 5, in <module> import sqlite3 as sqlite ModuleNotFoundError: No module named 'sqlite3'" I've tried with pip install sqlite3 but it isn't found. I did install pony with pip successfully.

Hi! sqlite3 module is part of a system library. https://docs.python.org/3/library/sqlite3.html It looks like your Python was compiled without sqlite3 for some reason

Could be debian weirdness. Thanks! I'll try installing from another source.

Could be debian weirdness. Thanks! I'll try installing from another source.

Debian does like to cut standard python into parts. There's usually an apt package available to get the missing part back like python3-venv. Pretty sure modern debian doesn't cut sqlite3 from python3 though. I don't have a fresh debian box but I don't remember ever having issues with sqlite3 on my debian and ubuntu servers (except when compiling from source) and python3-bottle-sqlite only depends on python3.

Would like to avoid pony.orm.dbapiprovider.OperationalError: FATAL: too many connections for role "xxxxxxxx" error on Heroku

+1 Please Are you find Solve ?

Pony creates a single connection per process (or per thread, if you use multi-threading) for each Database instance (db variable). If you have too many connection, it may means you have too many Python processes, or you have many Database instances which points to the same database. Each db cache connection for later user, you can call db.disconnect() to close that cached connection

Thank You 🌹 Yes I use Threading I will try Solve it.

Hello! One question, if I use flush() to get the ID of an object (to then send to another external service), am I guaranteed that: 1. If I commit the ID when committing will be the same, 2. If I rollback the object will be removed?

Yeah, I think so.

Hello! One question, if I use flush() to get the ID of an object (to then send to another external service), am I guaranteed that: 1. If I commit the ID when committing will be the same, 2. If I rollback the object will be removed?

1) yes 2) yes. After rollback, on some databases ID of cancelled object can be reused by some other object

Hi, it's possible to generarte entities in Pony from database tables ? (some like reverser enginering)

Maybe something like this could help https://github.com/abetkin/pony-inspect

Maybe something like this could help https://github.com/abetkin/pony-inspect

Permanent link to the abetkin/pony-inspect project you mentioned. (?)

Thanks a Lot!!! All of you

pony.orm.core.BindingError: Database object was already bound to SQLite provider db.bind('sqlite', 'users.sqlite3', create_db=True) db.bind('sqlite', 'groups.sqlite3', create_db=True) db.generate_mapping(create_tables=True) Is it possible to bind two different databases?

I think you need to use a second Database object

I think you need to use a second Database object

Yes, I simply created another table instead of a new database file, by adding another class ("Class Table2(db.Entity):")

Hello, I want to use PonyORM with FastAPI. I follow the instructions, but I get an error pony.orm.core.BindingError: Mapping was already generated. Help me please. Picture1: https://ibb.co/JxhxXwb Picture2: https://ibb.co/R9vQzgk Picture3: https://ibb.co/n8NHWNp

Hello, I want to use PonyORM with FastAPI. I follow the instructions, but I get an error pony.orm.core.BindingError: Mapping was already generated. Help me please. Picture1: https://ibb.co/JxhxXwb Picture2: https://ibb.co/R9vQzgk Picture3: https://ibb.co/n8NHWNp

Try to generate db mappings before defining FastAPI app

Not sure if this would help but who knows

Also Pony doesn't play well with async code so uvicorn might be an issue

Try to generate db mappings before defining FastAPI app

https://ibb.co/3Bmk1Dz :(