I now have class User(db.Entity): # ... web_session = Optional('WebSession') # TODO: remove later. Kept for compatibility. web_sessions = Set('WebSession', reverse="user") And class WebSession(db.Entity): # ... user = Optional(User)

Now I am getting pony.orm.dbapiprovider.ProgrammingError: column "user" does not exist I assume on the WebSession table (the error could be more specific).

Thanks for reporting. That's a bug, I'll fix it

Hey team! How's migration going? Was looking to change a 0,1---0,1 relationship to 0,n---0,1

@here any idea why pony would ever throw an AttributeError: pony.orm.dbproviders.sqlite doesn't have attribute provider_cls ?

Not sure what postgres does with null::int

Thanks for reporting. That's a bug, I'll fix it

You can write something like that: def ignore_none(**kwargs): return {key: value for key, value in kwargs.items() if value is not None} Table.get(**ignore_none(name=name, version=version))

Well... you asked about get, and select is entirely different story. Pony should translate select generator code to SQL, and cannot translate your function, because it doesn't know what equivalent SQL should be. I don't know it too :) Also, the function that you wrote expects keyword arguments, not positional ones as you use inside the query. 1) I suggest you to write selects explicitly: games = select(g for g in Game if (g.name == name or name is None) and (g.version == version or version is None)) 2) You can also build query incrementally: query = select(g for g in Game) if name is not None: query = query.filter(lambda g: g.name == name) if version is not None: query = query.filter(lambda g: g.version == version) 3) Probably even something like that should work: def ignore_none(query, **kwargs): for key, value in kwargs.items(): if value is not None: query = query.filter(lambda x: getattr(x, key) == value) return query query = select(g for g in Game) query = ignore_none(query, name='X', version='Y') I didn't test it. Theoretically it should work, but there is an open issue 223, which describes a bug with getattr which is not fixed yet. You may follow approach (2) for safety

Could you guy please start with code documentation? Most importantly this would be for the public methods, so IDEs like Pycharm could give quick documentation. Also reading the source is really hard this way.

If you specify composite_index(attr1, attr2), the columns will be in that order

Also stuff like .to_json() is not documented online either: https://docs.ponyorm.com/api_reference.html?highlight=to_json#Query.to_json

@luckydonald so you mean add docstrings to all functions?

to_json() was an experimental API, we want to replace it with something completely different. I think we need to remove to_json mentions from the doc

to_json() was an experimental API, we want to replace it with something completely different. I think we need to remove to_json mentions from the doc

The problem with to_json is that it has weak possibilities in querying hierarchical data. We want to replace it with something like that: select(g for g in UniversityGroup if g.graduate_year == 2018).extract(""" number, major, department { number, name, faculty { name } }, courses foreach(c) orderby(c.name) { id, name, credits } students foreach(s) where(s.gpa > 3.5) orderby(s.name) limit(10) { id, name, gpa }, """)

The problem with to_json is that it has weak possibilities in querying hierarchical data. We want to replace it with something like that: select(g for g in UniversityGroup if g.graduate_year == 2018).extract(""" number, major, department { number, name, faculty { name } }, courses foreach(c) orderby(c.name) { id, name, credits } students foreach(s) where(s.gpa > 3.5) orderby(s.name) limit(10) { id, name, gpa }, """)

What I dislike in to_json function is not the data format, but the lack of expressive power when it is necessary to recursively specify fields and subcollections which need to be retrieved

But why implement another lanuage if you could just write python?

(sorry for not pastebin)