;)

I ditched pizza an got chickennuggets.

Also, is there a way to implicitly add a condition to every query for a given model? By default (or even a hard coded unconditional addition to every SELECT), I only want to select entities that have model.deleted=="N" (the stupid pseudo Boolean column).

Also, is there a way to implicitly add a condition to every query for a given model? By default (or even a hard coded unconditional addition to every SELECT), I only want to select entities that have model.deleted=="N" (the stupid pseudo Boolean column).

Maybe having views in the database, or you can have that lambda function stored in a variable

But then I still need to remember to use the lambda var...

I can look into views, no one else is using them currently, so maybe I can add them without affecting other apps using the DB. Hmm...

How do I use the result of them in the same file?

The problem is basically we can only access the table Entities from the same file where the `db=Database()` part is. And the database binding will be at the users side, as he gets to choose the database used. Mysql, posgres,... That means that the entities from the library aren't available inside the library to use, as i cannot import the db from the users code, that is independent

I've been toying with the idea of using punq to do light weight dependency injection, injecting the db somehow...

I think the idea of blueprints is all that is needed. Basically we’d have to change the EntityMeta to not directly act, but use a mehod call on the db to register it’s entity. That way the blueprints store that call. Basically flask Blueprints are really easy. You have a deferred_functions list, and every function from the Flask object get replaced by a call to record_once, which just adds that to the deferred_functions list.

As soon as you call blueprint.register(app), it will iterate though that list and call them functions

In new migrations you will be able to work with a new layer of entites (as we call it now VirtualEntities) and it gives you an ability to dynamically define entities and even attributes.

WOW! That sounds good! Is that what's in the orm-migrations branch?

I'd love to see more about VirtualEntities :)

Is there any way in the online editor to have a relationship with an entity that has a composite primary key? (I need to specify that two columns must match, not just one)

No, it's not published yet. It is in dev branch.

Out of curiosity, what databases are people here using?

Mainly Postgres, with some SQLite

Yeah, I've noticed that Postgres seems to be popular. Whenever I see a talk/presentation that involves a relational database, it's Postgres more often than not.

I imagine that SQLite is used a lot for testing

Yep

SQLite is also useful when having a single file rather than an independent database is a benefit

If you are new to databases and are using pony, any popular one is fine

I'm not new to databases, but would like to use PonyORM next time I have a project involving a relational database. I've spent quite a bit of time reading about database comparisons, and to me it looks like the best way is just to pick one, then worry about performance etc. later if needs be.

I agree

I use both Postges and SQLite in production. The latter is used to store historical and some other data. But Pony does not allow for entities backed by different databases to be interlinked with each other. I hope it will be implemented sometime 🙂

I have a timezone offset from an external system which is supplied to me as a float. Pony doesn’t seem to support a float for timedelta eg timedelta(hours=5.5)

Is that a bug or intended behaviour?

TypeError: 'hours' argument of timedelta(...) function must be of 'int' type. Got: 'float' (inside X.y)

Should I just do timedelta(seconds=value * 3600) ?

I am generating the timedelta iinside a method

calling the method directly works

using the method within a pony query causes issues

seems like it may be a bug

def current_time(self): return datetime.datetime.utcnow() + datetime.timedelta(seconds=int(self.y * 3600))

a query like select(.... if x.z < a.current_time()) blows up

pony==0.7.9

Error is?

AssertionError: [CallFunc(Name('int'), [Mul((Getattr(Name('self'), 'facebook_timezone_offset_hours_utc'), Const(3600)))], None, None), CallFunc(Name('int'), [Mul((Getattr(Name('self'), 'facebook_timezone_offset_hours_utc'), Const(3600)))], None, None), AssName('seconds', 'OP_ASSIGN'), AssName('seconds', 'OP_ASSIGN'), CallFunc(Name('print'), [Name('seconds')], None, None), CallFunc(Getattr(Name('datetime'), 'timedelta'), [Name('seconds')], None, None), AssName('delta', 'OP_ASSIGN'), CallFunc(Name('print'), [Name('delta')], None, None)]

It seems like the query translator gets confused

And what is the line of exception inside pony? (before throw() method)

Why is the dev branch in a private repo? Is that an artifact of how Pony used to be dual licensed? (is it still dual licensed to people that originally bought the license requiring you to keep a separate repo?)

What's the best way to add a string constant to a pony model?

class X(db.Entity): ... my_constant = 'blah'

?

yes

thank you

Hi all, I’m working on a script where I need to create entities on the fly, that is I do not know the full scheme before hand, is it possible to do this in pony ?

It is possible, but not very convenient. After migrations release it will be easier. Right now you can do it using metaclasses:

In simple cases (if you don't use composite keys) you can do something like that: from pony import orm from pony.orm import core def define_entity(db, entity_name, attrs): return core.EntityMeta(entity_name, (db.Entity,), attrs) db = orm.Database() entity = define_entity(db, 'Person', dict( name=orm.Required(str), age=orm.Optional(int) )) orm.sql_debug(True) db.bind('sqlite', ':memory:') db.generate_mapping(create_tables=True)

Thanks, I’ll try it out, should fit quite well with what I’m doing :)

Hello sweet people of the Pony community

Hello sweet people of the Pony community

Hi Jeff, welcome!

Hello sweet people of the Pony community

Hello

Hi, i have a question I use pony orm with sqlite in wal mode I create two different orm.db_session to work with In first transaction , i have deleted all rows from some table In second transaction , I'm trying to read count of rows in this table. And i have got 0 not as expected for me .The trouble is the fact that when i'm trying to read in transaction2, the deletions from transaction1 have not already committed, so read operation from transaction2 should return not 0, because rows exists in this table What may be the case of this trouble? I also have tried to use embedded in python sqlite3 library instead of second pony transaction. And with this library all is fine - i have got correct count of rows What's wrong?

It is not nested transaction It's just two different transactions

I have written some code of this https://pastebin.com/TueTriqC

In logs we will see "table has 0 rows ", but first transaction was not committed

Seems like second transaction running inside of first.. but why and how to deal with it?

I don't think you can use await in a db_session context manager, as it is not an asynchronous context manager, so it commits as soon as you await because you exited the context.

You can decorate an async function however according the docs (search for coroutine)

With it decorated, you have to explicitly commit when you're ready

I don't think you can use await in a db_session context manager, as it is not an asynchronous context manager, so it commits as soon as you await because you exited the context.

You are right, thanks As i understand, pony orm is not so good in using with async code

Database transactions is not so good together with asynchronous code

I'm beginning a project that uses aiohttp in serving a web API with pony as the interface for a DB. This article convinced me this could work. It went into some of the problems of trying to use await in the middle of a DB transaction. https://techspot.zzzeek.org/2015/02/15/asynchronous-python-and-databases/

Hey there. I'm getting ValueError: Attribute State.chat_id is required while looking up an value, failing the State.get(...) call, and I did provide the chat_id as argument: class State(db.Entity, self.State): user_id = orm.Required(int, nullable=True) # can be None (e.g. channels) chat_id = orm.Required(int, nullable=True) # can be None (e.g. inline_query) state = orm.Required(str) data = orm.Optional(orm.Json, nullable=True) # can be None orm.PrimaryKey(user_id, chat_id) # end class user_id=10717954 chat_id=None db_state = self.StateTable.get(chat_id=chat_id, user_id=user_id)

What I think is strange is that it is happening on read access. Edit: also happening on creation.

Setting is_volatile=True fixed it.

I created a Pull Request #444 for fixing that.

In the editor.ponyorm is the autosave enabled again?

That’s really bad for my workflow. I often make minor changes, quite quickly until I get what I have envisioned, which I don’t want to snapshot. At the end of that I get a version which I copy in my project, to have under version control. Usually I don’t remember to open up the editor one more time just to save a snapshot. If there has passed quite some time and I open the editor up again, that most of the time implied that the model there was worth a snapshot, but I opened up the editor, to make changes I need now, and that’s what I gonna do, without thinking about doing a snapshot of the old version before changes. Here is how it typically goes: - Open existing schema - Make changes - Remember „Oh, I should have made a snapshot of the old version“ - Open same diagram in new tab - Create Snapshot of version before - Close Tab - Save (overwrite) the version with just made changes. Obviously, with autosave it’s not possible to get back to that earlier version.

Maybe you could keep autosave (I.e. browser crashed, halp) and manual save (I.e. I’m happy with the diagram) two separate things?

In the editor.ponyorm is the autosave enabled again?

Or is that happening when I delete an Entity?

Can confirm, it is saving without my consent when I add/remove entitites, making my previous changes forever lost

:(

I accidentially removed an entity.

Now it’s gone

I accidentially removed an entity.

(He he, Don’t tell the Catholics)

😭

Hi, what do you think would be correct UI behavior?

Don’t save unless I press save?