🙂

They will also throw money at any kind of premium support package. Like every startup in Silicon Valley (hi from san jose by the way) has a paid AWS support tier, probably paid support tier for their major dependencies. The companies look at it like an insurance policy, their app is mission critical so why would you not pay easily $250-500 per month to be able to relax knowing the devs will help you in an outage 🙂

extremely good point

I think the kind of support you offer fore free in here, but via private messages / email would work

so you just need a webpage and some payment buttons :)

Great! We are working on such a page. We are going to provide one time and subscription payments via Stripe. Do you think we'd better use a service (e.g. Patreon, OpenCollective) or it doesn't matter?

Probably doesn’t matter, depends what’s easier for you. The most important thing is what exactly you plan to “sell” ;)

I see, cool

Start with stripe and add others if there is demand later

👍

I think the kind of support you offer fore free in here, but via private messages / email would work

The public responses still offer help for everyone else, too

pony.orm.core.UnrepeatableReadError: Value of X.y for X[493] was updated outside of current transaction (was: datetime.datetime(2019, 2, 23, 16, 22, 14, tzinfo=tzoffset(None, -18000)), now: datetime.datetime(2019, 2, 23, 21, 22, 14))

Is this the timezone confusing pony?

field is a datetime.datetime

PonyORM doesn't support timezones. You need to convert datetime to datetime without timezone, probably in UTC

That seems pretty reasonable, since time stamp in DB should always be UTC anyways

Ah ok thank you

@matthewrobertbell pytz can help you with the more tedious aspects

https://eureka.ykyuen.info/2015/04/17/python-convert-a-timezone-aware-datetime-to-utc-datetime-using-pytz/

I'm using dateutil which seems fine

Or that

hello everyone, I want use default value None, But I got "Default value for required attribute Position.exit_order cannot be None" Could someone please help me?

maybe something like it : aa = orm.Required(str, default=None)

you should use Optional

show us your code

the exit order should be Optional since Required means "could not be None"

I got it, Thanks

I got this, I dont know what happened

this is my code

https://docs.ponyorm.org/relationships.html#multiple-relationships-between-two-entities

each relationship defined using exactly two attributes. Like: Group.students <-> Student.groups Order.products <-> Product.orders Order.user <-> USer.orders etc In your class you have three attributes in Position which are pointed to Order, but only one attribute in Order which is pointed to Position

Now, in Position, I want use two order, I got error

Then you need to have two attributes in Order, like "entry_positions" (which refers to all Position which has entry_order linked to this specific order), and "exit_positions"

bingo thanks sincerely,

I think I missed some conception, it's very hard for me. Why does this work

Each relationship is two-sided. When A refers to B in some way, this means that B refers to A in some corresponding way Relationships are modeled in database with using of additional columns (for one-to-many relationships) or intermediary tables (for many-to-many relationships) You can search some information about Entity-Relationship model. Pony entities and relationships are similar to ER model with "crow notation"

yeah, Thanks again😘

Hi guys!

Does pony support "datetime - datetime" in queries?

TypeError: Unsupported operand types 'datetime' and 'datetime' for operation '-'

This is the context of the problematic part from my query `(datetime.utcnow() - car.created_at).total_seconds() / timedelta(days=1).total_seconds()` and in python this normally returns 0.5 for if half a day has passed

https://github.com/ponyorm/pony/issues/438 I've created an issue, if that's relevant

Thanks, think we need to add this operation. It seems currently only datetime - timedelta is supported

@sashaaero considered adding it three months ago, https://github.com/ponyorm/pony/issues/115#issuecomment-451453989 maybe he has already some progress on it :)

Thanks!

Meanwhile, trying to think of workarounds, I think the best way is to create float attribute days_passed_since_created_at and fill it before each query

Yeah, Anon . Since now Im fully focused on migrations (because we decide to force the implementation) all these things are delayed. Sorry. We understand that some of these small additions can be done quickly, but everytime we distract from the main goal we postpone it very hard. At the moment the most Pony users ask for migrations and we put all our efforts to this goal.

Migrations are very welcome, though!❤ 😊

Migrations 100% most needed! Thanks guys!

Hello Alexander! I've discovered a bug, that dropped some of my data in production. And I have no backup.

I have such entity: class Payment(EntityMixin,Entity): id = PrimaryKey(int, auto=True) paid = Required(bool, default=False) created = Required(datetime, default=<built-in method utcnow of type object at 0x7f9d1ea00d80>) updated = Optional(datetime) type = Required(str, 16) amount = Required(Decimal, 10, 2) term = Optional(Json, default={}) rate = Required(Decimal, 10, 4, default=Decimal('1')) metadata = Required(Json, default={}) method = Optional(PayMethod) user = Optional(User) servers = Set(Server)

I want to cleanup old unpaid payments: select(p for p in Payment if not p.paid and coalesce(p.updated, p.created) + timedelta(days=31) < datetime.utcnow())

This query translates into: SELECT "p"."id", "p"."paid", "p"."created", "p"."updated", "p"."type", "p"."amount", "p"."term", "p"."rate", "p"."metadata", "p"."method", "p"."user" FROM "payment" "p" WHERE ("p"."paid" OR (coalesce("p"."updated", "p"."created") + INTERVAL '744:0:0' DAY TO SECOND) < %(p1)s) LIMIT 10 {'p1':datetime.datetime(2019, 3, 24, 8, 24, 7, 469374)}

ACHTUNG!!! ("p"."paid" OR (coa.... ___________________________^ OR here, not AND

Hmmm.... really? I'll look into it immediately

But if I change query to: select(p for p in Payment if coalesce(p.updated, p.created) + timedelta(days=31) < datetime.utcnow() and not p.paid) all is fine: SELECT "p"."id", "p"."paid", "p"."created", "p"."updated", "p"."type", "p"."amount", "p"."term", "p"."rate", "p"."metadata", "p"."method", "p"."user" FROM "payment" "p" WHERE (coalesce("p"."updated", "p"."created") + INTERVAL '744:0:0' DAY TO SECOND) < %(p1)s AND NOT ("p"."paid") LIMIT 10 {'p1':datetime.datetime(2019, 3, 24, 8, 26, 20, 256209)}

This is postgres, maybe it works in another databases

Are you sure this is the same query? In SQL you have LIMIT 10 which is not present in your Python query

Is query really specified as a generator expression, or constructed step by step with lambda?

I tested on SQLite and Postgres abd it works correctly

Yes, this is right queries, I ran them in console and debugged with .limit() and .show()

No, query is not constructed step by step

>>> select(p for p in Payment if coalesce(p.updated, p.created) + timedelta(days=31) < datetime.utcnow() and not p.paid)[0:1] SELECT "p"."id", "p"."paid", "p"."created", "p"."updated", "p"."type", "p"."amount", "p"."term", "p"."rate", "p"."metadata", "p"."method", "p"."user" FROM "payment" "p" WHERE (coalesce("p"."updated", "p"."created") + INTERVAL '744:0:0' DAY TO SECOND) < %(p1)s AND NOT ("p"."paid") LIMIT 1 {'p1':datetime.datetime(2019, 3, 24, 8, 51, 46, 982292)} [] >>> select(p for p in Payment if not p.paid and coalesce(p.updated, p.created) + timedelta(days=31) < datetime.utcnow())[0:1] SELECT "p"."id", "p"."paid", "p"."created", "p"."updated", "p"."type", "p"."amount", "p"."term", "p"."rate", "p"."metadata", "p"."method", "p"."user" FROM "payment" "p" WHERE ("p"."paid" OR (coalesce("p"."updated", "p"."created") + INTERVAL '744:0:0' DAY TO SECOND) < %(p1)s) LIMIT 1 {'p1':datetime.datetime(2019, 3, 24, 8, 52, 4, 768955)} [Payment[16207]] >>>

Here is screen dump made just now

What Python version do you use?

Python 3.7.0 (default, Aug 22 2018, 09:16:41) [GCC 6.3.0 20170516] on linux

pony 0.7.9

It seems to be caused by bytecode changes in 3.7

Will fix it today and release new Pony version

Thank you

We tried to fix decompiler today, but It turns out to be pretty complex task. It is doable, but requires some time. I think it is better not to use Pony with Python 3.7 until the fix will be ready: conditions of type not A and B can be incorrectly translated. If possible, rollback to Python 3.6 Actually Python 3.7 is the first Python version where Pony decompiler may work incorrectly. This means that we need to change the way decomper works. I have a plan, but need about a week to implement it. This will be a high-priority task

We tried to fix decompiler today, but It turns out to be pretty complex task. It is doable, but requires some time. I think it is better not to use Pony with Python 3.7 until the fix will be ready: conditions of type not A and B can be incorrectly translated. If possible, rollback to Python 3.6 Actually Python 3.7 is the first Python version where Pony decompiler may work incorrectly. This means that we need to change the way decomper works. I have a plan, but need about a week to implement it. This will be a high-priority task

Good luck :)

Good morning

I am unable to download a diagram (as an image) from the editor

not a big deal but wanted to let you know :)

Hello. What's wrong with it? What's happening when you try?

get a 404 reply from 45.79.156.60