It should work: from pony.orm.examples.university1 import * populate_database() with db_session(sql_debug=True): students = select(s for s in Student if s.gpa > 3).prefetch(Student.courses)[:] for s in students: print(s.name) for c in s.courses: print(c.name) print() Note [:] after the query, it forces the query execution inside db_session. Without [:] of .fetch the query will be lazy and don't prefetch anything

hmm i tried, lot of thanks!

It should work: from pony.orm.examples.university1 import * populate_database() with db_session(sql_debug=True): students = select(s for s in Student if s.gpa > 3).prefetch(Student.courses)[:] for s in students: print(s.name) for c in s.courses: print(c.name) print() Note [:] after the query, it forces the query execution inside db_session. Without [:] of .fetch the query will be lazy and don't prefetch anything

DAMN xDD, my mistake was that i used Courses class in prefetch instead of Student.courses

Thank you very much! i recommend everybody to take a look pony orm, awesome

?

sorry, it is possible to make 3 level join in a select? I want to get Diary object and name attr of Person. Diary points Contact and Contact to Person Person - Contact - Diary select((d, p.name) for d in Diary for p in c.contact.person if blabla ) But pony says NotImplementedError: for p in c.contact.person

currently im iterating to get the value

just asking if its possible

This is because d.contact.person is not a collection, but a single object, so you cannot iterate over it You can just write: select((d, d.contact.person.name) for d in Diary if something) Alternatively, you can write it in reverse direction: select((d, p.name) for p in Person for d in p.contacts.diary if something)

Or you can write select((d, p.name) for d in Diary for p in Person if p == d.contact.person and something)

Or you can write select((d, p.name) for d in Diary for p in Person if p == d.contact.person and something)

DAMN

thanks again XD

Sure

is more simple than i thought

Good morning! There is a good pattern design to apply and work with ponyorm (or an another python orm)? Im relatively new on python, im searching on internet but just in case..

Hi! I don't know any specific design patterns for use with orm. PonyORM itself uses multiple design patterns, such as IndentityMap https://martinfowler.com/eaaCatalog/identityMap.html UnitOfWork https://martinfowler.com/eaaCatalog/unitOfWork.html LazyLoad https://martinfowler.com/eaaCatalog/lazyLoad.html QueryObject https://martinfowler.com/eaaCatalog/queryObject.html etc

Just asking to get a good logic design layers, i have a look to these links

When using Pony, with say Tornado, i instantiate the database in one file, declare the tables in another and Do all he manipulations in my other files. This allows one to pass the database object around as necessary and lets you selectively use particular models. Also it’s as close as I could get to using pony/tornado as one might use models in Django.

When using Pony, with say Tornado, i instantiate the database in one file, declare the tables in another and Do all he manipulations in my other files. This allows one to pass the database object around as necessary and lets you selectively use particular models. Also it’s as close as I could get to using pony/tornado as one might use models in Django.

And this design don't break the low coupling rule?

i mean

what i understand to good design is that is necessary to make every layer do one thing and relegate to another layers other things, to work with db i thought that is a good idea have an access db layer with all methods wich access to the entities and then all the files or ocmpoents access to that access layer

im a newbie, im just asking

xD

I think PonyORM is *already* that separate DB layer that abstracts database usage from application code. In my practice, any separation of additional db layers leads to serious code complications, and looks like an anti-pattern

Additional separation make sense when you need to pass information to another process, for example, to JavaSctipt front-end layer. Then you need to convert retrieved entities to JSON structures

Not sure what that is. But it, in combination with relative imports (Python > 3) allows you to pull in the models wherever it is necessary. So you can use “from . import models/database” everywhere within your package. It is a bit tricky to setup. I don’t have a public repo of this at the moment I’m afraid.

I think PonyORM is *already* that separate DB layer that abstracts database usage from application code. In my practice, any separation of additional db layers leads to serious code complications, and looks like an anti-pattern

okey, thats logic, is very abstract the access to the db, yep. I was just asking because pony is different as normal db workflow (without orm) xD

thanks a lot guys

Carel, regarding import of db/models, it is also possible to import just db object and then access entities as an attributes of db object. Sometimes it may simplify imports from pony import orm from models import db with orm.db_session: result = orm.select(p for p in db.Person)

Carel, regarding import of db/models, it is also possible to import just db object and then access entities as an attributes of db object. Sometimes it may simplify imports from pony import orm from models import db with orm.db_session: result = orm.select(p for p in db.Person)

omg xD

The drawback is IDEs (like PyCharm) don't understand what db.Person is and auto-complete Person attributes may not work

The drawback is IDEs (like PyCharm) don't understand what db.Person is and auto-complete Person attributes may not work

I'm waiting to http://mypy-lang.org/

i like the concept of static tiped in other languages, is a pain in the ass the object inspection in python imho

and, i have a last question, wich is better, pass the db object to threads class i launch or import the db on all the threads?

i see on the docs ponyorm is thread safe

I think there is no difference between importing and passing of db object, use what is more convenient

Also, do you write web application or desktop application?

mm, service-backend application?

in the next weeks i will integrate with flask

but now only is a service wich read/writes the db and make thinks with the data

@akozlovsky thanks. I’ll double check what I did in my code. I recall I needed to throw the dB instance about somewhat, hence the separate package structure.

@ask does pony have a tool to generate from raw sql to pony object?

To generate what? Or you mean to generate raw sql?

From raw sql into pony object

You mean generate entities from tables declaration?

Yes

I do not speak english well

https://github.com/abetkin/pony-inspect

https://github.com/abetkin/pony-inspect

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

The case if I have already make tables in sql languages and I want to try Pony and use pony query.Is there any tools to convert tables into pony class?

If I remeber correctly you can use this.

Ah .. let me check the link

If you okay with redeclare entities, you can specify table name for each entity by youself

If you okay with redeclare entities, you can specify table name for each entity by youself

I need a tool to fast converting

So take a look at pony-inspect

Yes I will. Is it being tested and ready to be used ?

It should work for simple cases, but entity declarations sometimes may be not ideal, for example instead of many-to-many relationship you will get intermediate entity. So you may need clean up entities manually at the end

Yes I had read it that python-inspect has limitations so I probably should specify tables into pony class manually

Is there a way to import existing pony database into pony ERD?

At this moment, no

Okay

Hi guys! ` File "<auto generated wrapper of process_channel_dir_file() function>", line 2, in process_channel_dir_file File "/home/vader/.local/lib/python2.7/site-packages/pony/orm/core.py", line 480, in new_func reraise(exc_type, exc, tb) File "/home/vader/.local/lib/python2.7/site-packages/pony/orm/core.py", line 468, in new_func commit() File "/home/vader/.local/lib/python2.7/site-packages/pony/orm/core.py", line 349, in commit transact_reraise(CommitException, exceptions) File "/home/vader/.local/lib/python2.7/site-packages/pony/orm/core.py", line 319, in transact_reraise reraise(exc_class, new_exc, tb) File "/home/vader/.local/lib/python2.7/site-packages/pony/orm/core.py", line 343, in commit primary_cache.commit() File "/home/vader/.local/lib/python2.7/site-packages/pony/orm/core.py", line 1727, in commit cache.database.provider.commit(cache.connection, cache) File "/home/vader/.local/lib/python2.7/site-packages/pony/orm/dbproviders/sqlite.py", line 315, in commit provider.transaction_lock.release() CommitException: error: release unlocked lock` This error appears when I return a new ORM object creater @db_session decorated function to another @db_session decorated function. Anyone knows how to handle this?

Hi guys! ` File "<auto generated wrapper of process_channel_dir_file() function>", line 2, in process_channel_dir_file File "/home/vader/.local/lib/python2.7/site-packages/pony/orm/core.py", line 480, in new_func reraise(exc_type, exc, tb) File "/home/vader/.local/lib/python2.7/site-packages/pony/orm/core.py", line 468, in new_func commit() File "/home/vader/.local/lib/python2.7/site-packages/pony/orm/core.py", line 349, in commit transact_reraise(CommitException, exceptions) File "/home/vader/.local/lib/python2.7/site-packages/pony/orm/core.py", line 319, in transact_reraise reraise(exc_class, new_exc, tb) File "/home/vader/.local/lib/python2.7/site-packages/pony/orm/core.py", line 343, in commit primary_cache.commit() File "/home/vader/.local/lib/python2.7/site-packages/pony/orm/core.py", line 1727, in commit cache.database.provider.commit(cache.connection, cache) File "/home/vader/.local/lib/python2.7/site-packages/pony/orm/dbproviders/sqlite.py", line 315, in commit provider.transaction_lock.release() CommitException: error: release unlocked lock` This error appears when I return a new ORM object creater @db_session decorated function to another @db_session decorated function. Anyone knows how to handle this?

what?

you make a dbsession in the scope of another dbsession?

no. I return object from one method to another.

Both methods a wrapped in db_session

And this happens only if I call some_binding.exists(foo=bar) before return

i.e. something inside .exists releases the lock

and it happens on .get too

but does not happen on .select

Do you have an example?

it's a long one, in our codebase. I'll try to make a minimal now.

Hi Grigori! Do you use async libraries like Tornado in your code?

Twisted, that is

btw, I can't reproduce it cleanly from bottom up

but if I run any select before exists, the bug goes away

and it is 100% reproducible on our code base

nothing asyncrhonous is used in this case. It's just unit test directly calling some method.

both the unit test and the method are wrapped in db_session

So, no async functions are decorated with db_session?

no

I need some time to understand what's going on

I kinda minimized it

the outer function: ` @db_session def test_bug(self): t = self.metadata_store.TorrentMetadata(title='test') t.delete() self.metadata_store.bug( ` Inner function: ` @db_session def bug(self): self.Metadata.exists(signature="123") return self.TorrentMetadata(title='dfdfd') `

however, I still can't reproduce it cleanly without our stack

this delete is required to trigger the bug

if I insert select(...)[:] before exists, bug goes away

yay!

I got a minimal example!