won't be caught either! only output in the console: Process finished with exit code -1073741819 (0xC0000005)

Consider using valgrind

class ViewFilterThreadClass(QThread): thread_finished = pyqtSignal(bool) def __init__(self, parent=None): QThread.__init__(self, parent=parent) self._dal: DataAccessLayerClass = DataAccessLayerClass() self._tw: QTableWidget = QTableWidget() self._net: str = "" self._view_from: datetime = datetime.now().date() self._view_to: datetime = datetime.now().date() self._view_mode: ViewMode = ViewMode.ACTIVATED self._reporter: str = '' self._reporter_func: str = '' self._line: str = '' self._train: int = 0 self._location: str = '' self._code: str = '' self._category: str = '' self._delay: int = 0 def set_properties(self, dal: DataAccessLayerClass, tw: QTableWidget, net: str, view_from: datetime, view_to: datetime, view_mode: ViewMode, reporter='', reporter_func='', line='', train=0, location='', code='', category='', delay=0, parent=None): self._dal = dal self._tw = tw self._net = net self._view_from = view_from self._view_to = view_to self._view_mode = view_mode self._reporter = reporter self._reporter_func = reporter_func self._line = line self._train = train self._location = location self._code = code self._category = category self._delay = delay def run(self): try: result = self._dal.get_filter_in_time_delta(self._view_from, self._view_to, self._net, self._view_mode, self._reporter, self._reporter_func, self._line, self._train, self._location, self._code, self._category, self._delay) except Exception as ex: print(ex.args) pass self._tw.setRowCount(0) query_result_to_table_widget(self._tw, result) self.thread_finished.emit(True)

In the last line the application crashes! No braik point and no try-except will help

I think my knowledge is not good enough for that.

What I don't understand, with Python 3.9 the code works!

I think you need the event loop started in the main thread to process signal emitting

Then I suspect the error in some different place, and the query execution just triggered a problematic code path execution

Yes, I am of the same opinion now!

like app = QApplication() app.exec_() or something like that

def _on_btn_view_all_clicked(self): """ with thread """ if self.frame_csv.isVisible(): return self.can_refresh = True # view progress window dt_view_from: datetime = self.dt_view_from.dateTime().toPyDateTime() dt_view_to: datetime = self.dt_view_to.dateTime().toPyDateTime() self.prg_wnd.show() if (dt_view_to - dt_view_from).days > 0: # self.setEnabled(False) self.prg_wnd.show() self._thread_running = True # initial and run thread self.view_all_thread.set_properties(self.dal, self.table_widget, self.net, dt_view_from, dt_view_to, ViewMode(self.table_view_index)) self.view_all_thread.start() while self._thread_running: qApp.processEvents() # self.prg_wnd.close() self.table_widget.resizeRowsToContents() # notwendig ! wird im thread (query_result_to_table_widget) nicht ausgeführt

In the main thread I'm waiting for the end of the rendering in the TableWidget : self._tw.setRowCount(0) query_result_to_table_widget(self._tw, result) Then I close the widget self.prg_wnd (Progress)

Hi @metaprogrammer ! It seems I found a bug in 0.7.15rc1. Query example that FAILS: select(s for s in Session if s.lastseen > datetime.utcnow() - timedelta(days=30)).first() Query example that PASS: select(s for s in Session if s.lastseen > datetime.utcnow() - timedelta(30)).first() I've digged into file pony/orm/asttranslation.py, line 210, function `postCall(translator, node)`: There args and kwargs of some function used inside query combined in the single list: `args = [ arg.src for arg in node.args ] + [ kw.src for kw in node.keywords ]` Further follows a check: if len(args) == 1 and isinstance(node.args[0], ast.GeneratorExp): return node.func.src + args[0] It will cause IndexError: list index out of range if we use ONLY kwargs inside function. If we change isinstance(node.args[0], ast.GeneratorExp) to isinstance(args[0], ast.GeneratorExp) then the error is not thrown.

Cool, thank you!

Hello everyone, I had this question for a long time. What is the most optimal data type for financial transactions in Pony and/or what approach would be recommend? Obviously float is out of the question due to rounding error. I've used Decimal for that purpose but its not very convenient to use (due to conversions and etc) and it has some unexpected bugs time to time. Another solution I had in mind for a while is using int with size=64 (max value 9223372036854775807) and it would be ok for storing values with 2 decimal places but it won't be enough for values with more decimal places (for example Bitcoin amounts has 8 decimal places and Ethereum 18 and etc). How you approach this issue in your projects while using Pony?

Any feedback would be appreciated :)

Hi @metaprogrammer ! It seems I found a bug in 0.7.15rc1. Query example that FAILS: select(s for s in Session if s.lastseen > datetime.utcnow() - timedelta(days=30)).first() Query example that PASS: select(s for s in Session if s.lastseen > datetime.utcnow() - timedelta(30)).first() I've digged into file pony/orm/asttranslation.py, line 210, function `postCall(translator, node)`: There args and kwargs of some function used inside query combined in the single list: `args = [ arg.src for arg in node.args ] + [ kw.src for kw in node.keywords ]` Further follows a check: if len(args) == 1 and isinstance(node.args[0], ast.GeneratorExp): return node.func.src + args[0] It will cause IndexError: list index out of range if we use ONLY kwargs inside function. If we change isinstance(node.args[0], ast.GeneratorExp) to isinstance(args[0], ast.GeneratorExp) then the error is not thrown.

This one is already fixed, but thank you anyways.

Hello everyone, I had this question for a long time. What is the most optimal data type for financial transactions in Pony and/or what approach would be recommend? Obviously float is out of the question due to rounding error. I've used Decimal for that purpose but its not very convenient to use (due to conversions and etc) and it has some unexpected bugs time to time. Another solution I had in mind for a while is using int with size=64 (max value 9223372036854775807) and it would be ok for storing values with 2 decimal places but it won't be enough for values with more decimal places (for example Bitcoin amounts has 8 decimal places and Ethereum 18 and etc). How you approach this issue in your projects while using Pony?

Could you share what Decimal bugs do you mean?

Hi! What do you mean? Can you provide an example?

For example I have a table "Client" with a field "name". I could have: class Client(db.Entity): _description_ = 'This table contains end-clients' name = Required(str, description='The first name of the client') And pony would do something like "COMMENT ON TABLE client IS 'This table contains end-clients';"

Hey @metaprogrammer recently I been stuck with following issue pony.orm.core.UnrepeatableReadError: Value of Balance.balance for Balance[171] was updated outside of current transaction (was: Decimal('50001.605299999996'), now: Decimal('50001.60530000')) Problem is, I have only one transaction in code. I assumed that it happened because I was get balance object twice inside this transaction but after caching it one after it got fetched this error keeps popping up. Any idea what could be the cause? Sql debug: https://paste.ubuntu.com/p/r4hy2G5QdV/ Full error traceback: https://paste.ubuntu.com/p/9nbYpMFvxW/

@alexey115 this bug happens to me time to time

Maybe it's just sqlite issue, but still

Also working with Decimal library feels bit janky

@alexey115 this bug happens to me time to time

How do you get the balance value? Do you convert it from a float value at any point?

Also working with Decimal library feels bit janky

It is, but no other Postgres type supports uint256. I believe that for full compatibility the PG type should be declared as numeric(78). Like it is discussed here https://stackoverflow.com/questions/50072618/how-to-create-an-uint256-in-postgresql

How do you get the balance value? Do you convert it from a float value at any point?

It was floating point issue with SQLite, so there is no way around it (for SQLite)

It is, but no other Postgres type supports uint256. I believe that for full compatibility the PG type should be declared as numeric(78). Like it is discussed here https://stackoverflow.com/questions/50072618/how-to-create-an-uint256-in-postgresql

Decimal is ok for the most part, but I with there was better solution :)

Thats why I asked, in case somebody found better way to do that

If you're not doing anything with it on db side you can store it as string.

Hi everyone. I've got a problem. Pony generates this sql: [2022-01-23 02:52:55,532] INFO: UPDATE `dbtrait` SET `data` = %s WHERE `dn` = %s AND `game_object` = %s AND (`data` = CAST(%s AS JSON)) [pony.orm.sql] [None] And throws: ... File "/home/xunto/.local/share/virtualenvs/tales_system_service-2lRreFFq/lib/python3.9/site-packages/pony/orm/dbapiprovider.py", line 64, in wrap_dbapi_exceptions raise ProgrammingError(e) pony.orm.dbapiprovider.ProgrammingError: (1064, "You have an error in your SQL syntax; check the manual that corresponds to your MariaDB server version for the right syntax to use near 'JSON))' at line 5") I'm using mariadb:10.7. It happens when I update this Model using upsert method: https://gist.github.com/xunto/c1871c9c6cbf28b2805ebda79b27d14f

Any idea what to do?

Hello. It seems that this syntax is supported in MySQL (https://dev.mysql.com/doc/refman/5.7/en/cast-functions.html#function_cast) but not supported in MariaDB (https://mariadb.com/kb/en/cast/#comment_3157)

Ah, interesting. Any workarounds for that? Also, sorry for late messaging, I hope I didn't wake you up. I expected to get answer in the morning.

I'm thinking about it. No worries, telegram isn't that aggressive in terms of notifications

Can you show the query itself?

You mean the upsert call?

Ah sorry, didn't notice that huge gist link

Why don't you try something like trait = DBTrait[db, owner] or DBTrait() trait.data = data ?

Still happens.

But isn't DBTrait[...] throwing ObjectNotFound if not found? So this should work for situation where record is missing?

Ah, I meant DBTrait.get(...)

Still happens with: @classmethod def upsert(cls, db_game_object: DBGameObject, dn: str, data: Dict) -> DBTrait: trait = DBTrait.get(dn=dn, game_object=db_game_object) or DBTrait() trait.data = data return trait

If any json updates don't work for MariaDB, I think the only way is to create some function with raw sql generation.

Or you can wait for another Alexander to reply. He sometimes creates some magical workarounds)

Hah, okay. Thanks, It would be great if he found a way. I'll stick to raw sql for now.

Any chances to get new release?

Any chances to get new release?

As far as I know recently there was release for Python 3.10

I'm preparing a stable release, should be released tomorrow

:O

https://pypi.org/project/pony/0.7.15/ 🤫

https://github.com/ponyorm/pony/releases/tag/v0.7.15

:D

Fantastic!

Now PonyORM supports Python 3.10. Also, now you can create SQLite memory database which is accessible from multiple threads by specifying :sharedmemory: "filename". (The previous way of creating in-memory database with a :memory: filename does not allow using the database from different threads)

Nice! Well done on overcoming that milestone! 💪

thanks for a wheel!

Sure!

Quick bugfix release, which fixes PostgreSQL-related issue: https://github.com/ponyorm/pony/releases/tag/v0.7.16

hi

Welcome if you are not a bot :)

Hello dear developers! I have found new bug related to isinstance function. For simplification consider reduced entities: class Session(db.Entity): id = PrimaryKey(int) user = Required('User') class User(db.Entity): id = PrimaryKey(int) type = Discriminator(str) sessions = Set(Session) class Worker(User): pass class Admin(Worker): pass Build a query to get staff online now: five_min_ago = datetime.utcnow() - timedelta(minutes=5) select(s for s in Session if s.lastseen >= five_min_ago if isinstance(s.user, Worker) It produces following SQL (at least for MySQL): SELECT `s`.`id`, ...bla bla..., `s`.`user` FROM `session` `s` WHERE `s`.`lastseen` >= %s AND `s`.`type` IN ('Admin', 'Worker') But it should be something like this: SELECT `s`.`id`, `s`.`user` FROM `session` `s`, `user` `u` WHERE `s`.`lastseen` >= 0 AND `u`.`type` IN ('Admin', 'Worker') AND `u`.`id` = `s`.`user` To workaround I can add for u in User if u == s.user to the query and change isinstance to check u instead of s.user. Another way is to know what exactly types of users and its heirarchy I have and check as strings if s.user.type in ('Worker', 'Admin'). But both ways are inconvenient.

Looks like a bug, thank you for reporting it!

When i add record - how i can get record ID?

Just search only?

obj = MyEntity(foo=1, bar=2) flush() print(obj.id)

obj = MyEntity(foo=1, bar=2) flush() print(obj.id)

спасибо

Hi all, How Can I get the value with highest occurrences for a specific column? Tried: SELECT some_str, COUNT(some_str) AS some_str_occurrences FROM some_table GROUP BY some_str ORDER BY some_str_occurrences DESC LIMIT 1; But if more than one value appear same no of times (which is max) I want to return empty result.

You can do something like from pony.orm import count, desc q1 = select( (obj.some_str, count()) for obj in MyEntity ).order_by(-2) # order by second column, descending results = q1[:2] # limit to 2 rows if not results: print("empty table") return elif len(results) == 1: row1 = results[0] else: row1, row2 = results if row1[1] == row2[1]: print("no highest value") return print("highest value is", row1[0])

You can do something like from pony.orm import count, desc q1 = select( (obj.some_str, count()) for obj in MyEntity ).order_by(-2) # order by second column, descending results = q1[:2] # limit to 2 rows if not results: print("empty table") return elif len(results) == 1: row1 = results[0] else: row1, row2 = results if row1[1] == row2[1]: print("no highest value") return print("highest value is", row1[0])

I want pure sql

CREATE VIEW some_str_count AS SELECT some_str, COUNT(*) as cnt FROM some_table GROUP BY some_str SELECT some_str FROM some_str_count t1 WHERE NOT EXISTS( SELECT * FROM some_str_count t2 WHERE t2.some_str != t1.some_str AND t2.cnt >= t1.cnt )

https://github.com/ponyorm/pony/releases/tag/v0.7.15

migrations ... ? 🥺

migrations ... ? 🥺

Now, after implementing Python 3.10 support, we can focus on migrations :)

Now, after implementing Python 3.10 support, we can focus on migrations :)

Hi! I'm getting the following error: pony.orm.core.UnexpectedError: Object Item[new:1] cannot be stored in the database. OperationalError: 1364 Field 'publish_date' doesn't have a default value 2022-01-31T10:07:02.257980132Z But, i understand that publish_date is getting the default value at database level: class Item(db.Entity): id = PrimaryKey(int, auto=True) item_id = Required(str) chat_id = Required(str) title = Required(str) price = Required(float) url = Required(str) user = Required(str) publish_date = Required(datetime.datetime, sql_default="CURRENT_TIMESTAMP") observaciones = Required(str) item = Required(str) What is wrong?

Hi! What database do you use?

Hi! What database do you use?

Im using Mariadb. Also, the Item query: new_item = Item(item_id=item_id, chat_id=chat_id, title=title, price=price, url=url, user=user, observaciones="prueba", item="prueba")

It is possible that you added the sql_default option after you had already create the tables, in that case it will not affect existing tables. You can apply change manually by executing the following command in db admin interface: alter table item alter column publish_date set default current_timestamp

no, because i dropped all the database before hehe, but give me a minute, i will drop all again

It is possible that you added the sql_default option after you had already create the tables, in that case it will not affect existing tables. You can apply change manually by executing the following command in db admin interface: alter table item alter column publish_date set default current_timestamp

You was right Alexander, i dropped the DB (now, yes) and it working

Good evening! I was testing the ORM's features when stumbled upon a problem. Here's my code. I'm trying to create a one-to-one connection, and for some reason this code just won't create the 'profile' column for the 'User' table. Do you have any suggestions on how I can fix this?

In one-to-one relationship, the column is created on a side with Required attribute. It is not necessary to create columns on both sides, one column is enough