Memory leak

[littlej-x]

Describe the bug
Bellow is an example of usage of this library that causes a memory leak. Running this code will quickly use up the system's memory. With use of a memory profiler, I traced the source of the leak to line 323 of classbuilder.py if isinstance(val, ProtocolBase):. At this point val is "city1" and self is city1 <address_<anonymous> attributes: city> from bellow example.

Example Schema and code

import python_jsonschema_objects as pjo

schema = {
    "title": "person",
    "type": "object",
    "properties": {
        "address": {
            "type": "object",
            "properties": {
                "city": {"type": "string"},
            },
            "required": ["city"]
        }
    }
}


def person_from_data():
    classes = pjo.ObjectBuilder(schema)    
    Person = classes.get_class('person')
    Address = classes.get_class('person/address_<anonymous>')

    return Person(address=Address(city="city1"))


def main():
    while True:
        person_from_data()


main()

Expected behavior
No memory leak should result from use of this library.

[cwacek]

This is the same as issue #210. You’re seeing this because you’re calling build_classes inside the loop. Move that out and you won’t see the memory increase each time.

…

On Nov 17, 2022, at 9:39 PM, littlej-x ***@***.***> wrote:  Describe the bug Bellow is an example of usage of this library that causes a memory leak. Running this code will quickly use up the system's memory. With use of a memory profiler, I traced the source of the leak to line 323 of classbuilder.py if isinstance(val, ProtocolBase):. At this point val is "city1" and self is city1 <address_<anonymous> attributes: city> from bellow example. Example Schema and code import python_jsonschema_objects as pjo schema = { "title": "person", "type": "object", "properties": { "address": { "type": "object", "properties": { "city": {"type": "string"}, }, "required": ["city"] } } } def person_from_data(): classes = pjo.ObjectBuilder(schema) Person = classes.get_class('person') Address = classes.get_class('person/address_<anonymous>') return Person(address=Address(city="city1")) def main(): while True: person_from_data() main() Expected behavior No memory leak should result from use of this library. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.

[littlej-x]

Ah ok, thank-you.

[littlej-x]

For my understanding, why then does this example not cause memory usage to increase.

import python_jsonschema_objects as pjo

schema = {
    "title": "person",
    "type": "object",
    "properties": {
        "address": {
            "type": "object",
            "properties": {
                "city": {"type": "string"},
            }
        }
    }
}


def person_from_data():
    classes = pjo.ObjectBuilder(schema)    
    Person = classes.get_class('person')
    Address = classes.get_class('person/address_<anonymous>')

    person = Person(address=Address())
    person.address.city = "city1"

    return person


def main():
    while True:
        person_from_data()


main()

Where the line isinstance(val, ProtocolBase): is never reached
?

[cwacek]

That’s a good question. I’ll need to look at that in more detail.

…

On Nov 17, 2022, at 11:02 PM, littlej-x ***@***.***> wrote:  Why then does this example not cause memory usage to increase. import python_jsonschema_objects as pjo schema = { "title": "person", "type": "object", "properties": { "address": { "type": "object", "properties": { "city": {"type": "string"}, } } } } def person_from_data(): classes = pjo.ObjectBuilder(schema) Person = classes.get_class('person') Address = classes.get_class('person/address_<anonymous>') person = Person(address=Address()) person.address.city = "city1" return person def main(): while True: person_from_data() main() Where the line isinstance(val, ProtocolBase): is never reached ? — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.