I'm not sure SERIALIZABLE can solve this

I think you can solve this by using PostgreSQL INSERT ... ON CONFLICT (...) DO UPDATE ... statement https://www.postgresql.org/docs/current/static/sql-insert.html#SQL-ON-CONFLICT Pony does not support it yet, but you can write raw sql query db.execute("insert ...")

Hey guys, I'm wondering what's the best practice for password management when connecting to a database with pony? Leaving it in clear in the code seems dangerous, how do you guys deal with this?

Use config files

Is that the standard way? It doesn't feel great security wise, I feel like one of my users could accidentally share theirs.

use python-dotenv

U can set file attr 640 in unix

Oh that's great, didn't think to store them as environment variables, I'm gonna do that thank you both =)!

Hey guys, I'm wondering what's the best practice for password management when connecting to a database with pony? Leaving it in clear in the code seems dangerous, how do you guys deal with this?

Good practice to store "secret" data in environment variables

Hi guys. Can you recommend good swagger library for flask? Most of them seems abandoned.

Flask swagger 2

do you mean this one? https://github.com/soerface/flask-restful-swagger-2.0

do you mean this one? https://github.com/soerface/flask-restful-swagger-2.0

Permanent link to the soerface/flask-restful-swagger-2.0 project you mentioned. (?)

Yes

this project is not maintained

Hello, I typed up my question about PonyORM here - https://stackoverflow.com/questions/50875270/schema-migrations-in-ponyorm And there is an old, unanswered question here: https://stackoverflow.com/questions/50016739/mock-pony-orm-entity

Hi Terrence! Thanks for poining out, I'll answer the questions

Done

Как дел?)

Good )

Hi. I want to save 3 files en disk, they correspond to a Document entity. The thing is for each file I wanna be sure that writting to disk ANd to database succeeds. before_insert or after_insert can't do that. I was thinking a bout a context manager. Do you see another way to do it ?

this what I want to accomplish : python @db_session def save_document(path): some_files = [...] new_entity =[] for file in some_files: try: d = Document(*args) d.flush() savefile except orm.OrmError: raise someother error except IOError: rollback raise someerror else: new_entity .append(d) return new_entity

So you want it to be one atomic transaction?

Not necessarily. I do not want save file if database fails and i don't want save to database if file save fails. I don't want every thing fail, if only one file fails. the thing is that the underlying function is already inside a db_session

if flush() succeeds, does it mean that commit() will ?

I think the problem fundamentally is that either system (filesystem / database) can fail independently

You could write to database, then write to file, and then if writing to file fails, remove your database record?

or have a flag for file written in the database table

Flush can be roll backed

If exception raised

this is why I propose flush better than commit

I think it may be something like that: def save_file(filename, saved_files, ...): <do save> saved_files.append(filename) def save_document(path): files_to_save = [...] saved_files = [] new_entities = [] try: with db_session: for file in files_to_save: d = Document(*args) new_entities.append(d) d.flush() save_file(file, saved_files, ...) except: for filename in saved_files: remove_file(filename) raise return new_entities

is it a problem if save_document is already decorated by @db_session ?

In that case inner db_session will be ignored, so commit to the database happens too late. You can do manual commit: def save_file(filename, saved_files, ...): <do save> saved_files.append(filename) @db_session def save_document(path): files_to_save = [...] saved_files = [] new_entities = [] try: for file in files_to_save: d = Document(*args) new_entities.append(d) d.flush() save_file(file, saved_files, ...) commit() except: for filename in saved_files: remove_file(filename) raise return new_entities

thank you guys

Nice. "Hot chicks" join our comminity, not a spam for sure.

Alexander, It's a sexism. It may be a real developer, why do you now. Btw, some time ago your avatar was an anime girl with panties on her head. So, not be so fast with conclusions )

Hmm one after the other and all written in upper case. For sure spam.

Shame on me

Hi, everyone!

I'm trying to use buffer as PrimaryKey. PonyORM allows this, but select always fails afterwards.

from pony import orm db = orm.Database() class Bla(db.Entity): smt = orm.PrimaryKey(buffer, auto=False) db.bind(provider='sqlite', filename=':memory:') db.generate_mapping(create_tables=True) a = Bla(smt=str(0x01)*20) q = orm.select(p for p in Bla)[:]

Python 2.7 mumbles something about TypeError: writable buffers are not hashable

Does anyone have some thoughts about this problem?

Don't use mutable types as pk

but how I can use binary type as pk then?

Str in 2.7 is bytes string

In python 3 u can use bytes

I tried string, but I get 'can't convert to unicode' error

ValueError: Value for attribute MetadataGossip.sig cannot be converted to unicode: '8\x13D\xb2\xcd3\xf62\xa2\xbe\x83\xa3ca<3\xb3u\xe5M\xb9n\td\x9d}\xf8K\x8do\xad\xea+\x94\xfc\x90Tl...

i.e. I can't create the entry

but that's another problem

Are you using python 2.7

Python 2.7.14 (default, Jan 17 2018, 14:37:42) [GCC 6.3.0 20170406] on linux2

Try to set attr as bytes

I'm using bytes type for binary data on python 3

still cannot be converted to unicode

hm

Error looks like python try to convert bytes to Unicode with invalid encodings

funny thing, PonyORM seems to expect the value to be unicode. Therefore, It is impossible to store true binary data in str with Pony.

```

``` from pony import orm db = orm.Database() class Bla(db.Entity): smt = orm.PrimaryKey(str, auto=False) db.bind(provider='sqlite', filename=':memory:') db.generate_mapping(create_tables=True) a = Bla(smt=str('\x13D\xb2\xcd3\xf62\xa2\xbe')) q = orm.select(p for p in Bla)[:] ```

triggers it

PrimaryKey(bytes, auto=False)

Try this

same result

'cant convert'

From the docs: > Starting with the Pony Release 0.6 the attributes of str type in Python 2 behave as if they were declared as unicode attributes. There is no difference now if you specify str or unicode as the attribute type – you will have unicode string in Python and in the database.

yep. Unfortunately, the docs don't tell that its impossible to use buffer as PrimaryKey

Hi Grigori! At this moment, Pony cannot use binary attribute as a primary key, because, if I understand correctly, most databases don't allow it. In SQL binary values are typically stored using BLOB datatype, and BLOB column cannot be indexed. In some databases there are other binary types besides BLOB, like VARBINARY for small binary data. Maybe in the future we can allow primary key VARBINARY columns, if it is supported by the database. Right now an attempt to make binary attribute a primary key leads to obscure error, we need to fix the error text and make it less obscure Regarding Python types, Pony treat both str and unicode as the same unicode string type, for easier porting of applications to Python3: you can define attribute as str in Python2 application and it will work without any changes after porting application to Python3 For binary data, you can use buffer in Python2 and bytes in Python3. Unfortunately Python2 define bytes as an alias to str in a misguided attempt to make porting to Python3 easier. So you need to use buffer type in Python 2.7 for binary type

Alexander, thanks a lot for making it clear to me!