Similar problems arise with e.g. datetime.datetime.

if you're using the Decimal type you really want to use strings

there's no point in it otherwise, you'll be missing out on precision which is kind of its entire point

Well, in JSON it's all a string, right? { 'd': Decimal('10.000000000000000001') } -> '{ "d": "10.000000000000000001" }' or -> '{ "d": 10.000000000000000001}' In both ways, no precision is lost. I would assume it will only be lost when a default decoder does '{ "d": 10.000000000000000001}' -> {'d': float(10.)} but not when I give an object_hook to json.JSONDecoder which converts numbers.

But you have a point: At least in Postgres Pony uses jsonb, so '{ "d": 10.000000000000000001}' will probably loose precision in the Database, as the database will interpret it as float.

Pony allows to query JSON fields in queries, like select(obj for obj in MyObject if obj.json_field[x][y] > z) In such queries we expect that JSON items have the same type in Python and in the database. If values can be converted to some other type when loading from the database it may be surprising and non-intuitive. Maybe, it is better to perform such conversion explicitly. Say, you know that some fields are stored as strings, but actually datetime or decimal values, and convert them to that type when you read that JSON value

Hm, ok, I see. I never understood the magic that translates these generators to sql. Thanks anyway, I will use the explicit way, converting them manually.

Hey, sorry to interrupt can someone help me with creating ERD on a database?

This is some part of python code in a project of me:

https://pastebin.com/raw/MFf3GzFf

Any idea how to make this code shorter? Not including the json inside each function?

It's not that related to PONY but the community here is that nice, so I ask...

if request.form['token'] != config.secret_token: return False else: return True Replace with return request.form['token'] == config.secret_token

Okay...thought I could include this check with lower lines of code, but thanks

Hello guys, I'm using Jinja + Pony to create and run sql scripts. Actually, the generated script runs properly if executed from the shell on MySQL, but it never finishes when I run the db.execute(<script>) command. Is there any character limitation? Example: DROP TRIGGER IF EXISTS test_trigger_update; DELIMITER @@ CREATE TRIGGER test_trigger_update BEFORE INSERT, UPDATE ON test FOR EACH ROW BEGIN SET NEW.updated = CURRENT_TIMESTAMP; END @@ DELIMITER ;

I think you need to execute four separate commands commands = [ """DROP TRIGGER IF EXISTS test_trigger_update;""", """DELIMITER @@""", """CREATE TRIGGER test_trigger_update BEFORE INSERT, UPDATE ON test FOR EACH ROW BEGIN SET NEW.updated = CURRENT_TIMESTAMP; END @@""", """DELIMITER ;""" ]

for command in commands: db.execute(command)

I have removed the first command (the drop trigger) + run it as suggested and now I get an error when I run the DELIMITER command, so I think the DELIMITER is not supported.

Try without DELIMITER command

Maybe it is unnecessary

I couldn't make it work, but I see it should be unnecessary. https://github.com/PyMySQL/mysqlclient-python/issues/64

Any way to get the "query" and check into a one line if statement? https://pastebin.grey.pw/?6e32153ecdd11688#x+z65JcDdm9E32e2JeUUJP8lAfNKzQSr0WHASwb+nmA=

if interna_id is primary key and you are sure that the object with this id exists in the database, you can write if Server[3].is_busy:

It's NOT the primary key

Then you can write: if Server.get(interna_id=3).is_busy: But your current code looks more clear

Then you can write: if Server.get(interna_id=3).is_busy: But your current code looks more clear

Also it looks happy =3

Thanks I prefer short code :)

How to get the inserted object JSON serialized? https://pastebin.com/raw/VuuWXaYi

Try to use to_dict method: https://docs.ponyorm.com/api_reference.html?highlight=to_dict#Entity.to_dict

Okay, much easier :) http://stackoverflow.com/a/34765354/7878100

Since I need to pass the "data" somehow to celery

Any idea where things go wrong? https://puu.sh/vHuZi/63827b7051.png

I've changed from Optional back to Required in the meantime.

Any idea where things go wrong? https://puu.sh/vHuZi/63827b7051.png

Well, I think that config module does not have daemon_id variable :)

Of course it has... https://puu.sh/vHvDZ/cee6006e46.png

I think you have circular import. When entity definition processed, config module is not fully loaded yet

Also, right now you specified a constant (which is not defined yet). You can use callable instead, which will retrieve actual value of daemon_id later, at the moment of object creation. It will help with curcular import, and also may be useful if daemon_id value can be changed dynamically later daemon_id = Optional(int, default=lambda: config.daemon_id)

So just to make sure I understand, this will make python load the value before it's used?

Without lambda it will remember the value at the moment of entity definition. With lambda the default value will be calculated at the moment of object instance creation

By the way, are python related questions allowed too, like best practices?

I wouldn't mind

Just do it Can be deleted anyway xD

Well here we go

Any suggestion how to monitor a file for changes using Python? I'm using console commands to update some software. The output of this software is written to a file. Now I start a worker process (using celery), this worker now needs to "monitor" the file and check if it contains "Success", how to do so using Python?

There actually are some approaches to monitor file changes, some relying on the system used, the operating system which they are more efficient as polling would be

So the worker code needs to check the file again and again for changes, only when the "Success" part is found it will stop (of course I can include a timer, waiting 5 minutes, if still no "success" in file, it will notify me about failed updated).

Any links for this?

As I am on the bike right now, and it starts to rain, probably you googling makes mire sense xD

Already done, all I've found until now is related to windows by hooking into win32 API functions, but since I run my code on LINUX doesn't make sense to me :)

There must be a linux one too

Let me DuckDuckGo that for you: 🔎 python linux listen for file system changes

Pyinotify?

found it too atm, seems like it's perfect for my need.

tail -f file.txt | grep Success?😄

@Romet since the update command itself takes some time, using this wouldn't be helpful

Sometimes it takes 5 minutes, in case of good bandwith speed it may take 1 minute until the textfile contains "Success", so I need somekind of "watchdog" to monitor the file.

Alexsander hi, I checked ponyjs... Is it main realization of rest? Will you plan to use KO into library or not?

hey, all

Hi Maxim, currently we use PonyJS on fineartbiblio.com, but didn't release it officially yet. At this moment PonyJS uses KnockoutJS observables. As it turns out, KnockoutJS has some drawbacks - its templates are hard to maintain, and Knockout observables are slow. But we like the idea of observables in general, and want to re-implement PonyJS using another implementation of observables. PonyJS is not REST-like, it allows to serialize multiple objects into a single data packet, use it on JS side, and then send changes back to Python.

Thanks, for fast anwser :)

Hi Maxim, currently we use PonyJS on fineartbiblio.com, but didn't release it officially yet. At this moment PonyJS uses KnockoutJS observables. As it turns out, KnockoutJS has some drawbacks - its templates are hard to maintain, and Knockout observables are slow. But we like the idea of observables in general, and want to re-implement PonyJS using another implementation of observables. PonyJS is not REST-like, it allows to serialize multiple objects into a single data packet, use it on JS side, and then send changes back to Python.

Is the search based on sql too?

With PonyJS we select objects on the backend side using standard Pony queries

So it is a `text in <field>` query? How is that field indexed?

You mean full text search? We use standard PostgreSQL search capabilities for that. Table has additional indexed column of tsvector type which value is calculated by trigger as a concatenation of all other fields. And then we add raw_sql fragment to the query in the way similar to that: search_text = request.args.get("search_text") query = query.filter(lambda: raw_sql( "obj.fts_column @@ plainto_tsquery($search_text)" ))

https://www.postgresql.org/docs/current/static/textsearch.html

Uh, cool.

Somehow still didn't had the speed I needed

First time I use relations inside pony. Each server has a "default_map" attribute, with the id of the map. Now how to link them? I tried this: https://puu.sh/vPSZD/b7e1f993e4.png

But when running my app I get: raise exc pony.orm.core.ERDiagramError: Entity definition CounterStrike_Server_Go_Maps was not found

Get a little further, but using the generated code from the Pony Web based generator now I get: https://puu.sh/vPUXC/c7172a7922.png

1) All related entities should belong to the same database. If you want to split entity definitions betweens several files you can define db object in one module and then import it into another module. 2) When you define an attribute of Set type it will keep a collection of related entities. It is better to name that attribute in a plural form, like items, students, orders, etc., because when you retrieve a value of attribute it will be set of objects. 3) Any entity should have a primary key. If you don't define any primary key, Pony will create a default integer primary key with name id. In your entity you didn't define any primary key and the name id is already taken by another attribute. You should rename id attribute of Map entity to something like servers

omg

pony for javascript

sorry it took me a while to read that

@akozlovsky 1.) Yes it's the same database, object is initialized in config (I store SQL params inside this file, just to make sure I can edit settibgs easier). 3.) I've already creates the SQL code before, so how to go now? My 'Counterstrike_server_go_maps' table should just look like this: |id|name|display_name|

All I try to setup is a simple relation, one server is related to exactly one map. one map is related to many servers, but the relation from map to server isn't needed, because I only query from the server model.

Pony always creates accessors in both directions on a relation.

So how to setup this by using 1:1 relation (one server is related to exactly one map) & (1:n -> 1 one map is related to many servers) by using the pkey of map as relation id?

Like this: https://editor.ponyorm.com/user/ptmcg/Server_to_map

screenshot from designer of creating 1-many relation