#pyramid

there's one hideous problem with toml

no nulls

the dev says "well, just omit the value from the doc, and have the code assume a missing value is a null"

but because of this you can't roundtrip json to toml and back

and "add special handling to all your code reading these files" is a lousy decision

does anyone implement the CQRS paradigm with python web development? specifically, driving development through Models / Controllers (views) really only gets us so far. I've seen the pyramid_services concept, and i think that's a step in a health direction, but I'd love to get some architecture thoughts here.

I found needing to put quotes around everything in toml to be pretty annoying... seems like it could easily have a null value as a result of that restriction

dfee: well, I've using something beyond pyramid_services, but it would be something that Fowler nevertheless abhors (a business layer with anemic domain model lol)

ztane: i can imagine how to do to that quite easily :)

i'm not entirely certian what he means by anemic domain model...

that seems to be the solution pyramid_services offers too, anyway.

either an anemic domain model, or a "fat model" can exist, but i'm not clear of how to build orthognal to that... without doing something like pyramid_services, or by abstracting a service layer that is really just the C in CQRS.

SQLA seems to handle the query bit fine. i've read a bit of zzzeek's stuff, but it appears he believes that SQLA models qualify as domain models. (the first half of: http://techspot.zzzeek.org/2010/12/12/a-tale-of...)

I am using the SQLA models as anemic domain model or most of it

good morning

morning

morning

morning

hey stevepiercy

I am trying to understand how testing works with testing.setUp

So my test is using a dummy request and I access the view function directly. I use the testing.setUp(settings=my_settings) without passing any request.

How the threadlocal.get_current_registry() returns <Registry Testing> ?

I mean how does it know that there is a request happening without me passing the dummy request to setUp(request=request, ...) ?

evagelos: I would *guess* that testing.setUp() sets up one default request automatically

moo-_-: thanks i just thought maybe another fixture(using pytest) sets it up, I mean i probably have conflicts :/

evagelos: I would suggest avoiding thread locals, even in tests

yeah i know i read that i just wanted quickly to say my problem

moo-_-: i saw in the setUp class that it passes the argument directly without any checking like: config.begin(request=request)

exept if in the begin instantiates a dummy one ill have to look that too :|

Ok so I am in the middle of my test and while my config fixture returns the settings(config.registry.settings) when i call the threadlocal doesnt :/

I mean the threadlocal.get_current_registry().settings is None