#pyramid

any example to integrate gevent-websocket for websocket app with Pyramid? In official repo, there is no demo/example with Pyramid. And I am getting hard time working on integration

sontek: I'll let you know on pyramid_celery, so far one issue is making it in fact share config VALUES :\ since celery only gets it's section

[pylons] bbangert pushed 2 new commits to master: http://git.io/v9NNgQ

pylons/master 616c241 Darren Yin: Access WebOb exceptions directly (instead of calling exception property)...

pylons/master 1f00579 Ben Bangert: Merge pull request #22 from dareonion/master...

[pylons] bbangert pushed 2 new commits to master: http://git.io/zV8Odg

pylons/master 4bd3e44 Ralph Bean: Changed add_options to options for nose plugin....

pylons/master 0614ebe Ben Bangert: Merge pull request #12 from ralphbean/master...

gavinc: I throw the pyramid settings in celery_app.conf['PYRAMID_REGISTRY']

gavinc: So the config values should be there in the worker

heh, okay so I may only have a doc patch ;)

oh, and I'll add some list handlers for the ini as well, since you have to specify which seralizers you want now in 3.1

ah... and the registry is parsed from the normal paster deploy so includes works vs just the INI parser for the celery part

ini parsing, huh

:P

inklesspen: oh much worse, celery by default wants a whole configuration module which it executes

yum

"If you’re using the default loader, you must create the celeryconfig.py module and make sure it is available on the Python path."

yeah

it is sadness

3.0 no longer does that I think

you can have a instance

nope, still does. but yeah you CAN specify a config module to execute in the command line

and you CAN as sontek has done replace the whole config process

gavinc: you can definately create your own instance

gavinc: I do not know how it hooks it up

you can either pass the commandline a App instance, or a config module

inklesspen: don't worry, once montague is ready pyramid_celery will switch over as well :D

Then people will get all kinds of celery config goodness

:D

I did a whole thing once where I did import of whatever was in celeryd and patched it to read configs

inklesspen: are people using the set semantics in the wild? Never seen it myself

from somewhere else

The celeryconfig.py is bad but its really hard to configure celery without actual python objects

it is all sad

for things like celerybeat

gavinc: 'set semantics'?

using set to replace global config values

oh.

is that in fact USED by anyone?

i don't know.

i thought the "use = config:foo.ini" stuff wasn't used by sane people either

but then someone in here said it's vitally important

+1 vitally important ;)

but I at least haven't seen set used in the wild anywhere

did you see the weirdness that results when you combine set with "use = config:foo.ini"?

I documented it here: https://github.com/inklesspen/montague/issues/4

it is pretty insane.

I've only used use = config:foo.ini because ini itself doesn't have a good inheritance model

it doesn't have _any_ inheritance model

just globals that aren't globals

we do use = for varying development inis with/without ssh tunnels with some local bits and some remote bits, but keeping the giant pile of other configuration the same

I saw a project the other day that added inheritance to ini but can't find it now

yeah hold on

Yeah, that one

sontek: i know https://github.com/axialmarket/betterconfig has something similar

(my ex-boss decided to reinvent ini parsing)

I mean I've done bad things with make + sed to fake it

gavinc: So what you are saying is we just document ini inheritance with make + sed and move on?

heh

tempting

noooooooooo

my ex boss also introduced makefiles

for python

:(

What's wrong with make files? ;)

they're not djb's redo

djb?

-.-

djb is the guy who invented qmail

We were using make a ton but recently started using click for building little clis instead

click makes me sad

decorators galore

unreadable

I love me some click, its so composable

Invoke is the only decent alternative

click is pretty ok

a function that has more more then 2 decorators basically is unreadable

goodwill, really? I used it a few times only, but it looked awesome

click and invoke sort of are for different things

can you show an example for how it becomes messy?

gah

that example there is already 1 step toward disaster

https://github.com/dstufft/virtualenv/blob/rewr... :D

explain! is the problem is with the decorators, or... ?

holy potato

that is unreadable

decorators

... that isn't even much shorter then arg parse

yeah that example dstufft has ... pretty much

its not that it is shorter

it is more that it is decorate the decorated the decorated the decorated the decorated the decorated the decorated

dstufft: They are sort of different, I wish invoke would split out its cli tools separate from its execution stuff

goodwill, I don't see why you see that unreadable.

I like click fwiw

so far anyways

ree: I guess I just do

dstufft: I actually use them together but feel dirty because invoke has a lot of duplication from what click does

and if it were readable, how would it be like?

although I'm kind of down on the fact it just bails out if you set LC_ALL=C

basically that is a method with like 40 lines

most of my methods are not that long

and again: the decorated the decorated the decorated the decorated