#pyramid

what's the root dir?

found it ;)

${request.resource_url(request.root)} did the trick

ah... probably just request.application_url as well

right, just wanted the root

raydeo: getting paste.ofcode.org to be serverless using AWS Lambda

so been updating it to run properly with the latest pyramid and such

ah fun... what are you using for managed cassandra? or is it on redis?

also fixing up the cloudfront CDN to use HTTP/2 for the resources and fixing all that

redis, so I just threw up an elasticache instance

cool

https://c71kffsm50.execute-api.us-east-1.amazon... is the lambda deploy

also switching to HTTPS

raydeo: oddly, when I deploy the cookies stop working

benbangert: why not use the url generation that allows you to point to a remote URL?

So that in testing you can just use local, then in prod switch it?

benbangert: http://docs.pylonsproject.org/projects/pyramid/...

yea, I need to fix this up a bit

first task was moving it to serverless, next task is fixing it up more ;)

I see a single cookie being set. Named zappa

yea, so, API Gateway is funky. heh

it only lets a single cookie move between the outside and inside

so all the internal app cookies are consolidated into a single big zappa cookie

Zappa compacts/expands them across the boundaries

this is pretty slick so far though, just hitting 'zappa deploy' to stick a new version online

morning

hi guys - how do you solve CSS/JS problem in your applications? does every release ship precompiled/compressed statics or do you handle that during deployment somehow?

Ergo: I run bundling script during the deployment on the server

but I'd prefer to do this locally pre-deployment

Ergo: I'd do everything in the http server

But I guess it depends how much you're shipping.

http2 makes a lot of the older strategies redundant.

Charlie_X: well you still need to transpile es6 to 5 etc

Charlie_X: do you have some examples of how do you do it?

ie. how do you set up nodejs and buil pipeline?

I don't have anything that complicated.

You might want to talk to blaflamme when he's around.

But it's probably worth checking what mod_pagespeed can do now.

I personally think a lot of the JS stuff is wheel reinvention.

Charlie_X: they love that :-)

Apparently so.

Ergo: I would do transpile locally

Ergo: and then package all JS it to a wheel and install with pip

if Python is your thing

moo-_-: thats what i usually did

It wouldn't surprise me if mod_pagespeed can now manage "transpiling" but I've not needed to worry about that.

i need to use gulp

We don't have any "appified" websites.

i was thinking about shipping every part of UI/SPA as separate python package - then run build on it and copy to statics dir for httpd to serve

I reckon it's still worth looking up what you can just let the server do.

but it might be a bit tricky

Charlie_X: im about 95% sure its impossible since i need to multiple build steps with webpack/gulp/wthatever

Charlie_X: i still have to live with it :D

or i can go the way moo-_- suggests - but for python develop to work in production usually that would mean versioning build results

Maybe. I can't really comment on stuff I don't do myself. See remark above about wheels

which is kind of crappy

certainly sounds crapy.

well its certainly error proof

thing is most of time i dont do "releases" - i have one "rolling" app running via python develop

You can use your own PyPI though which takes a lot of the pain away.

but i still need to update the js

I assume this "appification" is why some webpages are getting so fucking fat.

TransferWise user page: 6 MB. Why?

Charlie_X: 6mb is excessive :D

Charlie_X: then again react source is 750kb :D

Charlie_X: angular

Charlie_X: but think positively,

subsequent page loads are fast

No, they're not. It's just badly written crap.

Developers should spend more time on lower specc'd machines and slower connections.

everytime you use a carousel, god kills a kitten

If only it were just carousels…

Bedtime for you, young man!

this is the only reason for carousels

there are no other good reasons

Charlie_X are you going to any conferences this year?

Hopefully. Just don't know which ones.

i plan on going to http://ploneconf.org/

That might be possible.

i went to my first ploneconf last year, and i think it was better than any pycon i attended

I had hoped to be able to get to that but didn't have funds at the time.

Difficult comparison: domain specific stuff is generally going to be better. But I also think PyCon has lost its way a bit.

PyCon is all about big data, number crunching, science, and django

This year… it's always been "fashionable". I miss the lack of depth as much as anything.

I hope to be planning a move to Oregon instead of attending PyCon this year

It's become a recruitment exercise as well.

true, if i were looking for regular employment, i might be interested

being self-employed and enjoying it doesn't provide much motivation to become a regular employee