archonic / limestone Goto Github PK

Doo eet.

Update sprockets

After running the following command:

docker-compose run web yarn install --pure-lockfile

I get the following error:

Step 7/15 : RUN bundle install
 ---> Running in 92d984434d85
Your Ruby version is 2.7.1, but your Gemfile specified 2.7
ERROR: Service 'web' failed to build: The command '/bin/sh -c bundle install' returned a non-zero code: 18

When checking my ruby version:

limestone · master ⟩ ruby -v
ruby 2.7.0p0 (2019-12-25 revision 647ee6f091) [x86_64-darwin19]

I know I must be missing something obvious. But since this project uses docker, my local ruby version should not matter.

@archonic Love the project! Taking some time today to play around with it and I needed to do a bit of digging before I figured out I had to add the public_registration feature in Flipper - we should update README.md with some instructions so new Users can register for the site

Known issues include:

• Double compiling. Saving a file has the webpack service compile, and then refreshing the page has the website compile.
• Making changes to imported files doesn't make a difference to bundled output unless application.js or application.scss is changed. (this is normal due to how webpack watch works)
• Including /app/assets stuff in /app/javascript seems to be non-standard (is it?). It can't process requires so there isn't much point. (correct!)

Asides from these config issues, it's been hard translating example webpack configs into a webpacker compatible environment.js file. The abstraction so far has been hurting more than helping.

Apparently I've been running into some trouble with Docker but I'm certain it's not that complicated. My experience would tell me this is just a conflict of ports between my own local machine and Docker's redis and postgres containers.

I'm pretty lost as to where I would even begin to change settings for which port to use on compose. I have to rebuild the bundle each time I try something new on a new configuration.

ERROR: for squad_redis_1 Cannot start service redis: driver failed programming external connectivity on endpoint squad_redis_1 (c478b03bbdd6142824f4301b241c70a3c69fb9ccc6cca2c3dabf36315ed12a46): Error starting userland proxy: listen tcp4 0.0.0.0:6379: bind: address already in use

ERROR: for postgres Cannot start service postgres: driver failed programming external connectivity on endpoint squad_postgres_1 (83ab1031e351424120617a313502a267523ee2a52d37ee29bde5bb5513fa5521): Error starting userland proxy: listen tcp4 0.0.0.0:5432: bind: address already in use

With RUN yarn --pure-lockfile in the Dockerfile, setting up a new environment should be possible with just docker-compose up --build. For some reason docker-compose run webpacker yarn install --pure-lockfile is required.

some of the credentials are store on the .env file, what is your opinion about moving them to use the rails credentials:edit ?

I would be up to do it.

With the last webpacker upgrade, it stopped reloading the page after changes were made to files in the pipeline.

Step 4 of the setup process triggers the following error. ERROR: No such service: webpack

Although running step 5 does seem to work to an extent it fails at later stages.

I suspect it might be related to step 4 not getting completed properly but I might be wrong. Any suggestions.

Note: I am on Mac OS Mojave. Docker & Postgresql services are running
@archonic

See attached image - more my problem as a result of having a transparent profile pic, but should be addressed. I think if a Gravatar pic is detected, we should adjust the CSS to include a sane background color option with no initials.

Realized after provisioning the Docker containers that you need to have STRIPE_* environment variables defined in order for the docker-compose exec web rails db:setup to succeed.

We should add a note to README.md detailing this requirement 👍

The pay gem is mature enough for production use. Integrate it!

This also requires changing the receipt strategy and the Role behaviour. In fact role should be removed - billing status should rely on Pay methods.

Minimal working Kubernetes deploy instructions would be handy in a wiki article. Especially if it works with Codeship or CircleCI.

Expectations:

• SSL (probably using CloudFlare)
• Nginx ingress
• Cloud SQL service assumption (no Postgres container)
• Redis service? Trying to create my own HA Redis didn't go well. Should we assume people want to use MemoryStore or AWS Redis?
• Containers? Sidekiq and Rails on the same container? Separate containers is very RAM heavy and guarantees uncompetitive costs VS VPS and Heroku.

Running tests locally with docker-compose exec website rspec does not create plans in Stripe, but pushing to master and running tests on Codeship does. This is despite this line in app/models/plan.rb:

CreatePlanService.new(self).call unless Rails.env.test?

Not sure why this is happening.

Just need to update Pay to fix this. While working on this however it would be worthwhile to update to Pay 3.x which makes architectural changes.

Hi @archonic,

I'm experimenting with limestone to use in my latest venture and finding it pretty great so far!

I think your latest commits to the master branch may have introduced a bug with administrate assets, however my fork already has some diverging changes so maybe it's on my side.

I've added these lines to sprockets manifest.js to fix the issue.

//= link administrate/application.css
//= link administrate/application.js