This is a follow-up question to Ember.js app, Parse throws 404 for any path but root:
I have built an Ember-CLI app that I now want to host on Parse.com.
I can't figure out how to actually run the app server on Parse. I confess I don't understand how to serve an Ember-CLI in production from any server.
Do I run "ember server" on Parse somehow or do I need to make my own Express server for this to work?
Run parse new in some temp folder to create the correct config files for your existing parse app
Edit .ember-cli file and add "output-path" : "dist/public" to deploy your code to dist/public
Copy the folders code and config from your temp folder in step 1 to dist/code and dist/config
In a terminal window cd dist and parse deploy
Edit environment.js in your ember config folder and change locationType to hash
Related
So I have a Django project that acts as a backend REST server, and I would like to integrate that with an existing React project that makes AJAX calls to this REST server. Again, both of these projects already exist, I just want to connect them. Let's say the root of my Django project is DJANGOROOT and that is where manage.py is located. My Django project has a couple of Django apps that are part of the project. Let's say my Django project is called "mydjangoproj", the two Django apps are called "myapp1" and "myapp2" and the React project is called "myreactproj". Each of these two projects (the Django project with the two apps, and the React project) is stored in a separate GIT repo and to connect them on my server I have the following directory structure:
DJANGOROOT/
DJANGOROOT/mydjangoproj
DJANGOROOT/mydjangoproj/settings.py
DJANGOROOT/mydjangoproj/urls.py
DJANGOROOT/mydjangoproj/static/
DJANGOROOT/mydjangoproj/templates/
DJANGOROOT/myapp1
DJANGOROOT/myapp2
DJANGOROOT/myreactproj
DJANGOROOT/myreactproj/package.json
DJANGOROOT/myreactproj/src
DJANGOROOT/myreactproj/src/index.js
etc
Note that the React app root folder is inside DJANGOROOT and is parallel to my django project and app folders.
The React app uses an npm module called react-scripts (see https://github.com/facebook/create-react-app/blob/master/packages/react-scripts/template/README.md#npm-run-build) that has a build script that will build all the stuff in DJANGOROOT/myreactproj/src (staring with index.js) and will put all this in a folder called DJANGOROOT/myreactproj/build. When I run the build (using "npm run build") it fills up the folder DJANGOROOT/myreactproj/build like this:
DJANGOROOT/myreactproj/build
DJANGOROOT/myreactproj/build/index.html
DJANGOROOT/myreactproj/build/{several other files}
DJANGOROOT/myreactproj/build/static
DJANGOROOT/myreactproj/build/static/css
DJANGOROOT/myreactproj/build/static/js
DJANGOROOT/myreactproj/build/static/media
So it appears that to connect Django to my React app, I have to find a way to have the root path for Django (mydomain.com/) point to DJANGOROOT/myreactproj/build/index.html. I understand that I must create a "home" template that is launched by the Django view that runs at mydomain.com/, and that home template has to launch the React app something like this:
{% load render_bundle from webpack_loader %}
{% render_bundle 'main' 'js' 'DEFAULT' %}
But I'm not sure how exactly I need to tell Django where to look for main.js (or is it index.js) and I'm not sure how to get Django to see all the static stuff in DJANGOROOT/myreactproj/build/static.
I also ran "npm list webpack" inside DJANGOROOT/myreactproj to get the version of webpack, and it shows this:
└─┬ react-scripts#1.0.11
└── webpack#3.5.1
I assume this means that webpack 3.5.1 is being used and is somehow bundles with react-scripts v 1.0.11
Also note that the Django app is already running under HTTPS under NGINX and UWSGI, so this is a real server not a dev server.
Can someone who has been through this process shed some light on what I need to do to hook this all up properly?
If your react app is using django just to get data via REST, why you want to use react in django templates? I think preferred approach here is separation not integration. Create standalone react app with its own server (both can run on one physical server of course) - this way your react app and django app will communicate only by rest calls and nothing else needed. If you have some assets in django app - move them to frontend react app.
I just started a new little project for learning purposes and I want to try out Vue js with Django (with DRF). I'm trying to use vue loader (webpack-simple template).
The problem is that I don't know how to synchronize npm run dev and python manage.py runserver. I don't know how to access a template that is rendered by django in webpack-dev-server.
I mean I have a template with django-template specific keywords like {% load static %} that is not handled by webpack-dev-server, obviously.
I know I can build it everytime npm run build, but that's kinda annoying and boring to wait for it everytime I want to make a little change.
In webpack, it's specified to run on a default index.html file, how can I make it work on the template that is actually rendered on 127.0.0.1:8000 with python manage.py runserver running? I know it doesn't make any sense to run 2 dev servers, but I don't know how to explain in other way.
Is there an alternative?
Thanks in advance for answers!
Run your Django server as normal. webpack shouldn't serve your files. It should just build them (use webpack development settings and webpack --watch) and let webpack put them in the static directory of your Django project, e.g.
// in your webpack config
output: {
path: path.resolve(__dirname, 'project/static/js')
}
That way Django can serve the files that are run through your webpack pipeline.
On top you can use the webpack live reload plugin and the live reload browser extension to auto-reload when your assets change.
When you are ready to commit your changes, build your files in production mode and commit the build files in the static dir.
I'm trying to test an Ember addon project by running ember server inside an addon project. This runs the dummy app. However I am not getting my less files automatically being build with the ember-cli-less functionality (expecting /app/styles/app.less):
Some files I have:
/addon-project
/app
/styles
/less
app.less
/dist
/assets
dummy.js
dummy.css
/tests
/dummy
However I am getting an error when I run ember server:
version: 0.1.15
Could not find watchman, falling back to NodeWatcher for file system events
Livereload server on port 35730
Serving on http://0.0.0.0:4201/
File not found: /app/styles/app.less
in any of the following include paths:
/Users/me/github/ember-addons/header/tmp/tree_merger-tmp_dest_dir-qEC1hODR.tmp
Error: File not found: /app/styles/app.less
You should try moving the "app.less" file up one level (out of your less folder so that it is in the "styles" folder).
I've set up a website using Ember-CLI and it is now ready for production, so we're looking for a way to keep it running permanently.
Right now we're using $ ember serve --port 80
but obviously this only works whilst we're logged in. I've used Forever before to keep a node app running but am not sure how to make this work with Ember CLI, as the 'ember serve' command obviously does more than just running app.js?
Any input would be appreciated!
Ember-CLI apps are NOT node apps, they are Browser apps, so you don't need anything special to serve them. To keep and Ember-CLI app running permanently, I suggest doing:
ember build --environment=production
This will perform the necessary build steps so that the code works in browsers (e.g, transpiling ES6 modules) and put the code in the build folder. It will also minify JS files and fingerprint all resources (this only happens when the environment is production).
All you have to to then is put the files inside the dist/ folder on a Web Server.
I suggest Apache or Nginx, but anything will work.
Edit
As Omair Vaiyani pointed out, this might not work in some servers because Ember-CLI
uses the locationType: 'auto' which defaults to 'history'. For that to work, you have to configure your SERVER to serve the ember app from all routes.
What I do, and server me well because I don't have control over the server, is to simply change the locationType to 'hash', which will generate URLs with hashes (http://myemberapp/#/myroute/myid) and will work with any server. Just edit the environment.js file accordingly:
module.exports = function(environment) {
var ENV = {
/* other stuf ... */
locationType: 'hash',
/* other stuf ... */
},
/* other stuff */
```
Basically I'm structuring my app similar to this GitHub project:
https://github.com/zackargyle/angularjs-django-rest-framework-seed
Is it possible to deploy both the backend and frontend onto a single PaaS such as Heroku/Elastic Beanstalk?
Having a separated REST backend and JavaScript frontend seems like a cleaner/more scalable way to do things rather than trying to mix them together like [django-angular]: (http://django-angular.readthedocs.org/en/latest/index.html/), or having a REST backend mix with the Django app like http://blog.mourafiq.com/post/55099429431/end-to-end-web-app-with-django-rest-framework
If it is not possible to deploy it easily onto Elastic Beanstalk, is there an easy way to deploy the Django backend onto Elastic Beanstalk, and AngularJS frontend to Amazon EC2/S3 with minimal configuration?
I realize there's a similar discussion before this: Client JS + Django Rest Framework
but it lacks more specific details.
I'm in the exact same boat with AngularJS as my client and django-rest-framework as my service. I also have the same type of git setup where the server and client code are siblings in the same repository. I don't have any experience with Heroku and I'm new to beanstalk but I was able to deploy my site and it's working on AWS beanstalk.
With beanstalk there are two ways I know of to deploy your code.
Use eb and git described here.
Works well if you want to push your source code directly.
Create your own zip to upload to beanstalk via the AWS management console. Amazon has a walkthrough on it here.
Route I chose so I can 'grunt build' my client and zip with server code before deploying.
I automated the zip creation using a python script. Amazon's walkthrough provides an example python zip. You have to structure it properly, mine looks roughly like this
app.zip
/.ebextensions/
/.elasticbeanstalk/
/app/ <-- my django-rest-framework project (settings.py, wsgi.py, etc.)
/restapi/ <-- my django-rest-framework application (my api)
/static/ <-- AngularJS results of 'grunt build' put here
/manage.py
/requirements.txt
I know you didn't specifically ask but the .config file inside .ebextensions/ took me way too long to get working. It can be formatted as YAML or JSON (can be confusing at first as every blog shows it differently). This blog helped me out quite a bit just be careful to use container_commands: and not commands:. I lost a few hours to that...
container_commands:
01_syncdb:
command: "django-admin.py syncdb --noinput"
leader_only: true
option_settings:
"aws:elasticbeanstalk:container:python:environment":
"DJANGO_SETTINGS_MODULE": "app.settings"
"aws:elasticbeanstalk:container:python":
"WSGIPath": "app/wsgi.py"
"StaticFiles": "/static/=static/"
"aws:elasticbeanstalk:container:python:staticfiles":
"/static/": "static/"
"aws:elasticbeanstalk:application:environment":
"AWS_SECRET_KEY": "<put your secret key here if you want to reference from env variable>"
"AWS_ACCESS_KEY_ID": "<put your access key here>"
"AWS_S3_Bucket": "<put your bucket here>"
In the zip you create (if you follow the beanstalk guides on django) the client code in your /static/ folder is automatically pushed to s3 when you deploy.
This setup isn't perfect and I plan on fine tuning things but it's working. Here are some downsides I ran into that I haven't solved yet:
Since I put my client code in the static/ folder my site sits under mysite.com/static/. Ideally I'd want it to be served as the root at mysite.com with my django-rest-framework content under mysite.com/api/
If you use the self describing api on beanstalk by default the assets won't be pushed since they sit in your python directory and not with your source code.
UPDATE 4-17-2014
I further refined this setup so I no longer have to go to mysite.com/static/ to load my index.html. To do so I used a django class based view to map index.html to the root of my site. My urls.py looks like
urlpatterns = patterns('',
(r'^$', TemplateView.as_view(template_name="index.html")),
...
)
and in my settings.py I configured TEMPLATE_DIRS as follows
TEMPLATE_DIRS = (
os.path.join(os.path.dirname(__file__) , '../static').replace('\\','/')
)
I use ../static because my static directory is a sibling of my app directory.
The last piece was to update my Gruntfile.js so 'grunt build' prefixes all the relative URLs in my angular code with the static folder. I used grunt-text-replace for this. It's the last task that runs after my code is sitting minified in a /dist folder. The downside to this approach is I'll have to update this task if I ever add static content to a new subfolder besides scripts, bower_components, styles, etc.
replace: {
replace_js_templates: {
src: ['dist/scripts/*.js'],
overwrite: true, // overwrite matched source files
replacements: [{
from: /templateUrl:\s*"/g,
to: 'templateUrl:"static/'
}]
},
replace_index: {
src: ['dist/index.html'],
overwrite: true, // overwrite matched source files
replacements: [{
from: /(src|href)="(bower_components|styles|scripts)/g,
to: '$1="static/$2'
}
]
}
},
Now django will serve my index.html page but everything else in my /static/ directory can benefit from a CDN.