JSON.stringify doesn't correctly escape some strings (such as </script>) leading to potential client code injection. The use of a library such as serialize-javascript is recommended.

Vulnerability description:
http://benalpert.com/2012/08/03/preventing-xss-json.html

https://github.com/cerebral/cerebral-website/blob/v2/app/markdown/doc_servercontroller.md should be updated.

Now http://www.cerebraljs.com/#/6/1:

• link doesn't say where it would lead
• could lead in other place after rearranging in menu

After http://www.cerebraljs.com/#/signals/input

• link say where it would lead
• would lead to the same place even after rearranging in menu

Code examples with horizontal scroll does not work for some strange reason

The instructions for the tutorials say to go to localhost:8080 in your browser to view the application. It seems to actually be running at localhost:3000

See github for example :)

Check and update testing documentation with latest changes.
At least it contains old action signature.
PR are welcome.

Rationale

With many API and best practice changes it can be very confusing for users to figure how to do things if they're not on the latest versions of cerebral and related packages. We should maintain documentation and tutorials for all releases of cerebral, so we can always point users to docs for their versions.

This will also help us make changes and releases without confusing or disturbing people who can't always upgrade to the latest.

Implementation Steps

• create a heroku instance for each version of the website
• each version will have links (dropdown of version numbers probably) to each heroku instance of the website. Nothing fancy...just point to the heroku url
• Generate Github markdown docs in SeverController
• For each instance, map client side document filename (doc_operators) to a specific git commit:

{
  "doc_operators": "https://github.com/cerebral-operators#cdndd33jdfeeed<SHA>
}

• ServerController will fetch github docs and then render them to an HTML string sent to the client.
• This allows search to "just work" for each documentation version/instance

we can make to deploy website from ci. it could collect all docs from correspondent github repos instead of dublicating documentation both on repo and website. it should simplify support docs in actual state.

see bem.info for example

http://www.cerebraljs.com/ is good place to demonstrate best practices.
would be great to see it covered with tests

I found the FAQ a bit unhelpful when it comes to the difference between Cerebral and Redux. I think most important thing to highlight would be what Dan said in a Twitter conversation about this very topic: "I guess Redux helps when your data structure is complex but async interaction is not very complex. Cerebral is opposite."

Would be cool to have a diagram that explains the cerebral application flow:

State -> View -> Event -> Signal -> Action Chain -> State/Side effects

Not even sure how big the performance gain is, but in theory the rendering performance should be better when using immutable model in combination with shouldComponentUpdate and shallow checking.

I had a quick read through the V2 website in Vienna and think that we could maybe improve the content structure. Since I haven't had the time to discuss this on the meetup, I will just post it here as reminder.

• Get Started
  • Setup
  • Create State (define model)
  • Render State (define view and pull state)
  • Update State (add action chains to controller)
• Concepts
  • Single state
  • Data flow
  • Actions
  • Chains
  • Signals
• Advanced (is this really advanced?)
  • Routing
  • Choosing a model layer (immutable or not)
  • Serivces
  • Computed
  • Modules
  • Context Providers
• Best Practices
  • Folder structure
  • Naming conventions
  • Storing module state
  • …
• API
  • …

It opens the same site in a new tab.
Minimum: point it to https://github.com/cerebral/cerebral-website/tree/master/app/markdown
Maximum: point to current article markdown file on on github
PR are welcome

Should we use classes or functions in the react examples on website V2?

We should also mention that you should avoid internal state, when using classes.

Heroku is overkill for doc site. Let host it on github pages.

Refer to cerebral/cerebral#131

So maybe we could make a default rule:

"Never pass props into a Cerebral component. Use signals to trigger state changes and extract state from state store. Only pass props to stateless components"
So then you have statefulcomponents, which gets their props from Cerebral. And then you have stateles components which gets their props from parents
nice separation

// We just add the chains that puts the application in
// "messagesOpened" state
messageOpened: [...messagesOpened, ...messageOpened]

and

When the user goes to /messages/123 the messagesOpened and messageOpened chain will run.

looks a little strange to me.

At least it's hard to reason about ;)

To me it's always very helpful to look a frameworks recommended folder structure before reading about whats what. It's an easy way to get an overview of the bits an pieces. When you then later read about them you will already know where it sits in your project :)

Hope this makes sense - I had a bear or two before I took this note.

Both http://www.christianalfoni.com/cerebral/ and http://www.cerebraljs.com/ has glitchy scroll in menu.

http://www.christianalfoni.com/cerebral should redirect to the new site

On my first reading "Structuring State" in V2 I found it confusing to read about namespaces. Why exactly would you call it a namespace?