Alex

This just probably need to be set to `type Reducer = (state: T) => T` if one needs `undefined` thing just use `Reducer`

```ts R.compose( isolateDOMFromSibling('foo'), isolateHTTP('foo'), Child )(sources); ``` Should probably be: ```ts R.compose( isolateDOMFromSibling('foo'), isolateHTTP('foo'), )(Child)(sources); ``` @jvanbruegge > but I am strongly in favor of standardization and a default isolation...

@abaco I agree that current API is just simple and convenient to use in the simplest cases (which are mostly the case for me), though only with typings saved =)....

> As for total isolation in onionify, it would provide one more way (there are others) to keep a component's state local. Onionify already provides this kind of "isolation", which...

> Should each isolateXXX also take a channel name in case the default one isn't used? That could be implemented (and even typed well) without problems I pointed this [in...

> DOM event mutation is a write effect, so according to Cycle's principles it should go through a sink. This is probably more correct approach than passing `{preventDefault: true}` to...

The problem with current approach is: ```ts DOM.select('input').events('click', {preventDefault: true}) // and then somewhere: DOM.select('input').events('click') // still will be default prevent, this is a kind of declarative inconsistency ``` >...

I'm not trying to solve the "problem" of how DOM events works in this case. I'm saying about code consistency. If you say in sink that *click events this elemennt...

@staltz what do you think about the issue I pointed at in [the comment](https://github.com/cyclejs/cyclejs/issues/727#issuecomment-340188302)? The same issue will be actual with any method "invoked" using the source. I think this...

@staltz > supposed to be sent to sinks, only there would the actual preventDefault occur. preventDefault happens in a driver, the question is how in a code this instruction is...