Can Closure figure out that type extensions are never used and remove related code? (eg. using extend-type with a protocol in a library but the user ends up not using that).

I would say no based on what I see but who knows

yes in theory. in practice unlikely though. CLJS isn't "typed" enough for this to work reliably

And I guess there is no way to help the process? Some type annotation somewhere?

depends. can't explain the process really. yes, typehints and eliminating dynamic uses of the protocol

basically you'll have to spend a bunch of time going through pseudo-named code looking for reasons why the code stays alive when you think it should be removed

closure is surprisingly good but many of the more advanced features really only work in fully typed code

but thats like tweaking the last 10% when you already get 90% of the benefit

So essentially, when extending protocols, dead code is rather to be expected on native types and custom ones that are used at least once somewhere

hard to say. really depends on how you use it but I'd say that is accurate yes.

Makes one slightly paranoid about overusing protocols

well you are using protocols with the intent of using them no? I mean you don't intend to write dead code from the start?

In the context of an app yes indeed, in the context of a lib it's best to think twice and try no to be too clever

This is a good way to do it: https://github.com/andrewmcveigh/cljs-time/blob/master/src/cljs_time/extend.cljs#L1-L2

Hello there, I am new Clojure(script) world. I needed a little with clojurescript, so I am fetching information from data api which contain html tags <br/> \r\n that are in hashmap as values. This is so that lines breaks after at desired sentence in HTML dom. I expect the line to break everytime this <br/> \r\n , however this is what I am getting

Whenever has this tag <br/>
This is what i get 
Lorem ipsum dolor sit amet, consectetur adipiscing elit. <br/> Quisque nisl eros, pulvinar facilisis justo mollis.

this is when I have \r\n
This is what i get 
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Quisque nisl eros, pulvinar facilisis justo mollis.

A lot of HTML rendering engines are reluctant to render “external” strings as HTML due to the danger of hacking, so they render “</br>” as a literal string </br> instead of as a break. I would expect the \r\n to work though. What are you using to generate the HTML for your DOM?

I have a shadow cljs react application with helix dom

Ok, React definitely won’t let you inject a “</br>” into your DOM. At the javascript level it will let you call dangerouslySetInnerHtml() to do something like that, but I wouldn’t take the chance myself. I haven’t used helix, let me look that up real quick and I’ll make a suggestion.

Ok, quick and dirty, here’s an example with my fake d/div function

(def test-string "Lorem ipsum</br>dolor sit amet\r\nQuisque nisl eros.")
=> #'user/test-string
(defn my-block
  [from-api]
  (let [lines (clojure.string/split from-api #"\r\n|\n|</br>")]
    (apply div (for [line lines] (div line)))))
=> #'user/my-block
(print (my-block test-string))
<div>
  <div>Lorem ipsum</div>
  <div>dolor sit amet</div>
  <div>Quisque nisl eros.</div>
</div>
=> nil

Basically what I did was use clojure.string/split plus a regular expression to split the incoming string into separate strings wherever there was a \r\n or a </br>, and then render each string separately.

Couple extra lines, but it makes React feel safer.