Re: [WICG/webcomponents] The MountObserver api, an alternative lazy custom element definition proposal (#896)

Good question, @doeixd,

Without a doubt, this proposal would need to be looked at closely by an expert who fully understands the inner workings of the browser, in order to help find the right balance between what would work performance wise, vs, perhaps, more flexibility.

Without that inside knowledge, I am making my best guess as to what would be effective to produce the most efficient result, perhaps erring on the side of making sure we don't sacrifice performance.  If the expert comes back and says that "we can do this in a way that would empower the developer more, without sacrificing performance", then I'm all for taking advantage of that.

My mental model I'm applying is comparing the situation to that of middle tier code, that needs to query  data from a remote database.  A good rule of thumb that seems to have prevailed in that arena, is to do as much of the filtering as possible declaratively, with a single call, so that less information needs to cross through the wire back and forth.  I *think* some of that may apply here, because I think most of the discovery of these conditions would be taking place within the c++ code, and I have seen people I consider to be experts bring up the issue of latency between c++ and JavaScript interactions (for example, in the context of [JSON parsing](https://github.com/whatwg/fetch/issues/104) ).  Also, wonderful as events are, I think the power they provide does come at a bit of a performance sacrifice, compared to inline code and even chained functions, hence my thinking that we only want a single event that captures the most important question -- are all conditions satisfied or not?

Let me know if my response makes sense, or if I'm misunderstanding your question.

It gets murky in my mind if doing the check for whereInstanceOf could happen in the low level code, or the JavaScript layer.  My best guess is low level, but it's no better than a coin toss, really.  I'm going to keep that question open in my mind, still kind of mulling that one over.

-- 
Reply to this email directly or view it on GitHub:
https://github.com/WICG/webcomponents/issues/896#issuecomment-2081599082
You are receiving this because you are subscribed to this thread.

Message ID: <WICG/webcomponents/issues/896/2081599082@github.com>