- From: Getify <getify@gmail.com>
- Date: Mon, 18 Oct 2010 01:08:08 -0500
- To: "public html" <public-html@w3.org>
?> Legacy behavior is often illogical. If we agree that something is illogical, doesn't that make it a good place to start in terms of fixing things to be logical in a future revision of the spec? Moreover, you haven't supported by you suggested we are "hacking around with async" -- indeed, I think we're doing the opposite of "hacking around" but instead are trying to directly address a short-falling of the current spec with a solution that narrowly fixes a specific use-case. > As well as being declarative, the waitFor proposal also supports more > use cases than the other proposal because it doesn't force a linear > dependency graph. The reason I'm in favor of the `async` approach as opposed to "waitFor" or several other possible solutions discussed so far is because I think it satisfies the use-case while fitting the principle of least effort -- it's a narrow fix for a specific use-case. I'm not suggesting that the non-linear dependency graph you suggest is invalid, but it's not what anyone so far in the thread has been asking for. I think there'd be a burden-of-proof on someone to show why the more complicated solution is necessary and the current more simple solution is insufficient. *Could* LABjs use behavior like to solve what it needs to do? Perhaps. But it's definitely more complicated than what I'm pushing for. LABjs currently addresses the need to have non-linear dependency by telling its users to just use separate independent $LAB chains. For the proposal I've pushed for, it means basically that the decision to use `async=false` type behavior for order enforcement would be available per-chain, which would let LABjs users do pretty much any kind of loading behavior they're likely to need. There may be some really niche corner-cases not served, but it's not something that I've seen needed in any practice thus far. --Kyle
Received on Monday, 18 October 2010 06:08:45 UTC