Call with Jack Slocum Minutes Feb 26 2008

From RuntimeWiki

Jump to: navigation, search

Topic

Call with Jack Slocum on Feature Requests Summary Page

Attendees

  • Coach Wei
  • Jack Slocum
  • Kin Blas
  • Jon Ferraiolo

Minutes

  • Coach: (give an intro of the task force, and status). Thank Jack for participating today;
  • Jack: the list seems to be higher level than the issues i have, which are more specific to user interface;
  • Coach: ok. so let's go through the list first, and then go through Jack's issues. What do you think of the "HTTP connection limit issue"?
  • Jack: important, if we can get browser vendors to do something. It is a huge limit and seems to be the left from the dialup days;
  • Jack: for cross frame communications, definitely important and should have been done a long time ago. Be able to trigger events cross frames would be very cool. Though we should be careful about security aspects;
  • Jack: event transparency API: powerful, but also makes it more complicated. In terms of priority, it would be really great to have something like this, but before this, i'd prefer capturing events (reserse to event bubbling) on Internet Explorer.
  • Jack: capturing events can be done on other browsers.
  • Jon: regarding to drag and drop, you can do it using Javascript. Would it be helpful to provide browser support for drag and drop (native support)?
  • Jack: like the way Internet Explorer for it? Adobe AIR has very nice support for drag and drop (to and from native OS), though there are some security implications;
  • Coach: would you suggest browsers provide native support for drag and drop?
  • Jack: the way it is ok (via JavaScript). The problem is this event transparency issue listed here that makes it difficult and requires a lot of workaround;
  • Jon: if we lobby other browsers to implement a few things like "getClientRect" (only IE implements this today), it would hugely improve things;
  • Jack: Yes, "getClientRect" would be a significant performance booster for any Javascript toolkit if it is supported on all browsers beyond IE.
  • Coach: "mutation event"?
  • Jack: a good idea. I see it very useful; Not sure how critical it is;
  • Coach: next is about performance;
  • Jack: "String" manipulation on IE is top on my list. Extremely slow. A lot of people use array "Join" to handle this, but array "join" is slower on other browsers.
  • Jack: "Computed box and style" definitely needs an upgrade here.
  • Jack: DOM performance is improved with IE7. I haven't ran into "eval" and Object creation issues myself;
  • Jack: On IE, createDocumentFragment function and innerHTML functions do not work with "table". Getting better table support on IE would be huge;
  • Jack: the ability to specify "overflow: auto" on the TBODY element on IE is big (You can do this on FireFox). Supporting this for THead and TFooter would be even better;
  • Jack: "subclassing" array is not a big issue for me;
  • Jack: "Client Side Persistent Storage and Caching" would be huge. IE has "userData" behavior, but other browsers don't have it;
  • Coach: "Xpath"?
  • Jack: Xpath would be good. But WebKit's "getElementBySelector" would be even better. Better support for CSS selectors would my perference.
  • Jack: "2D Drawing/Vector Graphics" -another item that would be fantastic if all browsers support it, though not sure what browser vendors will do;
  • Jack: "performance", improvement on Javascript performance, would be my top item on the list. Actually, "DOM performance" would be important too. At EXT, we use a lot of document fragments and innerHTML, which are essentially Javascript operations; Within a browser, if you loop through 50 elements and change their width, at the end of it, the browser will try to re-flow the elements. The "re-flow" is a huge bottleneck. The performance of "reflowing" or "recalcualting of layout" in response to changes made through Javascript is a big bottleneck;
  • Jack: "scoped ID" would help a lot. At EXT, we use autogenerated IDs to get around this;
  • Jack: my list are mainly rendering issues:
    • some kind of overFlowChange event would be huge. When an element overflow=auto, when a scrollbar appears, it is very hard to figure out when the scroll bar appears and disappears;
    • With absolute positioned "DIV" overlap on top of another element, the other element may show through (such as Select box, scrollbar). IE 7 fixed this. Firefox 2 made this worse. A lot of peple use "iFrame" to cover this. IFrame has a signficant performance drag.
    • IE6 and IE 7: with strict doc type, anything you set "overflow: auto" for an element, when you scroll the page, the element will not scroll.
  • Kin: IE establishes a new Z-Index context when you change the position of an element (anytime you position an element other than static), which has been annoying;
  • Coach: I dont' think i captured this well. can you send me a short description on this?
  • Kin: ok.
  • Coach: thanks all. I'll send out the minutes.
Retrieved from "http://www.openajax.org/runtime/wiki/Call_with_Jack_Slocum_Minutes_Feb_26_2008"
Personal tools