[Webkit-unassigned] [Bug 24529] WebInspector: HTML5 Offline Web Applications Support (ApplicationCache)

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=24529





--- Comment #73 from Joseph Pecoraro <joepeck at webkit.org>  2010-07-07 21:05:26 PST ---
> > Reverting the patches leaves me with the same problem from before.
> 
> What problem is that?

  (1) Patches that grow larger and larger in bugzilla, getting outdated, requiring rebaselines.
  (2) Multiple people working on the same code blind. Kavita was using my baseline patch,
      so any changes she made I couldn't be aware of until her patch is done.


> I was asking you to consider reverting not so much because of the chrome issues, that can be
> patched one way or the other. I'm more concerned about unasked questions regarding other
> aspects of the patch.

In the past if a feature is landing that requires a lot of parts (Timeline Panel, Audits Panel,
even the new HTML5 Parser) we have chosen to land but disable the feature, by default.
I agree that this is nowhere near complete. I would rather we turn the feature off than take
it out of the tree. Does that sound good?

It is easier to land incremental patches than to make large patches that often get out of
date when they sit on Bugzilla. Breaking pieces like this down into separate patches also
makes it easier to review.

Your observations and concerns are very good. I'd suggest filing bugs with "Application
Cache" in the title.


> - When a page calls swapCache and gets associated with a new cache. [...]

Currently it does not. This would be a good incremental patch. 


> - Can you separate populating the list view on the left-hand-side from
> the detailed grid view on the right-hand-side and defer fetching the
> resource list until needed? That could help make the interface more responsive.

This is how it works right now. When you select the ApplicationCache from
the left-hand-side it then fetches the data lazily.


> - When multiple top-level pages are associated with the same appcache group, it looks
> like only one (actually could be none) will be notified of update status changes, what
> about the others? I'm not talking about the nested frames, multiple pages using the
> same cache. What is your plan for that?

> - How does InspectorApplicationCacheAgent::didReceiveManifestResponse figure out
> which appcache/appcacheHost its being called by? Same question for all interfaces
> related to "pushing" info from the backend to the frontend. What is the plan for that?
> 
> - Why is 'domain' used as the label in the left-hand-side? The cookie model doesn't
> really apply, there can be multiple application caches per domain and in use in the
> same page at the same time. This real estate could probably be put to better use than
> having the 'domain' on display.

These all deal with multiple application caches, or multiple domains. This initial patch
doesn't deal with this, and admittedly handles it wrong right now. I didn't intend to have
this working in the first patch, and I took what Kavita had in her patch regarding domains
and left it unchanged. I see its behavior is wrong. I filed a bug:
https://bugs.webkit.org/show_bug.cgi?id=41636

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.