There might already be a bug on this, but I can't find it.
This is a regression from when the <preference> element was first introduced quite a while back (for FF1.5, as I recall). Not all apps that use preferences will include a separate dialog box with <prefpane>s in it as a user interface for modifying those prefs. Fennec is one example, and an app I wrote while with MDG is another.
Created attachment 335257 [details] [diff] [review]
Created attachment 335258 [details] [diff] [review]
forgot to update the patch before attaching it
Comment on attachment 335258 [details] [diff] [review]
patch is incomplete
Created attachment 335557 [details] [diff] [review]
Would be nice to get some tests for this (see e.g. http://mxr.mozilla.org/seamonkey/source/toolkit/content/tests/chrome/test_preferences.xul ).
good point, I'll see what I can come up with
Comment on attachment 335557 [details] [diff] [review]
Don't remove properties that are defined for elements and are listed in the documentation.
Why do we need a separate new element here? Can't we just put all of the functionality in the <preferences> element directly?
Created attachment 336099 [details] [diff] [review]
(In reply to comment #7)
> (From update of attachment 335557 [details] [diff] [review])
> Don't remove properties that are defined for elements and are listed in the
> Why do we need a separate new element here? Can't we just put all of the
> functionality in the <preferences> element directly?
Because it needs to listen for change events from the input controls (textboxes, checkboxes, etc) which means it needs to be an ancestor of those controls. Merging the prefmanager tag with the preferences tag would that everyone using these features would need to update their code, moving all of their input controls underneath their preferences tags.
(In reply to comment #9)
> Because it needs to listen for change events from the input controls
> (textboxes, checkboxes, etc) which means it needs to be an ancestor of those
Why can't it just use an event listener?
xbl destructors are never called, so it would never get a chance to remove the listener.
With a few minor errors fixed, the following already works just fine:
<textbox id="textbox" preference="browser.startup.homepage"/>
It seems like we can just use <prefpane> here rather than using a new element.
No, I can't. The prefpane binding inserts a bunch of UI stuff that are specific to one particular app's preferences window in addition to handling the preferences attribtue. By splitting the prefpane into two elements, (called "prefpane" and "prefmanager") you can get the latter without having to swallow the former.
I'm not following. I can understand the prefwindow binding, but I don't see anything of note in the prefpane binding. There are a couple of properties and methods that might not be used, but I don't see an issue there.
erm. on looking at the file again, I see that you're right. I need to reexamine the symptoms to see exactly how to fix the prefpane binding. In the mean time, however, could I get a review on the change to the preference element? We need that at least for Fennec A1. That's just the second patch.
Created attachment 342205 [details] [diff] [review]
fix a good review commend from Gavin.
Comment on attachment 342205 [details] [diff] [review]
>diff -r b222f6e972b3 toolkit/content/widgets/preferences.xml
>+ <property name="type">
>+ return "type" in doc && doc.type
missing semi-colon, and should probably return doc.type || ""; to match the default prefwindow behavior (missing attribute).
>+ <property name="instantApply">
>+ return ("instantApply" in doc && doc.instantApply) || rootBranch.getBoolPref("browser.preferences.instantApply")
I prefer explicit references to |this| (also missing a semi-colon). Shouldn't need the "in" check anymore either.
Should mark these readonly I guess (even though it doesn't seem to affect settability without a setter defined).
r=me with those.
Created attachment 342492 [details] [diff] [review]
patch to be checked in
Created attachment 342504 [details] [diff] [review]
forgot one nit in the last patch
Created attachment 342505 [details] [diff] [review]
/me sighs. perhaps this one will be acceptable?
Comment on attachment 342505 [details] [diff] [review]
this is perfectly acceptable, minus the changes to js/src/jsdtoa.cpp! :)
lol. I completely forgot to leave those out.
This should be fixed, no?
*** Bug 293439 has been marked as a duplicate of this bug. ***
This landed on Thu Oct 09 16:16:10 2008 -0700: