Hm. I guess the id test is historical stuff. It is legal XML and the parser ignores it.
Regarding the NS prefix: cdv vs gap I am quite sure I did not invent cdv myself. Regarding the w3c widget standard: I am OK with ditching it. Should there be a poll or how do you handle such things? All app templates should remove namespaces then, right? Isn't there another thread around namespaces breaking the wp platform? Anyway, I think we should follow the phonegap way with e.g gap:density for now. And ditch namespaces later. Axel Am 12.02.2014 22:38 schrieb "Andrew Grieve" <agri...@chromium.org>: > On Tue, Feb 11, 2014 at 3:43 PM, Brian LeRoux <b...@brian.io> wrote: > > > Given we are the only 'widget spec' impl in use today I'm ok with > diverging > > and not adding namespace confusion. Def want config to be explicit and > not > > have magical implicit mappings. > > > > > > On Tue, Feb 11, 2014 at 12:24 PM, Axel Nennker <ignisvul...@gmail.com > > >wrote: > > > > > - My implementation does not use "id". Don't know what this or might > > mean. > > > Found it from your test in spec/test-config.xml > > > > > - I do not want to discuss the sense of xml namespaces in this issue > if > > we > > > can avoid it. The current template config.xml defines two namespaces > and > > > for this issue's implementation I do not want to change that. So I > would > > > not drop the widget namespace and would not support > > > "platform"-without-prefix. > > > > Having gap:platform there makes the property seem like an second-class > maybe-not-supposed-to-be-there kind of attribute to me. > I'd be happy to change the default template to not reference the widget > spec and to make cordova's the default namespace if that will make your > inner XML validator rest at-ease, but I really feel strongly against having > XML namespaces creep in. I don't think that most devs know what they do, > and our tools wouldn't support you changing the gap: namespace prefix. > > > > > > - I would follow the phonegap example > > > > > > > > > http://docs.build.phonegap.com/en_US/3.1.0/configuring_icons_and_splash.md.html#Icons%20and%20Splash%20Screens > > > that defines e.g. " > > > > > > <icon src="icons/android/ldpi.png" gap:platform="android" > > > gap:density="ldpi" /> > > > > > > " and I would like to don't parse the icon file to infer parameters. > > > Developers want need icons for their app on all platforms they support. > > So > > > they will create all of them in all polished sizes and densities. > > > My Android implementation puts icons without cdv:density into > > > "drawable/icon.png" regardless of width/height. > > > What behaviour would you suggest when both lines are present in one > > > config.xml > > > <icon src="icon48.png" width="48" cdv:platform="android" /> // would > end > > up > > > in drawable-mdpi by your suggestion > > > <icon src="icon-mdpi.png" cdv:density="mdpi" cdv:platform="android" /> > // > > > would end up in drawable-mdpi too > > > > My thinking here was that density says the same thing as size, so I would > just not support density (or make size="mdpi" an alias for size="48"). > > > > > > > > > I think that developers know what is the "platform-way" for each > > platform. > > > On Android the usual way is to specify densities. > > > > > > - I would not use "size" because that is not w3c widget style. > > > > > > -- Axel > > > > > > > > > > > > 2014-02-11 20:22 GMT+01:00 Andrew Grieve <agri...@chromium.org>: > > > > > > > Would love to move this along. Would like to get buy-in from others > > > > first though. > > > > > > > > The proposal in this PR is to add tags like: > > > > > > > > <icon id="icon" src="icon.png" /> > > > > <icon id="logo" src="logo.png" width="255" height="255" /> > > > > <icon src="logo-android.png" width="255" height="255" > > > > cdv:platform="android" cdv:density="mdpi" /> > > > > > > > > My feedback: > > > > - What is "id" for? > > > > - Supporting "cdv:platform" is fine, but we should also support just > > > > "platform=". I'd be fine to drop xmlns="http://www.w3.org/ns/widgets > " > > > > from the file. > > > > - I don't think there are any platforms that support non-square > icons. > > > > I think size="###" would be better than width= && height= > > > > - What happens if you don't specify a size? Do we sniff it from the > > > > png header? This might be nice as a follow-up, but I'd lean towards > > > > making it required for the first cut. > > > > - cdv:density seems redundant with respect to size. Icons on android > > > > are 46px at mdpi, so the size can be used to derive the density. > > > > > > > > > > > > On Mon, Feb 10, 2014 at 10:55 AM, Andrew Grieve < > agri...@chromium.org> > > > > wrote: > > > > > He Axel, thanks for spearheading this. Will have a look shortly. > > > > > > > > > > > > > > > On Mon, Feb 10, 2014 at 9:07 AM, Axel Nennker < > ignisvul...@gmail.com > > > > > > > wrote: > > > > >> > > > > >> Andrew, > > > > >> > > > > >> any comments to the current implementation? > > > > >> https://github.com/apache/cordova-cli/pull/126 > > > > >> > > > > >> Joe commented that the new class in config_parser.js named "icon" > > > should > > > > >> be named "Icon" but I left it as is because the other classes are > > > > lowercase > > > > >> too. > > > > >> There was another comment that namespaces in config.xml attributes > > are > > > > >> SchnickSchnack/chatter. > > > > >> I think that we should use the cordava namespace if config.xml > > > deviates > > > > >> from the W3C widget definition. > > > > >> These two are the only comments I got. > > > > >> > > > > >> I tested this on Android and FirefoxOS. > > > > >> > > > > >> Any chance to accept the request (at least the Android part)? > > > > >> > > > > >> -Axel > > > > >> > > > > > > > > > > > > > > >
