cordova-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Carlos Santana <csantan...@gmail.com>
Subject Re: Deprecation of Config and the embedded use case (4.0.x related)
Date Tue, 17 Mar 2015 14:10:36 GMT
@Michal
I would get together with Marcel and Martin, they are our Android goto SME

@Joe
Do we have all the test cases  created now are some missing?
How do you run this specific test cases on Android Studio? Google team any
help on gradle to run this test cases on CI I think is very important to
have a healhty CI running test cases on every pull request or commit

Hum I don't know how I feel about the Plugins not working, I think this is
the key on using Cordova.
If I'm just going to have a Web View with no Cordova then I would use In
App Browser.

The whole purpose of Including a "Cordova" Web View in to a full Native
Android App, It's the "Cordova" functionality which resides mostly in the
Plugins!
Maybe I misunderstood your statements about plugins not working.



On Mon, Mar 16, 2015 at 9:31 PM, Joe Bowser <bowserj@gmail.com> wrote:

> Well, this feature was tested using TDD, and when the tests were re-written
> I assumed that they would be run.  In this case, I'll blame Android Studio,
> since we're still battling with the learning curve on that one.  (I have no
> clue how to run the new tests from Gradle on the command line, only in
> Android Studio).
>
> The thing is that another refactor removing layouts broke the tests, which
> is how I know that they weren't run.  So, I landed a couple of commits to
> refactor the unit tests so that they test this use case with the new API
> and the tests now pass.  This works again, and we can update the
> documentation,
>
> There's still the matter of getting the plugins to work, but I'm fine with
> leaving that to be an exercise for the downstreams that support this, and
> not Cordova itself.
>
>
>
> On Mon, Mar 16, 2015 at 6:27 PM Michal Mocny <mmocny@chromium.org> wrote:
>
> > Carlos, thats great, then perhaps you could give 4.0 embedded webview a
> > shot to confirm that it is still adequately supported for your customers?
> >
> > I think this thread has been too much talk and not enough trying it out
> in
> > practice.  Everyone agrees the use case is important, what's left is to
> > confirm we got it right.
> >
> > -Michal
> >
> > On Mon, Mar 16, 2015 at 8:58 PM, Carlos Santana <csantana23@gmail.com>
> > wrote:
> >
> > > I just want to add that Joe is not alone on thinking that are
> developers
> > > with this use case.
> > > For us we have customers that start with Native Android alone, and then
> > > later want to add a Cordova Web View to a portion of their App.
> > > And they want an easy way to add a Cordova Web View.
> > > For 4.x, I would assume that the developer can choose to make this
> > embedded
> > > Cordova Web View CrossWalk based.
> > >
> > >
> > > On Wed, Mar 11, 2015 at 10:24 AM, Joe Bowser <bowserj@gmail.com>
> wrote:
> > >
> > > > That's why we have tests! I just changed the activity and saw that we
> > > have
> > > > one failure.  I'm not sure why this test in particular is failing,
> > since
> > > > there's too many assertions in one method, so I'll have to try and
> > debug
> > > it
> > > > today.
> > > >
> > > > The thing is that if we're deprecating something and replacing it
> with
> > > > something else, we should write tests for it.  Releasing a 4.0.x and
> > > > changing how we embed a WebView by changing class names but not
> fixing
> > up
> > > > the deprecation is bizzare.
> > > >
> > > > On Wed, Mar 11, 2015 at 7:15 AM Andrew Grieve <agrieve@chromium.org>
> > > > wrote:
> > > >
> > > > > I wanted to make sure that I didn't break the old way of doing
> > things.
> > > > >
> > > > > On Tue, Mar 10, 2015 at 2:24 PM, Joe Bowser <bowserj@gmail.com>
> > wrote:
> > > > >
> > > > > > The main issue is that this isn't documented anywhere, and this
> is
> > > > > > necessary for people to use a Third Party WebView.  Also, why
> > didn't
> > > > you
> > > > > > bother updating the test with the new API?
> > > > > >
> > > > > > On Mon, Mar 9, 2015 at 5:19 PM Andrew Grieve <
> agrieve@chromium.org
> > >
> > > > > wrote:
> > > > > >
> > > > > > > Here's an example:
> > > > > > >
> > > > > > > ConfigXmlParser parser = new ConfigXmlParser();
> > > > > > > parser.parse(activity);
> > > > > > > webView.init(cordova, parser.getPluginEntries(),
> > > > > > parser.getPreferences());
> > > > > > >
> > > > > > > Feel free to iterate if you think the API is too obtuse,
but I
> > > think
> > > > > it's
> > > > > > > good to allow a file-less mode, and to allow different
WebViews
> > to
> > > > have
> > > > > > > different settings.
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > > On Mon, Mar 9, 2015 at 8:08 PM, Joe Bowser <bowserj@gmail.com>
> > > > wrote:
> > > > > > >
> > > > > > > > Do you have an example of how this would work? This
seems to
> > be a
> > > > lot
> > > > > > > more
> > > > > > > > complex than it needs to be.
> > > > > > > >
> > > > > > > > On Mon, Mar 9, 2015 at 5:05 PM Andrew Grieve <
> > > agrieve@chromium.org
> > > > >
> > > > > > > wrote:
> > > > > > > >
> > > > > > > > > It's so that you can have multiple CordovaWebViews
that use
> > > > > different
> > > > > > > > > configs within one application. It's also so
that you don't
> > > have
> > > > to
> > > > > > > have
> > > > > > > > a
> > > > > > > > > config.xml if you prefer to build up your config
in code
> > > instead.
> > > > > > > > >
> > > > > > > > > I don't think loadConfig() is deprecated. It
has
> > > > > > > > > a @SuppressWarnings("deprecation"), which just
silences a
> > > warning
> > > > > > > about
> > > > > > > > it
> > > > > > > > > setting the config of the Config class (which
is done for
> > > > backwards
> > > > > > > > > compatibility).
> > > > > > > > >
> > > > > > > > >
> > > > > > > > > On Mon, Mar 9, 2015 at 3:54 PM, Joe Bowser <
> > bowserj@gmail.com>
> > > > > > wrote:
> > > > > > > > >
> > > > > > > > > > OK, this actually makes using the WebView
as a component
> a
> > > lot
> > > > > > > harder,
> > > > > > > > > > since you now have to have this loadConfig
method which
> you
> > > > also
> > > > > > > marked
> > > > > > > > > for
> > > > > > > > > > deprecation required to get all of the necessary
> attributes
> > > out
> > > > > of
> > > > > > > > this.
> > > > > > > > > > I'm pretty sure this is a major step backwards
in that
> > people
> > > > > > looking
> > > > > > > > to
> > > > > > > > > > use Cordova as a component now have to jump
through
> > > additional
> > > > > > hoops
> > > > > > > to
> > > > > > > > > get
> > > > > > > > > > this to work.  What is the benefit of deprecating
the
> > Config
> > > > > static
> > > > > > > > class
> > > > > > > > > > and replacing it with the ConfigXmlParser
again? I don't
> > > > remember
> > > > > > why
> > > > > > > > > this
> > > > > > > > > > was done.
> > > > > > > > > >
> > > > > > > > > > On Mon, Mar 9, 2015 at 9:04 AM Andrew Grieve
<
> > > > > agrieve@chromium.org
> > > > > > >
> > > > > > > > > wrote:
> > > > > > > > > >
> > > > > > > > > > > On Mon, Mar 9, 2015 at 11:56 AM, Joe
Bowser <
> > > > bowserj@gmail.com
> > > > > >
> > > > > > > > wrote:
> > > > > > > > > > >
> > > > > > > > > > > > On Mon, Mar 9, 2015 at 7:39 AM
Andrew Grieve <
> > > > > > > agrieve@chromium.org
> > > > > > > > >
> > > > > > > > > > > wrote:
> > > > > > > > > > > >
> > > > > > > > > > > > > You can now instantiate a
CordovaWebView without a
> > > > > > config.xml,
> > > > > > > > and
> > > > > > > > > > > > without
> > > > > > > > > > > > > using Config. This happened
when I added an
> "init()"
> > > > method
> > > > > > to
> > > > > > > > > > > > > CordovaWebView. You can pass
in a
> CordovaPreferences
> > > > > object,
> > > > > > > and
> > > > > > > > a
> > > > > > > > > > list
> > > > > > > > > > > > of
> > > > > > > > > > > > > PluginEntry. Maybe we just
need a better comment on
> > > > Config
> > > > > > > saying
> > > > > > > > > to
> > > > > > > > > > > use
> > > > > > > > > > > > > these instead?
> > > > > > > > > > > > >
> > > > > > > > > > > > >
> > > > > > > > > > > > Where does one get this PluginEntry
list when they're
> > > > > > embedding a
> > > > > > > > > > > WebView?
> > > > > > > > > > > > This needs to be documented or
at least put in the
> test
> > > > that
> > > > > > > tests
> > > > > > > > > this
> > > > > > > > > > > use
> > > > > > > > > > > > case.
> > > > > > > > > > > >
> > > > > > > > > > > > >
> > > > > > > > > > > >
> > > > > > > > > > > > > > That has nothing to
do with InAppBrowser, this is
> > to
> > > do
> > > > > > with
> > > > > > > > > > > embedding
> > > > > > > > > > > > a
> > > > > > > > > > > > > > WebView inside an Android
application. I don't
> > think
> > > > you
> > > > > > > > > understand
> > > > > > > > > > > > what
> > > > > > > > > > > > > I
> > > > > > > > > > > > > > mean when I say the
embedded use case.
> > > > > > > > > > > > > >
> > > > > > > > > > > > > Maybe try explaining a bit
more?
> > > > > > > > > > > >
> > > > > > > > > > > >
> > > > > > > > > > > > Even though you edited the test
that explicitly
> covers
> > > this
> > > > > > use,
> > > > > > > > > case,
> > > > > > > > > > > and
> > > > > > > > > > > > even though we've talked about
using CordovaWebView
> as
> > an
> > > > > > Android
> > > > > > > > > View
> > > > > > > > > > > for
> > > > > > > > > > > > over a year, you need it explained
more?
> > > > > > > > > > > >
> > > > > > > > > > > > So, not everyone wants to use
all of Cordova, for
> many
> > > > > reasons.
> > > > > > > > > > Instead,
> > > > > > > > > > > > they really just want to take
advantage of the
> WebView
> > > > > > component
> > > > > > > in
> > > > > > > > > > their
> > > > > > > > > > > > native apps so that they can create
hybrid apps that
> > are
> > > > > mostly
> > > > > > > > > native
> > > > > > > > > > > with
> > > > > > > > > > > > only some parts that use Cordova.
 This is where you
> > > would
> > > > > > > declare
> > > > > > > > > your
> > > > > > > > > > > > view in your layout XML like this:
> > > > > > > > > > > >
> > > > > > > > > > > > <org.apache.cordova.engine.SystemWebView
> > > > > > > > > > > >             android:id="@+id/WebViewComponent"
> > > > > > > > > > > >             android:layout_width="match_parent"
> > > > > > > > > > > >             android:layout_height="match_parent">
> > > > > > > > > > > > </org.apache.cordova.engine.SystemWebView>
> > > > > > > > > > > >
> > > > > > > > > > > > And then, in the activity start
up your view like
> this:
> > > > > > > > > > > >
> > > > > > > > > > > >   private CordovaWebView webInterface;
> > > > > > > > > > > >   private CordovaInterfaceImpl
systemInterface = new
> > > > > > > > > > > > CordovaInterfaceImpl(this);
> > > > > > > > > > > >
> > > > > > > > > > > > //Set up the webview
> > > > > > > > > > > >         SystemWebView webView
= (SystemWebView)
> > > > > > > > > > > > findViewById(R.id.WebViewComponent);
> > > > > > > > > > > >         webInterface = new CordovaWebViewImpl(this,
> new
> > > > > > > > > > > > SystemWebViewEngine(webView));
> > > > > > > > > > > >
> > > > > > > > > > > >         Config.init();
> > > > > > > > > > > >         webInterface.init(systemInterface,
> > > > > > > > > Config.getPluginEntries(),
> > > > > > > > > > > > Config.getPreferences());
> > > > > > > > > > > >         webView.loadUrl(Config.getStartUrl());
> > > > > > > > > > > >
> > > > > > > > > > > > Right now, we're getting the configuration
from the
> > > Config
> > > > > > class,
> > > > > > > > > > because
> > > > > > > > > > > > we at least have access to this.
 If we don't have
> > this,
> > > > how
> > > > > do
> > > > > > > > > people
> > > > > > > > > > > get
> > > > > > > > > > > > access to the list of plugin entries
specified in
> > > > Config.xml?
> > > > > > > I'm
> > > > > > > > > > pretty
> > > > > > > > > > > > sure we still want to support
this feature.
> > > > > > > > > > > >
> > > > > > > > > > > > Does that make sense?
> > > > > > > > > > > >
> > > > > > > > > > >
> > > > > > > > > > > I gotcha. So, I think the answer is
to use
> > > ConfigXmlParser()
> > > > to
> > > > > > > > extract
> > > > > > > > > > the
> > > > > > > > > > > information required by init. You shouldn't
need the
> call
> > > > > > > > Config.init()
> > > > > > > > > > at
> > > > > > > > > > > all.
> > > > > > > > > > >
> > > > > > > > > >
> > > > > > > > >
> > > > > > > >
> > > > > > >
> > > > > >
> > > > >
> > > >
> > >
> > >
> > >
> > > --
> > > Carlos Santana
> > > <csantana23@gmail.com>
> > >
> >
>



-- 
Carlos Santana
<csantana23@gmail.com>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message