Bugs of WebViewer from the new version.

[Anthony Chow]

Hi,

I just found my project didn't work this morning, and it worked perfectly last night. I noticed that there is a new appearance of the IDE, but just when I recompiled my project, things became abnormal. It seems to me that the WebViewer doesn't behave as it was before. So I made a simple test on the WebViewer immediately: 

Below is the screenshot and the aia file of the test. Thanks in advance.

Regards,

Anthony

[Juan Antonio]

This test works for me.

Use HorizontalArrangement1
Height
Fill parent...

[Peter Mathijssen]

Did a quick test in the companion and it is working even set to automatic automatic. It displays full screen. 

[TimAI2]

Some of the problem may be the new AppInventor site, which I understand has not been fully sorted for mobile viewing

However I compiled your test app and threw it at Genymotion, it only displayed the AI2 logo with a gap at the top, AND i got a PlayStore message (might be the vagaries of the emulator?)

But something not right - the gap at the top ?

 

[Anthony Chow]

Hi Juan,

I have to assign a number to the height, as I have three Webviewers arranged vertically to separate the screen into three parts: one part for the video player, the other 2 for the main menu and the playlist. I have been working on this project for almost half a year, and suddenly, it doesn't work overnight after the the change of the IDE (I also found a new property called "BlocksToolkit" for the "Screen" components). Is it a bug ? I have no idea how to fix it, anyone can help ?

Regards,

Anthony

[TimAI2]

See the release notes for nb178:

https://appinventor.mit.edu/ai2/ReleaseNotes

Are you also using the latest companion 2.54/2.54u ?  (Although this is only directly needed for the new Speech Recognizer features)

[Chad Marks]

I too am having nothing but problems since last night.  I can't even drag blocks around without getting errors.  I must have submitted the auto-bug reports about 6 times.  Really disappointed since this is my first foray into app design, all I can really think of to do is start my app over from scratch and see if it works.

[TimAI2]

Done some more testing, works OK using new Companion on my HTC10 (Android 7.0), but very strange behaviour on my Genymotion Emulator (PIxel XL 3) (Android 9) and Pixel 3 Android 9  (both api 28)

Also works OK on my Genymotion Nexus 7 (android 5.1)

Can someone test on a real Android 8 and/or 9 device ?

[Juan Antonio]

Android 9

I changed background to green.

[TimAI2]

Yep, that is what I am seeing

[Zavodov com]

There was a problem after the update (yesterday everything was fine).

After compilation (exactly after compilation) WebViewer does not work correctly. (Android 9)

Igor

[ABG]

(added to App inventor Issues section of FAQ)

https://groups.google.com/forum/#!msg/mitappinventortest/wqJin0qO8M0/8xUafdROAgAJ

ABG

[Jan Cees Venema]

Same here. An image displayed using WebViewer in landscape mode shows only 5% at the bottom of the screen (regardless of using Automatic or Fill Parent). 

Also just happened today for the first time. This counts for my Xiaomi Redmi Note 5. My Samsung S6, however, doesn't have this problem!

[Anthony Chow]

Hi,

I am not using any companion, I just built the apk file and installed my app on my Sony H4493 (updated to Android 9) and Samsung SM-AQ105G (Android 9) directly.

Regards,

Anthony

[Chris Ward]

Hello everyone

It's a known bug introduced with App Inventor nb178. MIT are aware and hopefully it's something simple that can be fixed soon.

[Evan Patton]

We believe that this is only a problem affecting Android Pie 9. Has anyone experienced this issue on any other version of Android? This will help us understand where to start debugging.

Regards,

Evan

[Andres hernandez]

an old jelly beam here...it doesn't work

[BodyMindPower]

For me it works on a Galaxy S3 (Android 4.3 / Jelly Bean).

It does not work on a Xiaomi Redmi Not 5 (Android 9). The WebViewer collapses after a few ms. However, if something is triggered / executed on the screen, the WebViewer will be displayed correctly again. So if, for example, a fast clock (TimerInterval = 30) does anything (eg, set Lable.Text = empty string), everything is ok. See attached aia. Maybe that helps to fix the error.

Anke

[Evan Patton]

Hmm. That's very interesting Anke. I think that gives me an idea of something we should look at...

Thanks,

Evan

[Anthony Chow]

Hi Evan,

Since I have a WebViewer for displaying video, I made another test for video display with height 25%. Move the video vertically, you can see the the screen below 25% is shaking rapidly.

Regards,

Anthony

[BodyMindPower]

Hi Evan,

maybe it's interesting that the same bug also arises on Thunkable, since the API 28 release a few days ago.

https://community.thunkable.com/t/problem-after-thunkable-6-0-update/175789/14

So maybe it's not a AI2 specific problem.

Anke

[Taifun]

@BodyMindPower: Thunkable itself does not do much concerning their Classic version anymore, they just use the updates from MIT to update their version

Taifun

[BodyMindPower]

Hi Taifun, ok thanks for the hint.

[BodyMindPower]

It seems to be an Android bug / issue:

https://stackoverflow.com/questions/55371630/webview-display-problem-in-new-versions-of-android

Maybe that helps and someone get an inspiration for a solution.

The same problem arises since yesterday (API 28 release, Eagle) also on Kodular.

[Mika]

https://developer.android.com/about/versions/pie/android-9.0-changes-28

Maybe this is the reason:

Web-based data directories separated by process

In order to improve app stability and data integrity in Android 9, apps cannot share a single WebView data directory among multiple processes. Typically, such data directories store cookies, HTTP caches, and other persistent and temporary storage related to web browsing.

In most cases, your app should use classes from the android.webkit package, such as WebView andCookieManager, in only one process. For example, you should move all Activity objects that use a WebView into the same process. You can more strictly enforce the "one process only" rule by calling disableWebView() in your app's other processes. This call prevents WebView from being initialized in those other processes by mistake, even if it's being called from a dependent library.

If your app must use instances of WebView in more than one process, you must assign a unique data directory suffix for each process, using the WebView.setDataDirectorySuffix() method, before using a given instance of WebView in that process. This method places web data from each process in its own directory within your app's data directory.

Note: Even if you use setDataDirectorySuffix(), the system doesn't share cookies and other web data across your app's process boundaries. If multiple processes in your app need access to the same web data, you need to copy it between those processes yourself. For example, you can call getCookie() and setCookie() to manually transfer cookie data from one process to another.

[Chris Ward]

Hello Mika

That might cause issues itself, but it does not sound as though it could be the source of the WebViewer display issues which are our current concern.

I'm sure Evan will be doing his utmost to get it fixed, but it looks like the weight is on Google's shoulders.

[ABG]

See https://docs.google.com/document/d/1W0TkJuF29-3ILFSJZXt2eia3rlrAGBha-QE9EucFnpY/edit?usp=sharing

for a testing request for webviewer fixes on the ai2-test server.

ABG

[BodyMindPower]

Great, thanks ABG for the hint.

[Peter Mathijssen]

The problem with the webviewer should be solved in the next release that is planned for the coming weekend.

http://appinventor.mit.edu/ai2/upcomingrelease