Web Chat release testing

Try it out

Click to try out hosted release testing.

What is?

Formalized directions for testing Web Chat for pre- and post-release.

Do what?

1. Navigate to VSTS Web Chat daily builds
   1. Queue a new daily build (upper right corner)
2. Extract into the drop folder:
   • All files from cdn_folder except those ending in *.LICENSE and *.json
   • All tgz files
3. Move to 01.create-react-app directory
4. Delete Web Chat and directlinespeech dependencies from package.json
5. npm i botframework-directlinespeech-sdk....tgz
6. npm i botframework-webchat...core....tgz
7. npm i botframework-webchat...api....tgz
8. npm i botframework-webchat...component....tgz
9. npm i botframework-webchat...tgz (no suffix)

Browsers, which?

https://1drv.ms/x/s!Aqv30r9qhHTIdWWqWl6egBqgaRk?e=F7jp1n

• Chrome
• Edge Anaheim
• Firefox
• Internet Explorer (renderWebChat es5.js)
• Safari (Mac)
• Safari (iOS)
• Chrome (Android)

Commands, test?

• Use the 'Run all tests' button it runs this:

  const commands = [
  'card bingsports',
  'card breakfast',
  'card broken:lang',
  'card broken',
  'card flight',
  'card flighttracking',
  'card inputs',
  'card ol',
  'card markdown',
  'card reminder',
  'card restaurant',
  'card review',
  'card richmessage',
  'card simple',
  'card sportsclub',
  'card ul',
  'card weather',
  'animationcard',
  'audio',
  'audiocard',
  'card-actions',
  'carousel',
  'channel-data',
  'document-data-uri',
  'document-plain',
  'document-word',
  'dump-activity',
  'echo Hello world',
  'emptycard',
  'file',
  'herocard',
  'herocard long title',
  'image',
  'image-svg',
  'input-hint accepting',
  'input-hint expecting',
  'input-hint ignoring',
  'invalidCard',
  'layout single',
  'layout single carousel',
  'layout double',
  'layout carousel',
  'layout',
  'localization',
  'markdown',
  'content-multimedia',
  'oauth',
  'oauth signout',
  'proactive',
  'receiptcard',
  'sample:backchannel',
  'sample:password-input',
  'sample:redux-middleware',
  'signin',
  'slow',
  'speech',
  'tell me a story',
  'suggested-actions',
  'text',
  'thumbnailcard',
  'thumbnailcard long title',
  'timestamp grouping',
  'typing',
  'typing 1',
  'unknown activity',
  'unknown attachment',
  'upload',
  'user',
  'user id',
  'user name',
  'video',
  'video vimeo',
  'video youtube',
  'videocard',
  'xml'
  ];
  function runCommands(commands, dispatch) {
  if (!commands.length){ return }
  const text = commands.shift();
  dispatch({
   type: 'WEB_CHAT/SEND_MESSAGE',
   payload: {
     text
   }
  });
  
  new Promise(function (resolve) {
   setTimeout(resolve, 3000);
  }).then(function () {
   runCommands(commands, dispatch);
  });
  }
  

• On the release checklist, go through manual tests... manually.

Why.

We assume the hosting app has 4 ways to load Web Chat:

• Without React
  • Pure JavaScript: use renderWebChat()
  • Direct Line Speech
• With React
  • Babel
    • Use renderWebChat()
    • Use JSX <ReactWebChat>
  • create-react-app: use JSX <ReactWebChat>

In a React app, assume the hosting page will call React Hooks. Because Web Chat may not violate any rules of hooks, React will always be used/tested with Babel.