Improving PWA Page Load - progressive-web-apps

I have a PWA, which is essentially a book reader. As a result, it needs lots of data (viz. the book text) to operate. When analyzed by Lighthouse, it scores poorly on the Page Load Check.
My question is: What methods could I employ to improve the page load, while still ensuring offline functionality?
I could have a minimal start page (e.g., just display a 'Please wait, downloading text' message) and then dynamically download (via injected script tag or AJAX) the JSON data file. However, I'm not sure how I would subsequently ensure that the data is fetched from the cache.
Just wondering how others have handled this issue...

Since this question has gone tumbleweed, I decided to post the results of my attempts.
Based on Jake's article, I used the following script and Chrome DevTools to study service worker events:
'use strict';
let container = null;
let updateFound = false;
let newInstall = false;
window.onload = () => {
container = document.querySelector('.container');
let loading = document.createElement('div');
loading.classList.add('loading');
loading.innerHTML = 'Downloading application.<br>Please wait...';
container.appendChild(loading);
console.log(`window.onload: ${Date.now()}`);
swEvents();
};
let swEvents = () => {
if (navigator.serviceWorker) {
navigator.serviceWorker.ready.then(() => {
console.log(`sw.ready: ${Date.now()}`);
if (!updateFound) {
loadApp();
return;
}
newInstall = true;
console.log(`new install: ${Date.now()}`);
}).catch((error) => {
console.log(`sw.ready error: ${error.message}`);
});
}
navigator.serviceWorker.register('/sw.js').then((reg) => {
reg.onupdatefound = () => {
updateFound = true;
console.log(`reg.updatefound: ${Date.now()}`);
const newWorker = reg.installing;
newWorker.onstatechange = (event) => {
if (event.target.state === 'activated') {
console.log(`nw.activated: ${Date.now()}`);
if (newInstall) {
loadApp();
return;
}
refresh();
}
};
};
}).catch((error) => {
console.log(`reg.error: ${error.message}`);
});
};
let refresh = () => {
console.log(`refresh(): ${Date.now()}`);
// window.location.reload(true);
};
let loadApp = () => {
console.log(`loadApp(): ${Date.now()}`);
let child;
while (child = container.firstChild) {
container.removeChild(child);
}
let message = document.createComment('p');
message.textContent = 'Application loading';
container.appendChild(message);
let tag = document.createElement('script');
tag.src = './app.js';
document.body.appendChild(tag);
};
Along the way, I learned that once a service worker is registered, it immediately begins downloading all cached resources. I had assumed that resources were cached only after the page loaded them. I also found some definitive event patterns to indicate which lifecycle phase was occurring.
For a new install, the following events are logged in the above script:
window.onload -> reg.updatefound -> sw.ready -> nw.activated
For this case, when sw.ready fires, all resources have been cached. At this point, I can switch the app from the 'please wait' phase and dynamically load the cached resources and start the app.
For a simple page refresh, the following events are logged:
window.onload -> sw.ready
This will be the event sequence if the app has already been downloaded and no updates are available. At this point, I can again switch phase and start the app.
For a page refresh when the service worker script has been updated, the following events are logged:
window.onload -> sw.ready -> reg.updatefound -> nw.activated
In this case, when nw.activated fires, all cached resources have been updated. Another page refresh is required to actually load the changes. At this point, the user could be prompted to update. Or the app would update on its own the next time it was started.
By tracking these event patterns, it is easy to tell which lifecycle phase the service worker is in and take the appropriate action.

Related

Repeatedly getting error in Word office add-in "RichApi.Error: Wait until the previous call completes"

After applying Content control on long document (around 10 pages) using office Js, when I am trying to select the content control form a document using below code snippet, I am getting an error
OfficeService.js:338 RichApi.Error: Wait until the previous call completes.
at new n (word-win32-16.01.js:25:246227)
at o.processRequestExecutorResponseMessage (word-win32-16.01.js:25:310053)
at word-win32-16.01.js:25:308456
Sometimes I am able to fetch the selected content control but it's very slow and many times I get the above error. Not sure why this issue is occurring, I have checked in Office Js documentation but couldn't find the resolution.
export const findSelectedContentControl = async () => {
return await Word.run(async (context) => {
try {
const selectedContentControl = context.document.getSelection().contentControls;
selectedContentControl.load("items");
console.log("selectedContentControl..",selectedContentControl)
return await context.sync().then(() => {
let tagArray;
if(selectedContentControl.items.length === 0)
return;
tagArray = selectedContentControl.items[0].tag.split("|");
return tagArray;
});
} catch (error) {
console.log(error);
}
});
};

PWA - Cache won't update for offline use

I have a PWA which works fine both online and offline (but only with the initial files). However, the offline cache (let’s say a javascript file) is not being refreshed so whenever I am offline the old javascript file is used, but when online the new version is used.
On an iPad I can use Safari to go to the website and add the PWA to the home page.
If I then go offline, it works fine – all pages work etc.
But if I make a change to say a javascript file (something like adding an alert) and also change the version in my service worker, when I am online the change is reflected but when offline it remains at the older version
To clarify let’s say from the start, on going into a page it alerts “A1”
I then change the javascript to alert “A2” and change the version in the service worker.
If I run the app when online, sure enough the app says New Update Available and All Good (some alerts from the main.js file)
Then when I go into the actual page o the alert says “A2” – so all good.
Then go offline.
The alert still says “A1”
It seems that when online it uses the server latest files but when it tries to use cache the files are old and at the moment seem to be the original files.
I have read many sites on this with no success – some suggest it will sort itself in 24 hours. Some suggest setting the maxage of the service worker to 0 (but how do you do this?). Some say the files need renaming each time they change which seems very clunky.
The service worker is definitely working
main.js
$(document).ready(function () {
'use strict';
if ('serviceWorker' in navigator) {
navigator.serviceWorker
.register("/sw.js")
.then(res => {
console.log("service worker registered");
res.onupdatefound = () => {
const installingWorker = res.installing;
installingWorker.onstatechange = () => {
switch (installingWorker.state) {
case 'installed':
if (navigator.serviceWorker.controller){
alert("new update available");
forceReload();
}
else {
alert("all good");
}
break;
}
}
}
})
.catch(err => console.log("service worker not registered", err))
}
});
const forceReload = () =>{
console.log("ForceReload");
navigator.serviceWorker
.getRegistrations()
.then((registrations) =>{
console.log(registrations);
//alert("reg");
Promise.all(registrations.map((r) => r.unregister()))
caches.keys().then(function(names) {
for (let name of names)
caches.delete(name);
});
},
)
.then(() => {setTimeout(() => {
location.reload();
}, 500);
})
}
sw.js
let version =5; // update this to send update.
var cacheName = 'cacheV5'
var filesToCache = [
'/',
'/manifest.json',
'/index.html',
'/sales10.html',
'/getdata.html',
....
....
'/js/siteJS/sales10.js',
'/js/siteJS/getdata.js',
'/js/jquery/3.4.1/jquery.min.js',
'/js/bootstrap/bootstrap.min.js',
'/js/bootstrap/popper.min.js'
];
/* Start the service worker and cache all of the app's content */
self.addEventListener('install', function(e) {
self.skipWaiting();
e.waitUntil(
caches.open(cacheName).then(function(cache) {
return cache.addAll(filesToCache);
})
);
});
/* Serve cached content when offline */
self.addEventListener('fetch', function(e) {
e.respondWith(
caches.match(stripQueryStringAndHashFromPath(e.request.url.replace(/^.*\/\/[^\/]+/, ''))).then(function(response) {
return response || fetch(e.request);
})
);
});
function stripQueryStringAndHashFromPath(url) { //added this so when url paramerters passed grabbing the cashed js works
return url.split("?")[0].split("#")[0];
}
self.addEventListener('activate', function(event) {
event.waitUntil(
caches.keys().then(function(cacheNames) {
return Promise.all(
cacheNames.filter(function(cacheName) {
return true;
}).map(function(cacheName) {
return caches.delete(cacheName);
})
);
})
);
});

Why does the service worker download a page instead of navigating to it?

I've got a strange scenario with a service worker where instead of navigating to a page it will download a page instead of navigating to it.
On the link that it's supposed to navigate to there's nothing special, it's an anchor tag with an href attribute pointing to a sign out URL.
The relevant part of the service worker follows:
self.addEventListener("fetch", (event) => {
if (event.request.mode === "navigate") {
// this part I added to circumvent the issue
if (event.request.url.match(/SignOut/)) {
return false;
}
event.respondWith(
(async () => {
try {
const networkResponse = await fetch(event.request);
return networkResponse;
} catch (error) {
// This section is irrelevant, I've confirmed it doesn't run when the error happens
}
})()
);
}
});
I couldn't find any information on why the browser would download the page instead of navigating to it, but it only happens when the service worker processes the /SignOut/ URL, and only on that URL that I have noticed. Any ideas?

How to wait the page to test is loaded in non angular site?

I've tried this:
browser.wait(function () {
return browser.executeScript('return document.readyState==="complete" &&' +
' jQuery !== undefined && jQuery.active==0;').then(function (text) {
return text === true;
});
}, 30000);
If jQuery.active==0 then page is completely loaded. This should work for sites with JQuery and non angular pages.
However, I have many problems of instability to test for non angular sites.
How to fix this?
By default protractor waits until the page is loaded completely. If you are facing any error then it is because protractor is waiting for the default time to be completed, that you have specified in your conf.js file to wait until page loads. Change the value to wait a for longer time if you think your app is slow -
// How long to wait for a page to load.
getPageTimeout: 10000, //Increase this time to whatever you think is better
You can also increase the defaultTimeoutInterval to make protractor wait a little longer before the test fails -
jasmineNodeOpts: {
// Default time to wait in ms before a test fails.
defaultTimeoutInterval: 30000
},
If you want to wait for any particular element, then you can do so by using wait() function. Probably waiting for last element to load is the best way to test it. Here's how -
var EC = protractor.ExpectedConditions;
var lastElement = element(LOCATOR_OF_LAST_ELEMENT);
browser.wait(EC.visibilityOf(lastElement), 10000).then(function(){ //Alternatively change the visibilityOf to presenceOf to check for the element's presence only
//Perform operation on the last element
});
Hope it helps.
I use ExpectedConditions to wait for, and verify page loads. I walk through it a bit on my site, and example code on GitHub. Here's the gist...
Base Page: (gets extended by all page objects)
// wait for & verify correct page is loaded
this.at = function() {
var that = this;
return browser.wait(function() {
// call the page's pageLoaded method
return that.pageLoaded();
}, 5000);
};
// navigate to a page
this.to = function() {
browser.get(this.url, 5000);
// wait and verify we're on the expected page
return this.at();
};
...
Page Object:
var QsHomePage = function() {
this.url = 'http://qualityshepherd.com';
// pageLoaded uses Expected Conditions `and()`, that allows us to use
// any number of functions to wait for, and test we're on a given page
this.pageLoaded = this.and(
this.hasText($('h1.site-title'), 'Quality Shepherd')
...
};
QsHomePage.prototype = basePage; // extend basePage
module.exports = new QsHomePage();
The page object may contain a url (if direct access is possible), and a pageLoaded property that returns the ExepectedCondition function that we use to prove the page is loaded (and the right page).
Usage:
describe('Quality Shepherd blog', function() {
beforeEach(function() {
// go to page
qsHomePage.to();
});
it('home link should navigate home', function() {
qsHomePage.homeLink.click();
// wait and verify we're on expected page
expect(qsHomePage.at()).toBe(true);
});
});
Calling at() calls the ExpectedCondidion (which can be be an and() or an or(), etc...).
Hope this helps...

How do I know that I'm still on the correct page when an async callback returns?

I'm building a Metro app using the single-page navigation model. On one of my pages I start an async ajax request that fetches some information. When the request returns I want to insert the received information into the displayed page.
For example:
WinJS.UI.Pages.define("/showstuff.html", {
processed: function (element, options) {
WinJS.xhr(...).done(function (result) {
element.querySelector('#target').innerText = result.responseText;
});
}
};
But how do I know that the user hasn't navigated away from the page in the meantime? It doesn't make sense to try to insert the text on a different page, so how can I make sure that the page that was loading when the request started is still active?
You can compare the pages URI with the current WinJS.Navigation.location to check if you are still on the page. You can use Windows.Foundation.Uri to pull the path from the pages URI to do this.
WinJS.UI.Pages.define("/showstuff.html", {
processed: function (element, options) {
var page = this;
WinJS.xhr(...).done(function (result) {
if (new Windows.Foundation.Uri(page.uri).path !== WinJS.Navigation.location)
return;
element.querySelector('#target').innerText = result.responseText;
});
}
};
I couldn't find an official way to do this, so I implemented a workaround.
WinJS.Navigation provides events that are fired on navigation. I used the navigating event to build a simple class that keeps track of page views:
var PageViewManager = WinJS.Class.define(
function () {
this.current = 0;
WinJS.Navigation.addEventListener('navigating',
this._handleNavigating.bind(this));
}, {
_handleNavigating: function (eventInfo) {
this.current++;
}
});
Application.pageViews = new PageViewManager();
The class increments a counter each time the user starts a new navigation.
With that counter, the Ajax request can check if any navigation occurred and react accordingly:
WinJS.UI.Pages.define("/showstuff.html", {
processed: function (element, options) {
var pageview = Application.pageViews.current;
WinJS.xhr(...).done(function (result) {
if (Application.pageViews.current != pageview)
return;
element.querySelector('#target').innerText = result.responseText;
});
}
};