State of Embedding in Gecko

Following up from my last post, I’ve had some time to research and assess the current state of embedding Gecko. This post will serve as a (likely incomplete) assessment of where we are today, and what I think the sensible path forward would be. Please note that these are my personal opinions and not those of Mozilla. Mozilla are gracious enough to employ me, but I don’t yet get to decide on our direction ­čśë

The TLDR; there are no first-class Gecko embedding solutions as of writing.

EmbedLite (aka IPCLite)

EmbedLite is an interesting solution for embedding Gecko that relies on e10s (Electrolysis, Gecko’s out-of-process feature code-name) and OMTC (Off-Main-Thread Compositing). From what I can tell, the embedding app creates a new platform-specific compositor object that attaches to a window, and with e10s, a separate process is spawned to handle the brunt of the work (rendering the site, running JS, handling events, etc.). The existing widget API is exposed via IPC, which allows you to synthesise events, handle navigation, etc. This builds using the xulrunner application target, which unfortunately no longer exists. This project was last synced with Gecko on April 2nd 2015 (the day before my birthday!).

The most interesting thing about this project is how much code it reuses in the tree, and how little modification is required to support it (almost none – most of the changes are entirely reasonable, even outside of an embedding context). That we haven’t supported this effort seems insane to me, especially as it’s been shipping for a while as the basis for the browser in the (now defunct?) Jolla smartphone.

Building this was a pain, on Fedora 22 I was not able to get the desktop Qt build to compile, even after some effort, but I was able to compile the desktop Gtk build (trivial patches required). Unfortunately, there’s no support code provided for the Gtk version and I don’t think it’s worth the time me implementing that, given that this is essentially a dead project. A huge shame that we missed this opportunity, this would have been a good base for a lightweight, relatively easily maintained embedding solution. The quality of the work done on this seems quite high to me, after a brief examination.

Spidernode

Spidernode is a port of Node.js that uses Gecko’s ‘spidermonkey’ JavaScript engine instead of Chrome’s V8. Not really a Gecko embedding solution, but certainly something worth exploring as a way to enable more people to use Mozilla technology. Being a much smaller project, of much more limited scope, I had no issues building and testing this.

Node.js using spidermonkey ought to provide some interesting advantages over a V8-based Node. Namely, modern language features, asm.js (though I suppose this will soon be supplanted by WebAssembly) and speed. Spidernode is unfortunately unmaintained since early 2012, but I thought it would be interesting to do a simple performance test. Using the (very flawed) technique detailed here, I ran a few quick tests to compare an old copy of Node I had installed (~0.12), current stable Node (4.3.2) and this very old (~0.5) Spidermonkey-based Node. Spidermonkey-based Node was consistently over 3x faster than both old and current (which varied very little in performance). I don’t think you can really draw any conclusions than this, other than that it’s an avenue worth exploring.

Many new projects are prototyped (and indeed, fully developed) in Node.js these days; particularly Internet-Of-Things projects. If there’s the potential for these projects to run faster, unchanged, this seems like a worthy project to me. Even forgetting about the advantages of better language support. It’s sad to me that we’re experimenting with IoT projects here at Mozilla and so many of these experiments don’t promote our technology at all. This may be an irrational response, however.

GeckoView

GeckoView is the only currently maintained embedding solution for Gecko, and is Android-only. GeckoView is an Android project, split out of Firefox for Android and using the same interfaces with Gecko. It provides an embeddable widget that can be used instead of the system-provided WebView. This is not a first-class project from what I can tell, there are many bugs and many missing features, as its use outside of Firefox for Android is not considered a priority. Due to this dependency, however, one would assume that at least GeckoView will see updates for the foreseeable future.

I’d experimented with this in the past, specifically with this project that uses GeckoView with Cordova. I found then that the experience wasn’t great, due to the huge size of the GeckoView library and the numerous bugs, but this was a while ago and YMMV. Some of those bugs were down to GeckoView not using the shared APZC, a bug which has since been fixed, at least for Nightly builds. The situation may be better now than it was then.

The Future

This post is built on the premise that embedding Gecko is a worthwhile pursuit. Others may disagree about this. I’ll point to my previous post to list some of the numerous opportunities we missed, partly because we don’t have an embedding story, but I’m going to conjecture as to what some of our next missed opportunities might be.

IoT is generating a lot of buzz at the moment. I’m dubious that there’s much decent consumer use of IoT, at least that people will get excited about as opposed to property developers, but if I could predict trends, I’d have likely retired rich already. Let’s assume that consumer IoT will take off, beyond internet-connected thermostats (which are actually pretty great) and metered utility boxes (which I would quite like). These devices are mostly bespoke hardware running random bits and bobs, but an emerging trend seems to be Node.js usage. It might be important for Mozilla to provide an easily deployed out-of-the-box solution here. As our market share diminishes, so does our test-bed and contribution base for our (currently rather excellent) JavaScript engine. While we don’t have an issue here at the moment, if we find that a huge influx of diverse, resource-constrained devices starts running V8 and only V8, we may eventually find it hard to compete. It could easily be argued that it isn’t important for our solution to be based on our technology, but I would argue that if we have to start employing a considerable amount of people with no knowledge of our platform, our platform will suffer. By providing a licensed out-of-the-box solution, we could also enforce that any client-side interface remain network-accessible and cross-browser compatible.

A less tenuous example, let’s talk about VR. VR is also looking like it might finally break out into the mid/high-end consumer realm this year, with heavy investment from Facebook (via Oculus), Valve/HTC (SteamVR/Vive), Sony (Playstation VR), Microsoft (HoloLens), Samsung (GearVR) and others. Mozilla are rightly investing in WebVR, but I think the real end-goal for VR is an integrated device with no tether (certainly Microsoft and Samsung seem to agree with me here). So there may well be a new class of device on the horizon, with new kinds of browsers and ways of experiencing and integrating the web. Can we afford to not let people experiment with our technology here? I love Mozilla, but I have serious doubts that the next big thing in VR is going to come from us. That there’s no supported way of embedding Gecko worries me for future classes of device like this.

In-vehicle information/entertainment systems are possibly something that will become more of the norm, now that similar devices have become such commodity. Interestingly, the current big desktop and mobile players have very little presence here, and (mostly awful) bespoke solutions are rife. Again, can we afford to make our technology inaccessible to the people that are experimenting in this area? Is having just a good desktop browser enough? Can we really say that’s going to remain how people access the internet for the next 10 years? Probably, but I wouldn’t want to bet everything on that.

A plan

If we want an embedding solution, I think the best way to go about it is to start from Firefox for Android. Due to the way Android used to require its applications to interface with native code, Firefox for Android is already organised in such a way that it is basically an embedding API (thus GeckoView). From this point, I think we should make some of the interfaces slightly more generic and remove the JNI dependency from the Gecko-side of the code. Firefox for Android would be the main consumer of this API and would guarantee that it’s maintained. We should allow for it to be built on Linux, Mac and Windows and provide the absolute minimum harness necessary to allow for it to be tested. We would make no guarantees about API or ABI. Externally to the Gecko tree, I would suggest that we start, and that the community maintain, a CEF-compatible library, at least at the API level, that would be a Tier-3 project, much like Firefox OS now is. This, to me, seems like the minimal-effort and most useful way of allowing embeddable Gecko.

In addition, I think we should spend some effort in maintaining a fork of Node.js LTS that uses spidermonkey. If we can promise modern language features and better performance, I expect there’s a user-base that would be interested in this. If there isn’t, fair enough, but I don’t think current experiments have had enough backing to ascertain this.

I think that both of these projects are important, so that we can enable people outside of Mozilla to innovate using our technology, and by osmosis, become educated about our mission and hopefully spread our ideals. Other organisations will do their utmost to establish a monopoly in any new emerging market, and I think it’s a shame that we have such a powerful and comprehensive technology platform and we aren’t enabling other people to use it in more diverse situations.

This post is some insightful further reading on roughly the same topic.

The case for an embeddable Gecko

Strap yourself in, this is a long post. It should be easy to skim, but the history may be interesting to some. I would like to make the point that, for a web rendering engine, being embeddable is a huge opportunity, how Gecko not being easily embeddable has meant we’ve missed several opportunities over the last few years, and how it would still be advantageous to make Gecko embeddable.

What?

Embedding Gecko means making it easy to use Gecko as a rendering engine in an arbitrary 3rd party application on any supported platform, and maintaining that support. An embeddable Gecko should make very few constraints on the embedding application and should not include unnecessary resources.

Examples

  • A 3rd party browser with a native UI
  • A game’s embedded user manual
  • OAuth authentication UI
  • A web application
  • ???

Why?

It’s hard to predict what the next technology trend will be, but there’s is a strong likelihood it’ll involve the web, and there’s a possibility it may not come from a company/group/individual with an existing web rendering engine or particular allegiance. It’s important for the health of the web and for Mozilla’s continued existence that there be multiple implementations of web standards, and that there be real competition and a balanced share of users of the various available engines.

Many technologies have emerged over the last decade or so that have incorporated web rendering or web technologies that could have leveraged Gecko;

(2007) iPhone: Instead of using an existing engine, Apple forked KHTML in 2002 and eventually created WebKit. They did investigate Gecko as an alternative, but forking another engine with a cleaner code-base ended up being a more viable route. Several rival companies were also interested in and investing in embeddable Gecko (primarily Nokia and Intel). WebKit would go on to be one of the core pieces of the first iPhone release, which included a better mobile browser than had ever been seen previously.

(2008) Chrome: Google released a WebKit-based browser that would eventually go on to eat a large part of Firefox’s user base. Chrome was initially praised for its speed and light-weightedness, but much of that was down to its multi-process architecture, something made possible by WebKit having a well thought-out embedding capability and API.

(2008) Android: Android used WebKit for its built-in browser and later for its built-in web-view. In recent times, it has switched to Chromium, showing they aren’t adverse to switching the platform to a different/better technology, and that a better embedding story can benefit a platform (Android’s built in web view can now be updated outside of the main OS, and this may well partly be thanks to Chromium’s embedding architecture). Given the quality of Android’s initial WebKit browser and WebView (which was, frankly, awful until later revisions of Android Honeycomb, and arguably remained awful until they switched to Chromium), it’s not much of a leap to think they may have considered Gecko were it easily available.

(2009) WebOS: Nothing came of this in the end, but it perhaps signalled the direction of things to come. WebOS survived and went on to be the core of LG’s Smart TV, one of the very few real competitors in that market. Perhaps if Gecko was readily available at this point, we would have had a large head start on FirefoxOS?

(2009) Samsung Smart TV: Also available in various other guises since 2007, Samsung’s Smart TV is certainly the most popular smart TV platform currently available. It appears Samsung built this from scratch in-house, but it includes many open-source projects. It’s highly likely that they would have considered a Gecko-based browser if it were possible and available.

(2011) PhantomJS: PhantomJS is a headless, scriptable browser, useful for testing site behaviour and performance. It’s used by several large companies, including Twitter, LinkedIn and Netflix. Had Gecko been more easily embeddable, such a product may well have been based on Gecko and the benefits of that would be many sites that use PhantomJS for testing perhaps having better rendering and performance characteristics on Gecko-based browsers. The demand for a Gecko-based alternative is high enough that a similar project, SlimerJS, based on Gecko was developed and released in 2013. Due to Gecko’s embedding deficiencies though, SlimerJS is not truly headless.

(2011) WIMM One: The first truly capable smart-watch, which generated a large buzz when initially released. WIMM was based on a highly-customised version of Android, and ran software that was compatible with Android, iOS and BlackBerryOS. Although it never progressed past the development kit stage, WIMM was bought by Google in 2012. It is highly likely that WIMM’s work forms the base of the Android Wear platform, released in 2014. Had something like WebOS been open, available and based on Gecko, it’s not outside the realm of possibility that this could have been Gecko based.

(2013) Blink: Google decide to fork WebKit to better build for their own uses. Blink/Chromium quickly becomes the favoured rendering engine for embedding. Google were not afraid to introduce possible incompatibility with WebKit, but also realised that embedding is an important feature to maintain.

(2014) Android Wear: Android specialised to run on watch hardware. Smart watches have yet to take off, and possibly never will (though Pebble seem to be doing alright, and every major consumer tech product company has launched one), but this is yet another area where Gecko/Mozilla have no presence. FirefoxOS may have lead us to have an easy presence in this area, but has now been largely discontinued.

(2014) Atom/Electron: Github open-sources and makes available its web-based text editor, which it built on a home-grown platform of Node.JS and Chromium, which it later called Electron. Since then, several large and very successful projects have been built on top of it, including Slack and Visual Studio Code. It’s highly likely that such diverse use of Chromium feeds back into its testing and development, making it a more robust and performant engine, and importantly, more widely used.

(2016) Brave: Former Mozilla co-founder and CTO heads a company that makes a new browser with the selling point of blocking ads and tracking by default, and doing as much as possible to protect user privacy and agency without breaking the web. Said browser is based off of Chromium, and on iOS, is a fork of Mozilla’s own WebKit-based Firefox browser. Brendan says they started based off of Gecko, but switched because it wasn’t capable of doing what they needed (due to an immature embedding API).

Current state of affairs

Chromium and V8 represent the state-of-the-art embeddable web rendering engine and JavaScript engine and have wide and varied use across many platforms. This helps reenforce Chrome’s behaviour as the de-facto standard and gradually eats away at the market share of competing engines.

WebKit is the only viable alternative for an embeddable web rendering engine and is still quite commonly used, but is generally viewed as a less up-to-date and less performant engine vs. Chromium/Blink.

Spidermonkey is generally considered to be a very nice JavaScript engine with great support for new EcmaScript features and generally great performance, but due to a rapidly changing API/ABI, doesn’t challenge V8 in terms of its use in embedded environments. Node.js is likely the largest user of embeddable V8, and is favoured even by Mozilla employees for JavaScript-based systems development.

Gecko has limited embedding capability that is not well-documented, not well-maintained and not heavily invested in. I say this with the utmost respect for those who are working on it; this is an observation and a criticism of Mozilla’s priorities as an organisation. We have at various points in history had embedding APIs/capabilities, but we have either dropped them (gtkmozembed) or let them bit-rot (IPCLite). We do currently have an embedding widget for Android that is very limited in capability when compared to the default system WebView.

Plea

It’s not too late. It’s incredibly hard to predict where technology is going, year-to-year. It was hard to predict, prior to the iPhone, that Nokia would so spectacularly fall from the top of the market. It was hard to predict when Android was released that it would ever overtake iOS, or even more surprisingly, rival it in quality (hard, but not impossible). It was hard to predict that WebOS would form the basis of a major competing Smart TV several years later. I think the examples of our missed opportunities are also good evidence that opening yourself up to as much opportunity as possible is a good indicator of future success.

If we want to form the basis of the next big thing, it’s not enough to be experimenting in new areas. We need to enable other people to experiment in new areas using our technology. Even the largest of companies have difficulty predicting the future, or taking charge of it. This is why it’s important that we make easily-embeddable Gecko a reality, and I plead with the powers that be that we make this higher priority than it has been in the past.

Web Navigation Transitions

Wow, so it’s been over a year since I last blogged. Lots has happened in that time, but I suppose that’s a subject for another post. I’d like to write a bit about something I’ve been working on for the last week or so. You may have seen Google’s proposal for navigation transitions, and if not, I suggest reading the spec and watching the demonstration. This is something that I’ve thought about for a while previously, but never put into words. After reading Google’s proposal, I fear that it’s quite complex both to implement and to author, so this pushed me both to document my idea, and to implement a proof-of-concept.

I think Google’s proposal is based on Android’s Activity Transitions, and due to Android UI’s very different display model, I don’t think this maps well to the web. Just my opinion though, and I’d be interested in hearing peoples’ thoughts. What follows is my alternative proposal. If you like, you can just jump straight to a demo, or view the source. Note that the demo currently only works in Gecko-based browsers – this is mostly because I suck, but also because other browsers have slightly inscrutable behaviour when it comes to adding stylesheets to a document. This is likely fixable, patches are most welcome.


 Navigation Transitions specification proposal

Abstract

An API will be suggested that will allow transitions to be performed between page navigations, requiring only CSS. It is intended for the API to be flexible enough to allow for animations on different pages to be performed in synchronisation, and for particular transition state to be selected on without it being necessary to interject with JavaScript.

Proposed API

Navigation transitions will be specified within a specialised stylesheet. These stylesheets will be included in the document as new link rel types. Transitions can be specified for entering and exiting the document. When the document is ready to transition, these stylesheets will be applied for the specified duration, after which they will stop applying.

Example syntax:

When navigating to a new page, the current page’s ‘transition-exit‘ stylesheet will be referenced, and the new page’s ‘transition-enter‘ stylesheet will be referenced.

When navigation is operating in a backwards direction, by the user pressing the back button in browser chrome, or when initiated from JavaScript via manipulation of the location or history objects, animations will be run in reverse. That is, the current page’s ‘transition-enter‘ stylesheet will be referenced, and animations will run in reverse, and the old page’s ‘transition-exit‘ stylesheet will be referenced, and those animations also run in reverse.

[Update]

Anne van Kesteren suggests that forcing this to be a separate stylesheet and putting the duration information in the tag is not desirable, and that it would be nicer to expose this as a media query, with the duration information available in an @-rule. Something like this:

I think this would indeed be nicer, though I think the exact naming might need some work.

Transitioning

When a navigation is initiated, the old page will stay at its current position and the new page will be overlaid over the old page, but hidden. Once the new page has finished loading it will be unhidden, the old page’s ‘transition-exit‘ stylesheet will be applied and the new page’s ‘transition-enter’ stylesheet will be applied, for the specified durations of each stylesheet.

When navigating backwards, the CSS animations timeline will be reversed. This will have the effect of modifying the meaning of animation-direction like so:

and this will also alter the start time of the animation, depending on the declared total duration of the transition. For example, if a navigation stylesheet is declared to last 0.5s and an animation has a duration of 0.25s, when navigating backwards, that animation will effectively have an animation-delay of 0.25s and run in reverse. Similarly, if it already had an animation-delay of 0.1s, the animation-delay going backwards would become 0.15s, to reflect the time when the animation would have ended.

Layer ordering will also be reversed when navigating backwards, that is, the page being navigated from will appear on top of the page being navigated backwards to.

Signals

When a transition starts, a ‘navigation-transition-startNavigationTransitionEvent will be fired on the destination page. When this event is fired, the document will have had the applicable stylesheet applied and it will be visible, but will not yet have been painted on the screen since the stylesheet was applied. When the navigation transition duration is met, a ‘navigation-transition-end‘ will be fired on the destination page. These signals can be used, amongst other things, to tidy up state and to initialise state. They can also be used to modify the DOM before the transition begins, allowing for customising the transition based on request data.

JavaScript execution could potentially cause a navigation transition to run indefinitely, it is left to the user agent’s general purpose JavaScript hang detection to mitigate this circumstance.

Considerations and limitations

Navigation transitions will not be applied if the new page does not finish loading within 1.5 seconds of its first paint. This can be mitigated by pre-loading documents, or by the use of service workers.

Stylesheet application duration will be timed from the first render after the stylesheets are applied. This should either synchronise exactly with CSS animation/transition timing, or it should be longer, but it should never be shorter.

Authors should be aware that using transitions will temporarily increase the memory footprint of their application during transitions. This can be mitigated by clear separation of UI and data, and/or by using JavaScript to manipulate the document and state when navigating to avoid keeping unused resources alive.

Navigation transitions will only be applied if both the navigating document has an exit transition and the target document has an enter transition. Similarly, when navigating backwards, the navigating document must have an enter transition and the target document must have an exit transition. Both documents must be on the same origin, or transitions will not apply. The exception to these rules is the first document load of the navigator. In this case, the enter transition will apply if all prior considerations are met.

Default transitions

It is possible for the user agent to specify default transitions, so that navigation within a particular origin will always include navigation transitions unless they are explicitly disabled by that origin. This can be done by specifying navigation transition stylesheets with no href attribute, or that have an empty href attribute.

Note that specifying default transitions in all situations may not be desirable due to the differing loading characteristics of pages on the web at large.

It is suggested that default transition stylesheets may be specified by extending the iframe element with custom ‘default-transition-enter‘ and ‘default-transition-exit‘ attributes.

Examples

Simple slide between two pages:

[page-1.html]

[page-1-exit.css]

[page-2.html]

[page-2-enter.css]


I believe that this proposal is easier to understand and use for simpler transitions than Google’s, however it becomes harder to express animations where one element is transitioning to a new position/size in a new page, and it’s also impossible to interleave contents between the two pages (as the pages will always draw separately, in the predefined order). I don’t believe this last limitation is a big issue, however, and I don’t think the cognitive load required to craft such a transition is considerably higher. In fact, you can see it demonstrated by visiting this link in a Gecko-based browser (recommended viewing in responsive design mode Ctrl+Shift+m).

I would love to hear peoples’ thoughts on this. Am I actually just totally wrong, and Google’s proposal is superior? Are there huge limitations in this proposal that I’ve not considered? Are there security implications I’ve not considered? It’s highly likely that parts of all of these are true and I’d love to hear why. You can view the source for the examples in your browser’s developer tools, but if you’d like a way to check it out more easily and suggest changes, you can also view the git source repository.

Linking CSS properties with scroll position: A proposal

As I, and many others have written before, on mobile, rendering/processing of JS is done asynchronously to responding to the user scrolling, so that we can maintain touch response and screen update. We basically have no chance of consistently hitting 60fps if we don’t do this (and you can witness what happens if you don’t by running desktop Firefox (for now)). This does mean, however, that you end up with bugs like this, where people respond in JavaScript to the scroll position changing and end up with jerky animation because there are no guarantees about the frequency or timeliness of scroll position updates. It also means that neat parallax sites like this can’t be done in quite the same way on mobile. Although this is currently only a problem on mobile, this will eventually affect desktop too. I believe that Internet Explorer already uses asynchronous composition on the desktop, and I think that’s the way we’re going in Firefox too. It’d be great to have a solution for this problem first.

It’s obvious that we could do with a way of declaring a link between a CSS property and the scroll position. My immediate thought is to do this via CSS. I had this idea for a syntax:

This would work quite similarly to standard transitions, where a limited number of properties would be supported, and perhaps their interpolation could be defined in the same way too. Relative scroll position is 0px when the scroll position of the particular axis matches the element’s offset position. This would lead to declarations like this:

This would define a transition that would grow and fade in an element as the user scrolled it towards 100px down the page, then shrink and fade out as you scrolled beyond that point.

But then Paul Rouget made me aware that Anthony Ricaud had the same idea, but instead of this slightly arcane syntax, to tie it to CSS animation keyframes. I think this is more easily implemented (at least in Firefox’s case), more flexible and more easily expressed by designers too. Much like transitions and animations, these need not be mutually exclusive though, I suppose (though the interactions between them might mean as a platform developer, it’d be in my best interests to suggest that they should :)).

I’m not aware of any proposal of this suggestion, so I’ll describe the syntax that I would expect. I think it should inherit from the CSS animation spec, but prefix the animation-* properties with scroll-. Instead of animation-duration, you would have scroll-animation-bounds. scroll-animation-bounds would describe a vector, the distance along which would determine the position of the animation. Imagine that this vector was actually a plane, that extended infinitely, perpendicular to its direction of travel; your distance along the vector is unaffected by your distance to the vector. In other words, if you had a scroll-animation-bounds that described a line going straight down, your horizontal scroll position wouldn’t affect the animation. Animation keyframes would be defined in the exact same way.

[Edit] Paul Rouget makes the suggestion that rather than having a prefixed copy of animation, that a new property be introduced, animation-controller, of which the default would be time, but a new option could be scroll. We would still need an equivalent to duration, so I would re-purpose my above-suggested property as animation-scroll-bounds.

What do people think about either of these suggestions? I’d love to hear some conversation/suggestions/criticisms in the comments, after which perhaps I can submit a revised proposal and begin an implementation.