---------- Forwarded message ----------
From: Mark S. Miller <erig...@google.com>
Date: Tue, Nov 16, 2010 at 3:44 PM
On November 10th and 11th, a number of Google teams representing a variety
of viewpoints on client-side languages met to agree on a common vision for
is the result. It was first announced on Buzz at
Please forward this message to people and groups that should know about
this. This internal list, java...@google.com, is the place we
should have the Google-wide discussion of this document and these issues. If
you'd like to join this discussion, please subscribe at <
- Harmony (low risk/low reward): continue working in conjunction with
- Dash (high risk/high reward): Develop a new language (called Dash) that
performance profile and be amenable to tooling for large projects. Push for
Dash to become an open standard and be adopted by other browsers. Developers
for browsers that do not support Dash natively.
That’s the 10,000 foot overview. For more detail (including an FAQ), read
the web today is far too difficult. The cyclone of innovation is
increasingly moving off the web onto iOS and other closed platforms.
web has begun to outgrown it. The web development community has been
backed into using large amounts of JS largely to work around the
deficiencies in the platform. Complex web apps--the kind that Google
specializes in--are struggling against the platform and working with a
language that cannot be tooled and has inherent performance problems. Even
smaller-scale apps written by hobbyist developers have to navigate a
confusing labyrinth of frameworks and incompatible design patterns.
The web has succeeded historically to some extent in spite of the web
platform, based primarily on the strength of its reach. The emergence of
compelling alternative platforms like iOS has meant that the web platform
today will likely not be a viable solution long-term. Something must
Overview of two-pronged solution
There are two ways to approach the problem: either we can try to evolve
case it will take years and will be limited by fundamental problems in the
historical baggage that cannot be solved without a clean break. Thus,
although it’s low risk, it’s also relatively low reward.
The “clean break” option is extremely high risk--it will be a huge challenge
to convince other browser vendors to rally around a new language--but is the
risk is matched by the potential for a very high reward--a classic leapfrog
Pursuing either strategy in isolation is likely to fail. The evolve
state and unable to compete against the encroachment of other, less open
platforms. The clean break strategy, in isolation, would leave us in an
slowed down or evolved in undesirable ways without our support, we would
still have the fundamental flaws, and--worst of all--Google’s leadership
position on the web would be seriously damaged.
The only solution is to execute the two strategies in parallel. When the
leapfrog attempt succeeds (that is, it is an open standard and browsers
covering a majority of market share implement it), web programmers will have
It is paramount that Google continue to maintain a leadership position on
important open web standards such as Harmony. Harmony is the name of the
agreed trajectory of EcmaScript in TC39. Our JS++ project (part of the
larger Parkour project) will join with our Caja project’s efforts to advance
Harmony. Together, we will focus on improving the public Harmony spec and
helping drive it forward at a faster pace in external standard committees
and by leading by example in Chrome wherever possible.
In order to help speed up what can be a long and drawn out standardization
process, the internal Harmony effort will experiment using a preprocessor
on top of V8 to prototype features in a way that allows real code to be
written against the proposal. Details of this approach are yet to be
determined. The effort will also work with other browser vendors (e.g.
Mozilla) to get experimental support included, providing further pressure to
get Harmony standardized and widely implemented quickly. Harmony will be
implemented in V8 and JSC (Safari) simultaneously to avoid a WebKit
Developers who can focus solely on Chrome can expect to be able to see some
Harmony features in Chrome (behind a flag) by the middle of 2011. Developers
focusing on all browsers will have to wait multiple years for direct Harmony
support, due to the relatively slow pace of the standardization process. To
enable Harmony developers to target all earlier browsers, we will enhance
our source-to-source translators (such as Caja’s ES5-to-ES3 translator) to
Harmony will continue to be evangelized by Google externally as the
building on the web platform who wish to write standards-compliant
Harmony for those that need it.
Dash: The Clean Break
Dash is the leapfrog effort that is designed to be a clean break from
successful, but fill in holes everyone agrees it has.
Dash is designed with three perspectives in mind:
- Performance -- Dash is designed with performance characteristics in
mind, so that it is possible to create VMs that do not have the performance
problems that all EcmaScript VMs must have.
- Developer Usability -- Dash is designed to keep the dynamic,
platform the clear winner for hobbyist developers.
- Ability to be Tooled -- Dash is designed to be more easily tooled (e.g.
with optional types) for large-scale projects that require
code-comprehension features such as refactoring and finding callsites.
Dash, however, does not require tooling to be effective--small-scale
developers may still be satisfied with a text editor.
Dash is also designed to be securable, where that ability does not seriously
conflict with the three main goals.
Dash will be designed to be consumed in a number of locations:
- Browser VM -- Our aspiration is that Dash will ultimately be a viable
across all browsers.
- Front-end Server -- Dash will be designed as a language that can be
used server-side for things up to the size of Google-scale Front Ends. This
will allow large scale applications to unify on a single language for client
and front end code.
- Dash Cross Compiler -- Dash will be designed so that a large subset of
commit to using Dash do not have to seriously limit their reach. Platforms
that have a Dash VM can operate on the original Dash code without
translation and take advantage of the increased performance. One of the
ways we will evolve Harmony is to be a better target for such compiled Dash
lingua franca of web development on the open web platform. We will
proactively evangelize Dash with web developers and all other browser
vendors and actively push for its standardization and adoption across the
board. This will be a difficult effort requiring finesse and
determination, but we are committed to doing everything possible to help it
While Dash is catching on with other browsers, we will promote it as the
language for serious web development on the web platform; the compiler
allows such developers to target other browsers before those browsers
The Dash language effort will be driven by Lars Bak and his team in the
Aarhus office. Bruce Johnson’s team in Atlanta will handle the tooling, and
Pavel Feldman in STP will provide Web Inspector level support for Dash and
Dash will be spec complete and have working bits for the browser in Q1 2011.
Developers who can focus solely on Chrome can expect to be able to rely on
some Dash features built into Chrome within a year. Developers focusing on
all browsers will have to make use of the Dash cross compiler to target
other browsers, and, depending on the success of the evangelizing effort,
might have to wait years for other browsers to implement native support for
Although Dash is in the early stages of development, work is progressing
rapidly. You can learn more about the current proposal in this
FAQ Who authored this document?
Brad Abrams, Erik Arvidsson, Lars Bak, Darin Fisher, Dimitri Glazkov, Dan
Grove, Peter Hallam, Bruce Johnson, Alex Komoroske, John Lenz, Kasper Lund,
Mark Miller , Ivan Posva, Alex Russell, and Joel Webber, who collectively
represent TC39 (the EcmaScript standards body), WebKit, Parkour, Brightly,
JSPrime, JS++, Closure, JSCompiler, V8, Dash, Joy, and GWT, among others.
What happened to JSPrime?
The JSPrime effort was begun to unify and be a (single!) successor to GWT
and Closure/JSCompiler, suitable for large-scale development inside and
outside Google, including being amenable to IDE-like tools and static
compiler optimizations. The JSPrime team is happily folding its efforts into
Dash now that everyone agrees Dash will explicitly include the same goals.
What happened to JS++?
The collection of features under the JS++ umbrella have been folded into
advantage of our DOM improvements. This is a better plan because it gives
What happened to Joy?
The Joy templating and MVC systems are higher-level frameworks that will be
built on top of Dash.
Where can I learn more about Dash?
Dash is still in the early stages of development, but work is progressing
rapidly. For an early look at the current proposal, see this
(JQuery, Analytics, etc)
Moving to a new language will be a very large undertaking. The specifics of
What about the existing code bases for large Google Apps? Won’t they have
to rebuild everything to take advantage of Dash?
The Dash Cross Compiler should be capable of taking typed Closure code (with
some restrictions) and converting to Dash. Although the migration process
won’t be fully automatic, it should make moving over to a Dash codebase
How does this affect Web Inspector?
of Harmony that we add to chrome.
How does this affect our cloud IDE (Brightly)?
Brightly will enable building any web application in V1 using today’s
will support Dash as well. We expect that the more prescriptive development
aspects of Brightly that will come on line in the future will be more Dash
We expect Brightly itself to be the first application written in Dash.
How will we get Harmony related changes into Chrome?
Very carefully ;-). V8 is carefully tuned for speed with the current
make experimental changes. We are considering pre-processors and a number
of other options, but ultimately the precise solution is still an open
What about Go?
Go is a very promising systems-programming language in the vein of C++. We
fully hope and expect that Go becomes the standard back-end language at
Google over the next few years. Dash is focused on client (and eventually
Front-end server development). The needs there are different (flexibility
vs. stability) and therefore a different programming language is warranted.
Will Dash run on the Server? Android?
Yes, but short term we are focused on the client.
Does Dash replace Java?
For many projects that will be a viable option but it requires significant
engineering effort on Dash tooling and an extensive set of libraries.
Is Dash statically typed and toolable?
Dash is optionally-typed and with judicious use of types is as toolable as
Java. This enables “grown up” developer tools such as code-refactoring,
while still allowing small-scale or experimental projects the flexibility
that dynamism provides.
What is the future of the JSCompiler and GWT?
JSCompiler and GWT were already on a merger path. This effort gives us a
direction for that unification around the Dash language. We will actively
support teams for a long time on the current generation of JSCompiler and
GWT and provide fantastic co-existence and migration tools to Dash.
We are not! Google has a huge interest in keeping the evolution of
standards body) will likely increase somewhat, and we will continue to
honestly and whole-heartedly improve the language within the constraints.
What are the time frames?
The Dash VM and Dash Cross Compiler will be developed in parallel with the
language specification, and so should be available not long after the spec
is settled (likely in early 2011). However, the initial versions will not
be heavily optimized (and thus not necessarily ready for production apps)
until later (likely later 2011).
Experimental Harmony features will begin showing up in Chrome (behind a
flag) by midway through 2011, and will later be implemented simultaneously
Why do you have two projects? Why not just one?
See the section above about why we’re pursuing a two-pronged strategy.
What will Google developers be using?
We will strongly encourage Google developers start off targeting Chrome-only
whenever possible as this gives us the best end user experience. However,
for some apps this will not make sense, so we are building a compiler for
using GWT and JSCompiler to eventually migrate to the Dash compiler.
What if other browsers don’t follow us with Dash?
Lars has promised to “sweet talk” the other browser vendors and, while we
are all eager to see this, we recognize this is a very difficult road. Our
approach is to make an absolutely fantastic VM/Language and development
environment and build great apps that fully leverage it in order to help
other browsers see the wisdom in following. Once Dash has had a chance to
prove its stability and feasibility, we are committed to making Dash an open
standard with involvement from the broader web community.
However, in the event that other browsers don’t follow, Dash can still be a
success. Developers that target only Chrome can rely on the Dash VM, and
developers that target other browsers as well can still make use of the Dash
standard language--and that’s precisely why we must continue investing in
Why are you circumventing the standards process?
We fully intend to cooperate fully with standards processes--the problem is
viable in the long term. Any effort with the historic baggage that
make progress, and then engage the community.
What will we say at Google IO about Dash/Harmony?
Google deeply cares about the web. We care about making the web
incrementally better today (Harmony) as well as making it substantially
better in the future (Dash). Large scale applications should probably build
on Dash; smaller-scale developers might want to stick with Harmony until the
Dash standard gains ubiquity. Given that Dash is such a big bet we are
likely to spend much more time at IO on Dash, though of course we will spend
some time on the leadership position Google is taking in Harmony.