W3C home > Mailing lists > Public > public-webapps@w3.org > January to March 2013

Re: IndexedDB, what were the issues? How do we stop it from happening again?

From: Jonas Sicking <jonas@sicking.cc>
Date: Tue, 19 Mar 2013 15:52:05 -0700
Message-ID: <CA+c2ei-2Agm8H5V4c5mvXymbDgmAruGGgDEdBbmEMhEp1PU9JA@mail.gmail.com>
To: Robert Ginda <rginda@chromium.org>
Cc: "Tab Atkins Jr." <jackalmage@gmail.com>, Glenn Maynard <glenn@zewt.org>, Alex Russell <slightlyoff@google.com>, Tobie Langel <tobie@fb.com>, Jarred Nicholls <jarred@webkit.org>, public-webapps WG <public-webapps@w3.org>
On Tue, Mar 19, 2013 at 2:25 PM, Robert Ginda <rginda@chromium.org> wrote:
> One option that will probably never happen: Stop making jank-inducing API's
> available on the main thread.

I think we're already doing this. As far as I know all browser vendors
are on board with the idea of not having APIs which run on the main
thread and which are slow enough that they can introduce jank.
Including anything that does IO, but also things like encryption and
image decoding/encoding.

Unfortunately we already have a pretty large debt in form of janky APIs :(

> Do them all as synchronous APIs only, and
> available only to workers.

This part is more controversial I think. Forcing people to use workers
even in simple cases might not be popular.

There's also the issue that synchronous APIs generally have worse
performance. But it would need to be measured.

/ Jonas

/ Jonas
Received on Tuesday, 19 March 2013 22:53:03 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 18:49:58 GMT