IndexedDB: Replace use of DOMError with DOMException

IndexedDB: Replace use of DOMError with DOMException

DOMError has been removed from the DOM standard in favor of
DOMException. We've added instrumentation to measure the use of the
constructors and methods (http://crrev.com/1151023004), but given that
DOMException is drop-in compatible (same name/message properties) when
minted by an API, we can try just switching over. This would only
break code that does 'instanceof' tests, and we didn't even have any
of those for IndexedDB.

Intent: https://groups.google.com/a/chromium.org/d/msg/blink-dev/XAlBg66UB7c/G4GNCFLnwCQJ
R=philipj@opera.com
BUG=460725

[jsbell, 2015-06-12 21:59:49]

philipj@ - we may want to wait for your user counters to gather data before
doing this, but the CL is pretty straightforward. And apparently we didn't even
have any tests that would have noticed.

[philipj_slow, 2015-06-15 10:23:37]

On 2015/06/12 21:59:49, jsbell wrote:
> philipj@ - we may want to wait for your user counters to gather data before
> doing this, but the CL is pretty straightforward. And apparently we didn't
even
> have any tests that would have noticed.

This is what http://w3c.github.io/IndexedDB/ says and sure seems safe, but I had
imagined that a more likely way out of DOMError would be something like an
IDBError interface with exactly the same API surface. Maybe that's because I'm
used to seeing MediaError in HTMLMediaElement.error, which is a similar pattern
to this.

At least in Blink, it looks like this would be the first case where DOMException
is used as an attribute rather than an object that is thrown. There wasn't any
discussion on https://www.w3.org/Bugs/Public/show_bug.cgi?id=27727 or on GitHub
about this, and while there's no technical difficultly it does make me want to
ask around. Has Cameron McCormack or Anne van Kesteren commented on it? A
blink-dev intent would be a way to make sure this isn't too strange.

[jsbell, 2015-06-15 15:17:19]

On 2015/06/15 10:23:37, philipj wrote: 
> This is what http://w3c.github.io/IndexedDB/ says 

Don't put too much weight on that; since I'm the active editor I'm making
changes directly and will pester others for feedback when there's a sufficient
volume of them.

> At least in Blink, it looks like this would be the first case where
DOMException
> is used as an attribute rather than an object that is thrown. There wasn't any
> discussion on https://www.w3.org/Bugs/Public/show_bug.cgi?id=27727 or on
GitHub
> about this, and while there's no technical difficultly it does make me want to
> ask around. Has Cameron McCormack or Anne van Kesteren commented on it? A
> blink-dev intent would be a way to make sure this isn't too strange.

I'll ask standards folks for feedback to see if a simple replacement was what
everyone else was envisioning.

[jsbell, 2015-06-19 16:59:05]

On 2015/06/15 15:17:19, jsbell wrote:
> I'll ask standards folks for feedback to see if a simple replacement was what
> everyone else was envisioning.

Not a peep of dissent (or anything...) on public-webapps. I guess I'll send out
an intent.

[jsbell, 2015-09-23 17:35:56]

migrated to https://codereview.chromium.org/1364823002/