Indexed DB: Prevent inspector from re-creating deleted databases

Indexed DB: Prevent inspector from re-creating deleted databases

The dev tools resources panel maintains some state (e.g. selected
database/store names) to support actions like paging through data, but
doesn't keep connections open as that would block script actions (like
upgrades). If a database is deleted by script and the UI is poked the
inspector will attempt to re-open the database by name. But with the
Indexed DB API this will implicitly re-create it (at version 1, no
stores).

Fix this by handling the "upgradeneeded" event and aborting the
upgrade transaction, which is the same way that script would avoid
implicitly recreating a database.

BUG=539931
R=cmumford@chromium.org

Committed: https://crrev.com/223dd6ea885c085cf5a600345dd164c3e03a0ca3
Cr-Commit-Position: refs/heads/master@{#355651}

[jsbell, 2015-10-21 22:31:36]

Needs tests.

[cmumford, 2015-10-22 16:10:43]

lgtm

https://codereview.chromium.org/1411633005/diff/1/third_party/WebKit/Source/m...
File third_party/WebKit/Source/modules/indexeddb/InspectorIndexedDBAgent.cpp
(right):

https://codereview.chromium.org/1411633005/diff/1/third_party/WebKit/Source/m...
third_party/WebKit/Source/modules/indexeddb/InspectorIndexedDBAgent.cpp:213: if
(event->type() != EventTypeNames::upgradeneeded) {
Why not just an ASSERT for this as we're only observing this one type?

https://codereview.chromium.org/1411633005/diff/1/third_party/WebKit/Source/m...
third_party/WebKit/Source/modules/indexeddb/InspectorIndexedDBAgent.cpp:219: //
had previously been enumerate was deleted. We don't want to
s/enumerate/enumerated/

[jsbell, 2015-10-22 19:57:31]

Added tests; please take another look?

https://codereview.chromium.org/1411633005/diff/1/third_party/WebKit/Source/m...
File third_party/WebKit/Source/modules/indexeddb/InspectorIndexedDBAgent.cpp
(right):

https://codereview.chromium.org/1411633005/diff/1/third_party/WebKit/Source/m...
third_party/WebKit/Source/modules/indexeddb/InspectorIndexedDBAgent.cpp:213: if
(event->type() != EventTypeNames::upgradeneeded) {
On 2015/10/22 16:10:43, cmumford wrote:
> Why not just an ASSERT for this as we're only observing this one type?

I'm leaving this as-is (copy/pasta from other handlers); I believe the intent is
just to be resilient in case the spec changes and the inspector wasn't adapted.
(Script can also generate/dispatch events, but script should also never see
these requests or we have a world isolation bug.)

If we want to change to an ASSERT we should change them all.

https://codereview.chromium.org/1411633005/diff/1/third_party/WebKit/Source/m...
third_party/WebKit/Source/modules/indexeddb/InspectorIndexedDBAgent.cpp:219: //
had previously been enumerate was deleted. We don't want to
On 2015/10/22 16:10:43, cmumford wrote:
> s/enumerate/enumerated/

Done.

[cmumford, 2015-10-22 20:30:08]

slgtm

[jsbell, 2015-10-22 21:47:26]

The CQ bit was checked by jsbell@chromium.org

[jsbell, 2015-10-22 21:47:26]

The patchset sent to the CQ was uploaded after l-g-t-m from
cmumford@chromium.org
Link to the patchset: https://codereview.chromium.org/1411633005/#ps20001
(title: "Added tests")

[commit-bot: I haz the power, 2015-10-22 21:47:51]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1411633005/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1411633005/20001

[commit-bot: I haz the power, 2015-10-22 21:52:57]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-10-22 21:53:41]

Patchset 2 (id:??) landed as
https://crrev.com/223dd6ea885c085cf5a600345dd164c3e03a0ca3
Cr-Commit-Position: refs/heads/master@{#355651}