The markup view will now inline a textnode in its container if and only if:
- the text node is the only child (pseudo elements included)
- the text node length is smaller than a predefined limit
If a container is expanded, its text nodes will now always be rendered in full,
no longer as a short version with an ellipsis. When selecting or navigating on
a textnode, the layout will no longer be modified on the fly.
MozReview-Commit-ID: HcDMqjbOesN
--HG--
extra : rebase_source : 908ba5c1cab86018116271402c90992c1fca6d62
extra : histedit_source : f0083b1e4c54cde0fdce12e1d139baa41e7d6cda
Previously, we removed records locally, but didn't notify the server.
We can be nice and avoid making the server buffer messages for
subscriptions that the client will never use again.
MozReview-Commit-ID: 5iohGQPHXuz
--HG--
extra : rebase_source : 00639fe1016ffd241c18643049f4b061740025f1
Previously, all `PushNotifier` methods checked the process type, and
either called `Notify*{Workers, Observers}` or sent an IPDL message.
The message handlers then called back in to `PushNotifier` from the
remote process.
This was clearer when we only sent worker event notifications to the
content process, and fired all observer notifications in the parent.
It became more complicated once we started notifying observers for all
subscriptions in both processes (bug 1266433). This makes it harder to
see omissions; for example, "push-subscription-modified" isn't
currently forwarded to the child, but "push-subscription-change" and
"push-message" are.
This patch moves the remoting code into `PushNotifier::Dispatch`, and
adds a base `PushDispatcher` class. Each notification type subclasses
this class and implements logic for sending messages and firing
observers and worker events. It's more code, but a bit easier to see
which methods are called where.
MozReview-Commit-ID: 7Q0eD7qXOrW
--HG--
extra : rebase_source : c69acb95a0cb5470cf1c804639971be41b976cc2
Now that digits have been filtered out of the message manager message
names to avoid creating thousands of similar keys, we can reenable the
telemetry by renaming it. Also, update the description to address
bsmedberg's comments, and add me to the list of alert emails.
There are a huge number of different message manager messages with
names of the form "ublock0:sb:{N}", where {N} is some number from 1 to
over 1000. Having so many different keys seems to cause problems for
telemetry and makes it harder to tell how many messages of each type
there really are, so this patch combines them by eliminating any
digits. It will also help for the webdev tools that use channels with
names like "debug:server1.conn5.child1:packet". This will create some
ambiguity (eg there are some messages of the form "ublock:sb:{N}"),
but that should be a minor issue.
Even if the event handler calls `subscribe()` or `getSubscription()`,
the "readwrite" IDB transactions in `clearIf` and `forEachOrigin`
should execute first.
MozReview-Commit-ID: ETYGmnOIuag
--HG--
extra : rebase_source : 8f8847a825d1fcdb09a421b852e86b81431f7e8e
This notably fails for me on Windows, for which I've filed bug 1276802. Until that's
figured out, this test is disabled on Windows.
MozReview-Commit-ID: Hx24driJ80w
--HG--
extra : rebase_source : 509454294eb3bc4babeba4dc7992b13dd7f175d5
extra : histedit_source : 405bf4126a767c3718e1511392edfb45bfee7433
There are a series of tests strewn about the tree that seem to exercise window
opening in one form or another, so I thought I'd put them under a tag.
MozReview-Commit-ID: 74JwLUTzaMU
--HG--
extra : rebase_source : 0c92084914ffe44a781f034b2416e17905656607
We need to preload both the inactive and the active images. All our image loads
need to be kicked off before the window document's load event fires. While the
window is hidden, it is inactive. So we need the preloading hack to kick off
the loads for the active images.
MozReview-Commit-ID: 2dPdJNORwhw
--HG--
extra : rebase_source : f7e86186440b10d170ed4c27e3e138457d945089
extra : amend_source : aaf196756c4b8250916942ecc99aca46a19484ae
