Mostly just declaring globals that Cu.imports defines but there are some actual
bugs here that have been fixed as well as one test that just never ran because
of a hidden exception.
MozReview-Commit-ID: J6uIpYp8ANx
This currently forbids unknown top-level schema properties, and unknown
permissions. In the future, I'd like to make those warnings rather than
errors, for compatibility purposes, but I think errors are fine for now.
Previously we just checked every newly sideloaded add-on to decide whether to
offer it to the user for opt-in. This adds a new "seen" property (naming could
be better if you have other suggestions) which tracks whether we've ever shown
the opt-in UI for the add-on. It defaults to true for all add-ons and is only
set to false for sideloaded add-ons that default to being disabled on install.
The seen flag can be set to true through the API but cannot be reset to false
as that would allow add-ons to forcibly re-present themselves to the user when
disabled.
The opt-in UI sets the seen flag to true only when it has focus which fixes a
long-standing bug where if you accept the first add-on you see and restart the
other tabs might not show up.
The one slight downside of this approach is that it now requires loading the
full add-ons database on every startup in order to check the seen flag for all
installed add-ons. There are hacky ways we might get around this but they all
involve overloading prefs with even more object data. The good thing is that
we do the load and check asynchronously after most of startup is complete and
the UI is fully loaded so there shouldn't be any percieved impact to startup
time. I've run multiple talos runs to verify that none of the numbers appear to
regress.
Right now, each incremental build produces a new set of GeckoView
library files, but the previous files don't get cleaned up, and you end
up with a bunch of old libraries in your objdir. This patch cleans up
the old files before producing new ones.
geckoview_example is broken and obsolete, and we haven't maintained it
for a long time. We should remove it from the tree, allow GeckoView AARs
to build, and rely on other example GeckoView projects that live on
GitHub.
Moved these mostly onto the prototype. We couldn't do this before without making
the target of the wrapper a property of the wrappers and we don't want to expose
that but now WeakMaps allow us to get the target without exposing it.
Once change with this approach is that when the test suite shuts down the
add-ons manager it kills the map and so wrappers cease to function. A couple of
tests were relying on accessing wrapper properties after that but that would
have likely been unsafe anyway.
We used to need explicit names for functions to make stack traces display
properly. The JS engine is smarter now so doesn't need them and they just
make the code messy and redundant.
For build speed, for correct line numbers in errors, for faster development, for so many reasons.
Still a couple of cases left mostly in XUL files for different strings on Windows.
Bonus: The new lexical scope means ADDON_SIGNING and REQUIRE_SIGNING can just
be declared as regular constants and outside code can't get to them easily.
Simple obvious fix. Adds tests by making BootstrapMonitor (which
test_experiments.js and others use for verifying bootstrap startup and shutdown)
verify the list of registered chrome manifests at various points. Without the
fix this makes test_experiment fail as expected.
The Android-specific AddonUpdateService has a bit of redundant code
because AddonManagerPrivate has a backgroundUpdateCheck method that does
a lot of the same thing. This patch makes AddonUpdateService call that
method so there's less code and more consistency.
This requires flipping the "extensions.update.enabled" pref, which was
disabled in bug 528588 for showing the XUL addon update dialog. I don't
think this is relevant anymore in native Fennec and with the later
rewrite of AddonManager, so I'm fairly certain it's okay to flip that
pref.
The patch also disables the AddonManager update timer because we have
our own update timer on Android.
Experiments should differ from normal add-ons in a few ways:
* They can always be enabled regardless of compatibility info
* They default to disabled when installed
* They cannot be checked for updates
* They only stay enabled for the lifetime of the current process
* The UI doesn't give users the ability to enable/disable
This makes a few changes to keep these differences but remove much of the special casing code for experiments.
Being able to use regardless of compatibility was mostly fixed by bug 1220198 but I've also removed the redundant override in isCompatible.
Previously the "enabled until restart" feature worked with by not updating the DBAddonInternal object and instead using a hack to make the wrapper still seem enabled. This seems likely to break other code that relies on the state of the DBAddonInternal object so instead we update that as normal and simply don't persist the enabled state to disk.
Also switch the DBAddonInteral.prototype code to use some newer JS features.
I've removed the hack from addon.permissions which was hiding the enable/disable buttons in the UI and instead just hidden them in the UI stylesheet. This makes the API make sense and means callers can use addon.permissions to verify that enabling will work.
In order to meet the addon signing requirement for tests, specialpowers
needs to be installed at gecko runtime. This means it must be restartless.
This patch packages specialpowers as a restartless addon, but it does not
yet install it at runtime.
Attempt to get around the fact that XP SP2 (and below) systems won't be able to
handle Authenticode signatues with SHA-2 digests by disabling the maintenance
service, the only thing on our update path that uses Authenticode, on systems
running WinXP below SP3.
This patch both prevents the service from being included in new installations,
and prevents the service from being used where it is already installed.
I tried to keep the changes to existing tests as minimal as
possible. There were a few exceptions, though:
* test_update_ignorecompat.js was completely broken. I couldn't
figure out why it was suddenly failing after I changed it to use
`add_test`, and it turned out that it had been failing all along,
but in a way that the harness didn't pick up.
* I changed most of the `do_throw` in update callbacks to `ok(false`
because it took me about an hour to figure out where the test was
failing when I hit one of them.
* I made some changes to sync `test_update.js` and `test_update_ignorecompat.js`
where one appeared to have been changed without updating the
other.
* I made `promiseFindAddonUpdates` a bit more generic, because I was
planning to convert most of `test_update.js` to use it, rather
than nested callbacks. I changed my mind a quarter of the way
through, but decided to keep the changes, since they'll probably
be useful elsewhere.
I tried to keep the changes to existing tests as minimal as
possible. There were a few exceptions, though:
* test_update_ignorecompat.js was completely broken. I couldn't
figure out why it was suddenly failing after I changed it to use
`add_test`, and it turned out that it had been failing all along,
but in a way that the harness didn't pick up.
* I changed most of the `do_throw` in update callbacks to `ok(false`
because it took me about an hour to figure out where the test was
failing when I hit one of them.
* I made some changes to sync `test_update.js` and `test_update_ignorecompat.js`
where one appeared to have been changed without updating the
other.
* I made `promiseFindAddonUpdates` a bit more generic, because I was
planning to convert most of `test_update.js` to use it, rather
than nested callbacks. I changed my mind a quarter of the way
through, but decided to keep the changes, since they'll probably
be useful elsewhere.
If not building the Windows Maintenance Service, avoid a dependency
on crypt32.dll and wintrust.dll.
Also, avoid setting sUsingService inside the updater code; this
ensures that all Maintenance Service related codepaths are skipped.
Rather that trying to get the method from the sandbox global object which will
only work for var and function declared methods instead evaluate the function
name in the sandbox scope and get the result which will give us access to the
lexical scope.
