| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
Closes #3123
|
|
It is best to emphasize that the "controller" property needs to be min safe
Closes #3125
|
|
Sometimes IE returns an empty string for its normalized href on a tags.
This should pass the sanitation check in $compile.
Closes #2219, #2593
|
|
Give the java build process 2g of memory
|
|
Do not trasnform arrays into objects when generating the locale objects
Add unit test for this check
|
|
The stock Android browser doesn't support the current for-in body/style
detection for animations and transitions but we can manually fix this.
This is useful for PhoneGap web-views or traditional web-apps using the
stock browser.
|
|
|
|
Enhances sandboxing of Angular Expressions to prevent attacks via:
{}.toString.constructor(alert("evil JS code"))
|
|
Match jQuery behavior when prepending array into empty element
|
|
Previously an element like
<div class="foo ng-cloak">...</div>
would still be annoyingly visible if it matched a CSS rule like
.foo { display: inline-block; }, overriding ng-cloak's display: none.
|
|
Add tick and cross mark corresponding to their respective unicodes.
|
|
|
|
Previously if a template contained a directive that had a template
(sync or async) and the directive template was to replace the original
element and the directive template contained another directive on the
root element of this template and this new directive was an element
transclude directive then an infinite recursion would follow because
the compiler kept on re-adding and reapplying the original directive
to the replaced node.
This change fixes that.
Closes #2155
|
|
This reverts commit 15e1a29cd08993b599f390e83a249ec17f753972.
The original commit was fixing two issues - one of them was
preventing attributes that triggered directives that replaced
the compiled node to be merged into the new node.
This change was a breaking change (as seen in the diff of the
tests in this commit) and that's why it's being removed.
A proper fix will follow.
|
|
The quotes were causing the minerr processing to fail on Windows
|
|
- parallelize the tasks
- cache requests (e2e tests)
This reduces the time from ~18min to ~12min.
It makes the output little messy. We could buffer output of each task and display it once it's fully finished, nicely. I think giving instant feedback is better.
|
|
|
|
|
|
This reverts commit 0c6fb665a4e2e1e7ceb11372153963658d4b53b1.
The change invalidated the test because the point of the the test
was to test that an element directive works. Changing it to attribute
directive was wrong.
|
|
|
|
|
|
|
|
|
|
|
|
parseKeyValue and toKeyValue can now handle duplicate values in the query.
```
?x=1&x=2 <-> {x:[1,2]}
```
The algorithm looks like:
1)parseKeyValue looks for presence of obj[key]
2)detects and replaces obj[key] with [obj[key],val]
3)then pushes more duplicates if necessary
4)toKeyValue decodes array correctly
5)(not changed)$location.search({param: 'key'}) still replaces if necessary
6)(not changed)$location.search({param: ['key1', 'key2']}) sets the url with duplicates
BREAKING CHANGE: Before this change:
- `parseKeyValue` only took the last key overwriting all the previous keys;
- `toKeyValue` joined the keys together in a comma delimited string.
This was deemed buggy behavior. If your server relied on this behavior
then either the server should be fixed or a simple serialization of
the array should be done on the client before passing it to $location.
|
|
This plugin is shipped as a default one with Karma. It's specified as a peer dependency.
I assume, there's an old version of NPM on the CI server, which does not support peerDependencies and therefore it didn't get installed.
This will make the dependency explicit.
|
|
|
|
In Angular.toJson, any properties with a leading '$' character will be
stripped from the resulting string since angular uses this notation
internally for services. There have been complaints of not knowing
about this functionality until it breaks within their code.
|
|
img[src]="https://foo" has the unfortunate problem that the browser will
actually try retrieving the resource the non FQDN foo. The local DNS
might suffix a domain to this, resolve it, and try to present a
certificate for the https request and prompt the user to pick a
certificate. This commit avoids that by making foo a FQDN. Note that it
might be better to replace foo with example.com (ref
http://tools.ietf.org/html/rfc2606#section-3).
|
|
I use a .agignore file to skip the build directory and some other files
while searching using Ag (see
https://github.com/ggreer/the_silver_searcher).
|
|
|
|
This should not affect the Jenkins build at all.
Now, the Travis build uses Chrome on Sauce Labs, which in theory gives us opportunity to use any
browser/platform that Sauce Labs offers.
|
|
|
|
And also add shared config to make karma configs a bit simpler.
|
|
|
|
|
|
|
|
|
|
Closes# 2155
|
|
|
|
|
|
|
|
The description of the input selector made it seem that you were selecting
an input element based upon it's name attribute. In reality, you are
selecting an element by the string in the ng-model attribute.
|
|
Closes #3075
|
|
|
|
The chained scope creation example at the bottom of this document was using the childCtrl to create the babyScope, instead of the childScope.
|
|
Closes #3021
|
|
|
