angular-patternfly
angular-patternfly copied to clipboard
Notification drawer heading template is wrapped in h4
The current notification-drawer.html
template wraps the user defined heading.html
(from headingInclude
) in an h4
, which is restrictive for styling beyond a simple heading.
In progress web console PR here.
Here is where we are going with the openshift web-console:

From notification-drawer.html
it would be ideal to eliminate the h4
. If preferred for out of the box styling, move it inside the heading.html
example template:
<div class="panel-heading">
<h4 class="panel-title">
<a ng-if="!$ctrl.singleGroup" ng-click="$ctrl.toggleCollapse(notificationGroup)" ng-class="{collapsed: !notificationGroup.open}" ng-include src="$ctrl.headingInclude"></a>
<span ng-if="$ctrl.singleGroup" ng-include src="$ctrl.headingInclude"></span>
</h4>
<span class="panel-counter" ng-include src="$ctrl.subheadingInclude"></span>
</div>
@jeff-phillips-18 thoughts?
we will need to look into backwards compatibility issues that could arise from making this change but I can understand the reasoning.
Could we eliminate the h4
and update .panel-title
and set font-weight and size to match what an h4 would be? Then apps could overwrite the .panel-title
class.
We could. The issue would be any applications already overriding this using the h4 as the selector.
How closely is angular-patternfly trying to follow SemVer? Strictly no breaking changes until next major release?
Yes, that is what we are trying for.
So Angular (2) seems to follow SemVer, but is on version 4.4.2
with 5.0.0-beta.7
coming up soon. They seem to bump the major version semi-frequently, so the SemVer signal "something breaks" is there, but its not a big delayed rollout.
Curious if that would work for angular-patternfly. Following SemVer is great if 1.) you can go ahead and release the major as often as you have a breaking change (even if minor like an h4
), or 2.) if code changes are so infrequent that its ok to wait a relatively long time for a major release.
I'm guessing its hard to put enough architectural time into each new component to ensure tweaks (breaking tweaks) are avoided.
Yes, that is the struggle. I don't see the h4 really causing enough of an issue to warrant making the change and creating a major release. Is this something that you feel can not be (easily) worked around?
Agree. Not a problem for now, I think we have an acceptable workaround in place in the console.
OK, closing this issue. Feel free to reopen if you'd like it kept in the backlog for the next major release.
I'd probably prefer to keep it open. Its definitely still strange to work around the heading.
OK, thanks @benjaminapetersen