[xwiki-devs] [Proposal] Introduce a first set of UI Extension Points

Hi, following the previous thread I'd like to make a new proposal, introduce the following UIXPs in our templates : * header top: after <div id="xwikimaincontainerinner"> * header bottom : before the closing </div> * content top: after <div id="contentcontainer"> * content bottom : before the closing </div> * footer top: after <div id="footerglobal"> * footer bottom : before the closing </div> Since we don't have a builtin mechanism (as in: have top/bottom static list independent from the UIXP ID in the UIX) for the moment we have to emulate it by inserting multiple API calls (in the current proposal: at the top and bottom of the elements). Now we need to choose a syntax for the extension point IDs: 1) Use only dots in IDs: platform.header.top, platform.header.bottom, etc I was worried about using dots at first [1] but I guess that if we want to have a mapping between the HTML skeleton and UIXP IDs we'll use html5 custom attributes (<element data-uixp="platform.header" />), since we're talking about the future here. 2) Use a different separator for the position: platform.header#top If we decide to create a "position" static list in UIX, a specific separator for the position would allow to easily migrate existing UIXs. My preference goes to 2). Here are the UIXPs that would appear in the current homepage (after renaming the current IDs): * platform.header#top * platform.header#bottom * platform.content#top * platform.content#bottom * platform.panels.rightPanels * platform.panels.Applications * platform.panels.Applications.more * platform.footer#top * platform.footer#bottom [1] The problem dots is that if we decide to move towards a more generic insertion mechanism (http://dev.xwiki.org/xwiki/bin/view/Design/InterfaceExtensions#HInterfaceEx…) with a mapping between HTML IDs and UIPX IDs, the IDs will not be HTML friendly. It's not recommended to use dots in HTML elements ID and Name attributes because of HTML selectors (#id.class is a CSS selector that will look for <element id="id" class="class"/> and not <element id="id.class"/>) which then requires to escape dots in IDs when manipulating the DOM. WDYT ? JV.