Episode 418: Oct 29th, 2025
Links
Topics
- Sorry for the delays!
- More travel coming soon though…
- https://www.keanu.codes/?code=418 - mikes favourite
- Brought to you by HTTP code 418
- https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Status/418
- Previewing split view!
- browser.tabs.splitView.enabled
- Testing a punch out of a hole for newtab
- nsICategoryManager
- Hand-off Search Component
- Has a SearchComponents.manifest
- category browser-newtab-ui-component moz-src:///browser/components/search/ContentSearchUINewTabComponent.sys.mjs ContentSearchUINewTabComponent.definition
- ContentSearchUINewTabComponent.sys.mjs
- const ContentSearchNewTabUIComponent { definition() { return { type: NewTabUIComponent.TYPE_SEARCH, scripts: [dependencies, and script that defines moz-search-input], markup: { tagName: “moz-search-input”, attrs: { } }, } } }
- Something that reads from the category manager on initialization, also adds observers for “xpcom-category-entry-added” and “xpcom-category-entry-removed”, and on initialization OR when the category in question (browser-newtab-ui-component) changes, we reconstruct the list of components
- Potentially reuse CatManListenerManager, which does the above (but we need to recompute the categories on notifications)
- Interestingly, the CategoryManager has its entries populated from manifest files for each process, meaning that we’d get build-time registration for free in the privileged about content process. What we wouldn’t get is dynamic registration and unregistration
- Have the backend newtab code iterate the relevant categories, and get the return values for each entry in the category, and then to sort them by type.
- This will then create a structure which can be send down in the state object that each New Tab receives - A class (or component?) will be responsible for parsing that structure and holding it per page - Yes, it’s awkward that each instance of newtab has one of these, but that’s already a problem with the state object - The Search.jsx component will get the definition for the lone “search” item in the registry. When it finds it, it will insert the script and insert a web component (React, Lit, vanilla) that is responsible for adding stylesheet dependencies within its shadow dom or via CSS modules.
- I need to start an Engineering Design document to start shopping around with these ideas, at least show Gijs and Scott, but probably a bunch of others as well.
- File a bug to get rid of non-handoff behaviour to simplify Search.jsx and contentSearchUI
- This will then create a structure which can be send down in the state object that each New Tab receives - A class (or component?) will be responsible for parsing that structure and holding it per page - Yes, it’s awkward that each instance of newtab has one of these, but that’s already a problem with the state object - The Search.jsx component will get the definition for the lone “search” item in the registry. When it finds it, it will insert the script and insert a web component (React, Lit, vanilla) that is responsible for adding stylesheet dependencies within its shadow dom or via CSS modules.
-
Reviewing?
- Rate this episode
Chat
Resource Links
- Felicia Bacon
- npb hacks on the SpiderMonkey JS engine
- Compiler Compiler live stream
- Try out Mozilla VPN
- Firefox Source Code Documentation
- mozconfigwrapper: A Wrapper to manage different mozconfigs
- MyQOnly: Mike's addon to track your review queues on Mozilla Phabricator
- Mike's Firefox Color Theme
- Firefox Monitor: Stay informed on new data breach of services you're using online
- Create a free Bugzilla account
- Codetribute: Find a good first bug to start contributing on a Mozilla projects (Firefox, Thunderbird, and more). See the above link for creating a Bugzilla account first!
- Nathan Baggs livehacking / reverse engineering
- Alessandro Castellani has been streaming himself livehacking on Thunderbird
- Emilio hacking on Firefox