We — Manton Reece and Brent Simmons — have noticed that JSON has
become the developers’ choice for APIs, and that developers will
often go out of their way to avoid XML. JSON is simpler to read
and write, and it’s less prone to bugs.
So we developed JSON Feed, a format similar to RSS and Atom but in
JSON. It reflects the lessons learned from our years of work
reading and publishing feeds.
Sure, but the API also exposed at the top level feedparser.parse('http://example.com/feed.json').namespaces, and 100% of the behavior of feedparser is in FeedParserMixin which is 100% attached to XML. I mean I understand the spirit in which this comment of offered but it's hard to imagine feedparser changing enough that it would support json feeds without basically being rewritten, even the test framework of feedparser is all setup around XML
Might be interesting, and I'd be happy to generate this for my own stuff if it sees much usage, though one gets a little nervous about fragmentation. I feel like I remember an RSS-as-JSON spec a while back that didn't really get any uptake. On a related note, I'm still waiting for someone smarter than me to take the off-the-cuff thought of "sites could publish (and feedreaders consume) .git" and do something interesting with it.
It's hard to escape the media pronouncements that iPhones are now boring again after Samsung unveiled its latest Galaxy S8, Apple's Mac business is being overshadowed by more exciting Surface Windows PCs from Microsoft and that Apple Watch is a disappointing dud. But all of those media narratives are wrong, here's why.
The United States Senate continues the war against their own users. One Hackernews suspects some kind of massive federal conspiracy to censor comments on reddit.com. Another suddenly realizes that people might disagree about things for reasons other than ignorance, and becomes distressed. The rest of the comments are people arguing about technical methods to work around the user-tracking they implement in their day jobs.
Google continues the war against their own users. The XMPP Memorial Society trades barbs about whose fault it is that a misdesigned overengineered shitshow of a protocol failed to gain traction amongst non-erlang enthusiasts. Every single messaging platform in current existence is held up as Obviously The Future. Hackernews tries to figure out what Google's master plan is, and why Google is working so hard to make it look like aimless poorly-managed floundering. IRCv3 continues to be a retarded pile of solutions to the wrong problems.
The United States House of Representatives continues the war against their own users. Hackernews is outraged, presumably because the rules will now enable other companies to compete with Google in the lucrative Fuck Everybody's Privacy market sector. The entire comment thread is just Hackernews arguing about political shit and deciding which elected officials are betraying the American people. Not a single goddamn Hackernews makes the obvious connection to the shit they do at work all day for a living. The tacit consensus: Hackernews isn't bad for creating the tools of surveillance capitalism; Congress is bad for letting people use them.
Some academics figure out how to make shit in pictures look like shit in other pictures. One Hackernews notices that the machine learning papers have largely stopped relying on mathematics or any other scientific endeavor; the others are ready with reassurances that someone will get around to formal research sooner or later. All this stuff is super worthwhile in the meantime because we can just keep passing around training sets verbatim and treating them as infallible, just like we do with node.js libraries! Both the machine learning community and the web development community are completely free of charlatans! Scout's honor!
HELLO FRIENDS. I am announcing this everywhere because I'm very excited about
it. I released a new zine today! Read it here! Read all my zine things at jvns.ca/zines!
This zine is about some of my favorite Linux debugging tools, especially tools that I don't think are as well-known as they should be. It covers strace, opensnoop/eBPF, and dstat! netcat, netstat, tcpdump, wireshark, and ngrep! And there's a whole section on perf because perf is the best.
If you don't know what any of those tools I just mentioned are -- PERFECT. You
are who this zine is for!!! Read it and find out why I love them! Also, a lot
of these tools happen to work on OS X :)
I've been really delighted to see that a ton of people have enjoyed & learned
something new from this zine, whether they just started using Linux (!!!) or
have been debugging on Linux for 10 years.
As usual, there are 3 versions. If you print it, you can print as many as you
want! Give them to your friends! Teach them about tcpdump!
Due to popular demand, I am pleased to announce the launch of the Rails Tutorial screencats! The art of web development procrastination has never been cuter:
Although they’re not nearly as adorable as the screencats (for obvious reasons), I’m also pleased to announce the launch of the Ruby on Rails Tutorial screencasts, updated for Rails 5!
The Rails Tutorial screencasts are the most up-to-date resource for learning web development with Ruby on Rails. They are available for free via the Learn Enough Society, as well as being available for purchase as direct downloads. Those links include a 20% launch discount that expires in a week, so get them while it lasts!
The best way to get the new screencasts is via the Learn Enough Society, which includes all 15+ hours as integrated streaming video:
The Learn Enough Society also includes text and video for the three Developer Fundamentals tutorials (Command Line, Text Editor, Git), as well as immediate access to new tutorials as they’re released.
The Rails Tutorial screencasts are the ideal complement to the Rails Tutorial book, allowing you to see exactly how web applications are built in practice. There are video lessons corresponding to each chapter of the book, totaling over 15 hours of content. You can view a full sample lesson here.
As with the 3rd edition of the tutorial, the new 4th edition covers every major aspect of web development:
Creating both static and dynamic pages with Rails templates
Data modeling with a full database back-end
Creating a working signup page from scratch
Building a custom login and authentication system
Activating accounts and resetting passwords
Sending email in Rails, both locally and in production
Advanced data modeling to create a mini Twitter-like application
Coverage of software best practices, including test-driven development and version control
Emphasis on strong security throughout
Deploying to production early and often
Those familiar with the previous edition will find the following main differences in the Rails 5 version:
14 lessons instead of 12, due not to new material but to the two longest lessons being split in two (much more manageable)
Full compatibility with Rails 5, including the use of the rails command in place of rake
A shift toward integration-style testing for controllers, together with a new convention for passing parameters in tests
Because Rails 4.2 and Rails 5.0 are so similar, the new edition of the screencasts did not need to be created from scratch. Instead, the minor diffs mentioned above are highlighted as text notes in the videos themselves. The result is that it is immediately apparent which parts of the Rails framework have changed between versions.
Remember, all 15+ hours of the Rails Tutorial screencasts are available both via the Learn Enough Society and as direct purchase. Those links include a 20% launch discount* that expires in a week, so get them while it lasts!
* Note: The discount applies to any Rails Tutorial purchase or to the first month of the Learn Enough Society.