I recently switched my RSS/Atom feed reader away from Tiny Tiny RSS and its companion Android app. I had fallen behind on upgrades on the server, and the app had been removed from F-Droid, so I needed to do something. I had set up Tiny Tiny RSS in 2017, so it was time for a fresh start.
I wanted a similar setup: a server that would maintain track of my subscriptions, articles and (un)read status, and then an Android app with a good UI and offline support. I do most of my feed reading on my phone these days, so I'm a bit more picky on that side.
Rachel (by the bay) recently praised the FreshRSS developers, so that was the first thing I tried out. And it worked...fine? I set up a container with podman, hooked it up to a MariaDB database (also in a container), and imported all my feeds.
Mostly for due dilligence, I tried out Miniflux next. It only supports postgres, which knocked it down a peg for me since I prefer MariaDB. I was really impressed with its feature to download content from the link instead of what's in the feed to "fix" feeds that truncate content.
Then I learned that FreshRSS has the same feature, just a little less automatic! FreshRSS requires you to manually enter a CSS selector, which so far has just been pretty trivial since it's either article or main. It also supports manual feed construction using XPath, which seems really cool, but I haven't gotten that far yet.
The main limitation I encountered with Readrops was not having a preview or snippet when viewing all the articles (example). I also found the "estimated reading time" indicators kind of patronizing with no way to disable them.
As far as I could tell, Flux News didn't have any support for offline reading.
I like Capy Reader, it's a clean UI that's similar enough to what I had with Tiny Tiny RSS. FreshRSS worked right away, but I couldn't get it to connect to Miniflux over the Google Reader API. I'm pretty sure it's a me problem, since others are using it with Miniflux just fine.
In any case, I tried out FreshRSS+Capy Reader for a week, and while I still need a bit more time to get used to it, I think it will meet my feed reading needs. So yesterday, after taking a backup, I shutdown my Tiny Tiny RSS server after more than 8 years in service 🫡. Here's to 8 more years of feed reading with FreshRSS and Capy Reader!
As part of Boba Quest 2025 🧋, I'm trying and reviewing a new boba shop each week.
Flushing is one of my favorite neighborhoods in Queens because of how diverse and lively it is. Originally founded by the Dutch in the 1600s, today it's an incredible melting pot, primarily populated by Asian immigrants.
My coworker David Huerta originally recommended Xing Fu Tang to me at least two years ago. I went last year, loved it, and knew I had to review it for Boba Quest.
Even before you walk in, you get the sense you're in for a treat. They keep a giant pot of fresh boba on display and you can see employees scooping it up into cups.
Their speciality is brown sugar boba milk (not tea). So that's exactly what my girlfriend, Kajol, and I got — with light ice and regular sugar.
Probably the third different "No. 1" chain from Taiwan that I've tried.
My review:
Boba: 4/4 the boba looked fresh and it tasted fresh too. It was as close to perfect as I think you can get.
Tea: 4/4 it was everything I wanted out of a brown sugar boba milk. Excellent sweetness and they make it special by caramelizing a little bit of the brown sugar so you even have a bit of crunch texture.
Bonus: 1/1 it's a pretty small store, there's barely enough room for the line. But that ends up making the atmosphere even nicer, because you can tell everyone else is really there for the boba. Also if it's a nice day, you can sit across the street on the steps of the library.
Boba: 3/4 - It had a good flavor; it started out with a "melt in your mouth consistency" and then eventually got slightly chewier. I liked it better once it was chewier.
Tea: 2/4 - it was just milk so there was no tea flavor, but I would've liked the tea flavor! The caramelized bit on the top was wonderful, as it added some great texture.
Bonus: 1/1 - the ambiance was nice, the menu was clear, and there was a small area in the back where people can pick up a stick and then pick out a fortune based on the number on the stick.
Total: 6/10. Kajol said she would definitely want to go back to try the boba milk tea version of the drink.
Interestingly this is the first time Kajol and I have had our reviews diverge by more than a point. Of course, I think I'm right, but we'll have the opportunity to try it again in the near future.
For our next visit I think I'll be obligated to try Xing Fu Tang's more extravagant option: the gold foil boba milk. Yes, they really put gold leaf on top, and certainly charge you for it!
Shoutout to David for originally recommending Xing Fu Tang to me; if anyone else has tips, please let me know.
SecureDrop team member Kunal Mehta demo’d the SecureDrop Workstation at RightsCon 2025 on Feb. 25. You can watch a recording of the talk; the slides (PDF) are also available.
In the demo, Mehta explained the multiyear project to provide better and more streamlined tools for journalists to safely handle potentially malicious submissions. Built on top of Qubes OS, a security-focused operating system, SecureDrop Workstation leverages security-through-isolation to keep untrusted files isolated in virtual machines where journalists can review them without risking their safety and security.
Mehta also described a pilot phase in which the SecureDrop team collected and addressed feedback from journalists, before making SecureDrop Workstation generally available in 2024. It is now rolling out to all news organizations that use SecureDrop. In the session, he showcased the SecureDrop Workstation, while contrasting it with the classic SecureDrop workflow, to highlight its new usability and security features.
If you’re a journalist interested in using the SecureDrop Workstation or want to learn more, please contact us — we would be happy to help. Thanks to Access Now for hosting RightsCon and giving us the opportunity to present.
As part of Boba Quest 2025 🧋, I'm trying and reviewing a new boba shop each week.
How far are you willing to go for good boba? My rule of thumb is generally that if it's accessible via public transit, I'll make the trek.
Yesterday I took Amtrak down to Philadelphia, and a combination of the SEPTA subway and trolleys to get boba at Tea-Do, in the University City neighborhood — only around 100 miles from my apartment in Queens. Located right next to the UPenn campus, it's reachable using the T2, T3, T4, T5 trolleys from the 36th–Sansom station or the L1 at 40th Street station.
I was looking forward to the "contemporary tea house".
I'm kidding a little bit, I didn't go down to Philly solely to have boba. But since Tea-Do doesn't have any locations in New York, I thought it would be a good place to try out.
I ordered my usual, a jasmine milk tea with boba, less ice. As an aside: that was the moment when I learned Philly has a "sweetened beverage" tax.
Your standard jasmine milk tea with boba, less ice. And an elbow in the background.
Boba: 3/4 the boba was definitely above average, just slightly too chewy (possibly overcooked?) to be perfect.
Tea: 4/4 the tea was excellent, good flavor and good amount of sweetness (paying the tax was worth it!).
Bonus: 1/1 Tea-Do has a nice location, with long tables that make it convenient to sit down and hang out.
Total: 8/10. I would definitely come here again, it's a convenient location with good drinks.
This was my first time ever in Philly, and the boba didn't disappoint. I enjoyed the immense variety of public transit and just how convienient it was to get around. I definitely plan on visiting again soon.
Back in April 2023, I was in Brooklyn to attend a live taping of the FiveThirtyEight Politics Podcast. It was great, someone corrected Nate Silver on the U.S. constitution and former FiveThirtyEight writer Clare Malone returned for a lot of great banter backed by data.
The very next week, Disney laid off two-thirds of the staff, including Silver, and merged the website into the primary ABC News one. I have yet to ever attend another in-person podcast taping.
Nate Silver (left) and Galen Druke (right) discuss whether the New York Post headline was a "good use of polling" or not at a live podcast taping in April 2023. (It was bad.)
FiveThirtyEight really popularized and built up modern data journalism. The 2020 presidential election forecast was a fantastic case study of how to represent both probabilities and uncertainty. Its pollster ratings actually graded pollsters, and helped drive discussion on changing methodologies.
This is not to mention all of the fantastic work they did on sports predictions and statistics.
I'm particularly nostalgic about FiveThirtyEight since its peak was around the same time I started studying journalism. The "Significant Digits" newsletter was my mandatory morning reading and of course, I'd share the new statistics I just learned with whoever would listen.
Likewise, I'd listen to the weekly politics podcast within an hour or two of it being released to hear the latest "good use of polling or not" debate and polling details about the next election. Even though I stopped listening two years ago, it's still my #1 most listened to podcast at 354 hours (#2 is at 90 hours). My favorite series, "The Gerrymandering Project", was a deep, data-driven analysis of redistricting and gerrymandering. It's still worth a listen today.
Since being let go, Silver has criticized Disney for failing to monetize the site properly, which I agree with. There was no way to financially support them aside from buying merch, which I assume brought in basically nothing.
FiveThirtyEight was originally just Silver's blog, before it was bought by The New York Times (and then later sold to Disney/ESPN). It was just one of the many free web blogs the Times had, but if it were today, I could easily see them selling it as a separate subscription, like The Athletic or Wirecutter. Just ahead of its time unfortunately.
Some of the work FiveThirtyEight did has since been taken over by others, like The Upshot, Split Ticket, Harry Enten's Margins of Error, etc., but I have yet to find a similar data-focused media outlet to replace the void.
Now that I have two months of reviews under my belt, I wanted to create a proper website to make it easy to discover them.
Taking inspiration from Everything is Everything, I wanted a similar layout with a big map of all the reviews and then explanations of the project itself.
I figured all of this could be built as a static site and a bit of JavaScript. I've had my eye on Zola for a while, since it's written in Rust and by the same author as Tera, my preferred templating library.
It ended up being a great fit, largely because it has builtin support for loading content from a TOML file via a load_data function. I store all of the review metadata in a single file, which drives the entire site.
My main criticism is that Zola doesn't support so-called "ugly URLs" like foo.html (it only supports foo/). I guess I'm weird in that, as this blog demonstrates, I like URLs that end with .html.
I started by asking Claude to generate such a website for me (chat transcript), and kept slowly refining with prompts. While normally Claude is really good at letting you preview HTML websites, its own CSP prevented any of the map tiles from loading so I had to copy it locally and preview it that way.
I'm reasonably proficient at HTML/CSS/JS, but this was a much more fun way to design a website. Instead of having to focus on syntax, I could just tell it to shape it how I wanted, plus making my own manual improvements here and there.
People, often rightfully so, criticize AI-generated content as lacking creativity, but I feel that using Claude allowed me to design the website exactly how I, a human, wanted it to look like, and not get bogged down fighting syntax things.
I eventually hit Claude's output limit (since it kept repeating the whole HTML file) and had to start branching off into individual chats, like adding an escape key handler and learning about vh units in CSS (big TIL for me). I hit the limit again when I added the "Recent reviews" sidebar.
That was roughly the point I checked it into Git, so you can browse how the initial version looked like. It's come quite a long way since then.
