This looks really cool! Are there any plans to open source the code so that people (I) can run it locally? I doubt the service wants me adding my 1000+ feeds.
I haven't released the source code, it's an ungodly mix of Perl and Haskell, but maybe I will try and clean it up a little and make it public - it's quite hardcoded, eg the domain, currently.
How do you keep up with 1000+ feeds now? I feel using Gnus is quite efficient, but of the 664 feeds on Feedbase currently I'm subscribed to "only" 304.
I'm curious about how the project evolved, since "an ungodly mix of Perl and Haskell" is not a phrase one might expect to see very often. If not open-sourced, can you say more about which pieces are where?
In any case, I quite like the idea. I self-host TT-RSS, and flip back and forth between using the native web UI, and Elfeed with the TT-RSS API as a back-end.
I had been wanting to learn Haskell for a long time, and came to the conclusion that without an actual project, I would never get anywhere.
However when I started working on Feedbase (2015) I wasn't really good enough at Haskell to accomplish the task before I giving up in frustration, so I fell back to my usual preferred scripting language, Perl, for the part that fetches the feeds and updates the database and for the nntp server - I only implemented the (quite simple) website in Haskell.
I've gotten a little better since then, so my little blog-engine, Lantern, which also uses nntp as the main "interface", is written completely in (beginner) Haskell.
Thanks for the update! I find myself in a similar boat for small projects.
One approach that I have found useful is to make sure that any stored data is in a form that is approachable from a portable standpoint, not coupled tightly to any one language. It's also helped if I want to try re-implementing some existing solution, because I have that known data as a starting point for comparing what I know with what I want to learn.
Well, perl is my jam and I was a big haskell user back in university. But I understand how personal stuff can get quite hardcoded. Just knowing you're thinking about it is cool.
I don't neccessarily, or ever, read all of my feeds. I just quickly scan through the headlines in categories I'm currently feeling interested in. And most feeds don't update weekly or even monthly. It's kind of like how you can 'read' reddit, or HN, but not actually visit every site or read every comment thread. Except I don't depend on third parties for aggregation or moderation.
Nice! I've been wanting something like this in combination with hnrss.org to have my feed/news reader (Emacs?) allow me to revisit a HN thread and show unread comments (since I last looked at the comments). I think there's still some glue to be done too make that all work nicely.
Some blogs have a regular feed and a comment feed - I haven't really tried to use the latter, as comments in many places unfortunately is even more spammy than usenet.
It would be fun if you could post comments to the original blogs through Feedbase, alas, that would probably means custom integration for each and every one, and would be prone to abuse. So comments (follow ups) on Feedbase are local to Feedbase (and not used much).
I like Gnus so much that I try to put an nntp interface on most things, but the servers that come out of that are specialized, like Feedbase or my little blog-engine, Lantern.
Was confused by "news reader" since thought that's what RSS readers are, but then realized it's about Usenet. Never used that for news, only... stuff...
>Skip uninteresting items with newsreader scoring/killfiles
In my opinion, there is one step missing to bring nntp back to mainstream: votes. If there were a standard for votes, or feedbase.org would implement a standard, or clients would agree on using the ActivityPub protocol to keep track of votes, then reading nntp wouldn't be work like clearing a mail inbox but addictive like a Facebook feed.
There used to be a distributed, cooperative spam-marking system in usenet - the name escapes me right now - but I'm not sure it ever worked really well.
26 comments
[ 3.2 ms ] story [ 67.8 ms ] threadYou can use more old school clients such as slrn, tin, and nn - but usually displaying HTML takes some work.
Pan is another GUI client, I don't remember if it renders HTML.
How do you keep up with 1000+ feeds now? I feel using Gnus is quite efficient, but of the 664 feeds on Feedbase currently I'm subscribed to "only" 304.
In any case, I quite like the idea. I self-host TT-RSS, and flip back and forth between using the native web UI, and Elfeed with the TT-RSS API as a back-end.
However when I started working on Feedbase (2015) I wasn't really good enough at Haskell to accomplish the task before I giving up in frustration, so I fell back to my usual preferred scripting language, Perl, for the part that fetches the feeds and updates the database and for the nntp server - I only implemented the (quite simple) website in Haskell.
I've gotten a little better since then, so my little blog-engine, Lantern, which also uses nntp as the main "interface", is written completely in (beginner) Haskell.
One approach that I have found useful is to make sure that any stored data is in a form that is approachable from a portable standpoint, not coupled tightly to any one language. It's also helped if I want to try re-implementing some existing solution, because I have that known data as a starting point for comparing what I know with what I want to learn.
I don't neccessarily, or ever, read all of my feeds. I just quickly scan through the headlines in categories I'm currently feeling interested in. And most feeds don't update weekly or even monthly. It's kind of like how you can 'read' reddit, or HN, but not actually visit every site or read every comment thread. Except I don't depend on third parties for aggregation or moderation.
It would be fun if you could post comments to the original blogs through Feedbase, alas, that would probably means custom integration for each and every one, and would be prone to abuse. So comments (follow ups) on Feedbase are local to Feedbase (and not used much).
I don't know if you would define it as "modern".
I like Gnus so much that I try to put an nntp interface on most things, but the servers that come out of that are specialized, like Feedbase or my little blog-engine, Lantern.
In my opinion, there is one step missing to bring nntp back to mainstream: votes. If there were a standard for votes, or feedbase.org would implement a standard, or clients would agree on using the ActivityPub protocol to keep track of votes, then reading nntp wouldn't be work like clearing a mail inbox but addictive like a Facebook feed.