Post B1HBJ1ps2lbLfWSWum by Kolomona@podcastindex.social
(DIR) More posts by Kolomona@podcastindex.social
(DIR) Post #B137Lfrx3Dz0Nd027s by dave@podcastindex.social
2025-12-08T20:11:35Z
0 likes, 0 repeats
#api #feedparser Made a big structural change to the new feedparser to split all the tag handlers into their own modules to make things easier to manage going forward:https://github.com/Podcastindex-org/feedparserThis change will probably cause a lot of conflicts if you are working on the pre-changed code already. You'll want to start working on this base code instead.
(DIR) Post #B137kNQckaPYQVzqE4 by dave@podcastindex.social
2025-12-08T20:16:03Z
0 likes, 0 repeats
There is a tags folder in src. To support a new tag, create a module in that folder.Will keep refining the structure to make it as maintainable as we go.
(DIR) Post #B137lD1EwD0IQqWIsa by dave@podcastindex.social
2025-12-08T20:16:13Z
0 likes, 0 repeats
Also added a few initial unit tests.
(DIR) Post #B138RRFGR3n8iTDTCy by merryoscar@podcastindex.social
2025-12-08T20:23:49Z
0 likes, 0 repeats
@dave would love to see a short video walkthrough of how you are working with the model in your IDE!
(DIR) Post #B13AXpkeR0GmMo4gTI by ericpp@podcastindex.social
2025-12-08T20:47:23Z
0 likes, 0 repeats
@dave I have a bunch of stuff I worked on over the weekend to contribute after work
(DIR) Post #B13IafNIGqSigb0jwm by suorcd@podcastindex.social
2025-12-08T22:17:30Z
0 likes, 0 repeats
@dave let me know if https://github.com/Podcastindex-org/feedparser/pull/2 does not make sense. I see keeping the numerous files separate unneeded.
(DIR) Post #B13KRuKrTex0O7pBdA by dave@podcastindex.social
2025-12-08T22:38:23Z
0 likes, 0 repeats
@ericpp I hope this change doesn't wreck what you worked on too much. I think when you see the new layout it'll make things a lot easier.
(DIR) Post #B13KVQqd3NtS5mMdPc by dave@podcastindex.social
2025-12-08T22:39:01Z
0 likes, 0 repeats
@merryoscar Sure, I'm glad to record some stuff.
(DIR) Post #B13KaahppQzu7Ai88G by ericpp@podcastindex.social
2025-12-08T22:39:57Z
0 likes, 0 repeats
@dave I can adapt it. I was thinking of doing something similar, but with separate structs and separate state for each tag.
(DIR) Post #B13KpLXsTfi3RG5b96 by dave@podcastindex.social
2025-12-08T22:42:36Z
0 likes, 0 repeats
@ericpp I like the dispatch model. The only thing that annoys me a bit is having to go back to mod.rs to put in the extraction details. I feel like that could be genericized some. Thoughts?
(DIR) Post #B13LeNKpaFGbl2Jr5E by suorcd@podcastindex.social
2025-12-08T22:51:51Z
0 likes, 0 repeats
@dave did you want the discussions here, or in the project?
(DIR) Post #B13NZ7eCgL8PZ20HNg by dave@podcastindex.social
2025-12-08T23:13:17Z
0 likes, 0 repeats
@suorcd Here is faster for now. I find the GitHub discussion system to be tedious.
(DIR) Post #B13NpY8twQZGN2CAYy by suorcd@podcastindex.social
2025-12-08T23:16:16Z
0 likes, 0 repeats
@dave ok. Getting a build warning https://github.com/Podcastindex-org/feedparser/issues/4 not sure if it matters
(DIR) Post #B1455xJARLg35B4BXc by ericpp@podcastindex.social
2025-12-09T07:21:03Z
0 likes, 0 repeats
@dave Here's what I got so far: https://github.com/ericpp/feedparser/commits/main/It's supposed to implement everything that partytime does but I need to clean it up.
(DIR) Post #B145JISJ7rNSR0hf6W by ericpp@podcastindex.social
2025-12-09T07:23:28Z
0 likes, 0 repeats
@dave Traits could work to enforce the same function signatures. Are you think sort of like a plugin system where each tag would register itself with a central object?
(DIR) Post #B14qfYi64Fk942CpJQ by dave@podcastindex.social
2025-12-09T16:14:03Z
0 likes, 0 repeats
@ericpp Wow I will hold off on any more changes until you say it’s ready for merging. 😎
(DIR) Post #B14u1dr1vFRBFebcqO by dave@podcastindex.social
2025-12-09T16:51:43Z
0 likes, 0 repeats
@ericpp Did you see that I modified the .junie/guidelines.md and now the agent is writing the triggering prompt to the .llm_history/ log files? https://github.com/Podcastindex-org/feedparser/blob/tag-add-language/.llm_history/20251209_1.mdNot sure which IDE you are using, but I think it'd be good to have all the input prompts recorded as a teaching tool on how the code came about.cc: @merryoscar
(DIR) Post #B14u8OKa2JRX8Le7w8 by dave@podcastindex.social
2025-12-09T16:52:54Z
0 likes, 0 repeats
@ericpp @merryoscar Interesting that GH is making the tags referenced in the prompt invisible because of the angle brackets. Will need to make a change for that.
(DIR) Post #B14zE78LO7mHIzTG3U by ericpp@podcastindex.social
2025-12-09T17:49:58Z
0 likes, 0 repeats
@dave I still need to check out Junie. I've been using Cursor with Claude to do a lot of this stuff. Now I need to go back and clean it up. @merryoscar
(DIR) Post #B150AVbwXWOY15A0m0 by ericpp@podcastindex.social
2025-12-09T17:56:43Z
0 likes, 0 repeats
@dave The Junie guidelines doc seems pretty similar to the CLAUDE.md doc that Claude uses. I wonder if I can just copy it. I bet Cursor has the ability to export the prompts. My prompts are extremely basic and I tend to iterate over them during a chat session. I open a new session for each new feature or change I want to do.@merryoscar
(DIR) Post #B150AWrDuCDBslJjEG by dave@podcastindex.social
2025-12-09T18:00:30Z
0 likes, 0 repeats
@ericpp @merryoscar We can copy the guidelines.md over to agents.md which seems to be the cross-platform way to do it.h/t: @alecksgates
(DIR) Post #B15NJQQfTNp1Ut6i8W by Kolomona@podcastindex.social
2025-12-09T22:19:53Z
0 likes, 0 repeats
@dave @ericpp This may or may not be helpful for you but it's a very top level view of how I made v4v music without writing one single line of code myself.https://cdn.kolomona.com/tmp/vibecoding.html
(DIR) Post #B1GwgXoBpw0oB9QbVA by ericpp@podcastindex.social
2025-12-15T03:05:58Z
0 likes, 0 repeats
@dave Ok. I have it in a state that's good enough for me: https://github.com/Podcastindex-org/feedparser/pull/8Plus I'm tired of looking at it. I used Claude to build the previous version with tests. Then I thought it would be a fun challenge to delete the code that Claude added and have GPT 5.1 rebuild it from the tests, which worked out better than I thought. Then I went through manually and cleaned up some of the spaghetti code and bad assumptions it made.
(DIR) Post #B1GwgYy9WNZjmL64fY by dave@podcastindex.social
2025-12-15T12:18:01Z
0 likes, 0 repeats
@ericpp Awesome!!! 👀👀
(DIR) Post #B1H0OBag6TArsKbrAe by dave@podcastindex.social
2025-12-15T12:59:34Z
0 likes, 0 repeats
@ericpp Merged. Will build in a bit when I get to a computer.
(DIR) Post #B1HBJ1ps2lbLfWSWum by Kolomona@podcastindex.social
2025-12-15T15:01:52Z
0 likes, 0 repeats
@dave @ericpp ericpp++
(DIR) Post #B1J7cbSNomeg9CTHZA by dave@podcastindex.social
2025-12-16T13:30:01Z
0 likes, 0 repeats
@Kolomona @ericpp This is amazing Eric. I'm building the converter now based on your new input file format. I'm just reading files between processes for now (for simplicity) and will add the queueing/object storage later.I'll push it as soon as I get it stable.
(DIR) Post #B1JIjOcv18e7sWqCQa by ericpp@podcastindex.social
2025-12-16T15:34:30Z
0 likes, 0 repeats
@dave Excellent. I'm glad it's working well for you.I tried to keep it pretty close to what Partytime outputs except for database ids and separate tables for chapters, transcripts, etc (since it has no item ids).Some of the Partytime rules might need some cleaning up too. For example, it does string trims on newsfeed titles and links but not descriptions, but does do trims on item descriptions and itunes titles but not regular titles.@Kolomona