Mitch Downey :pci:

open source developer @ podverse.fm / fediverse / foss / javascript / node.js / podcasts / react / react native / typescript

⚡️:[email protected]

Public Key

npub1yfr5d05qj3qzz84kntze3c8accyz0unwczgm0fxuypuqdnenkwuqew6g6v

NIP-05 Address

[email protected]

Profile Code

nprofile1qqszy36xh6qfgsppr6mf43vcur7uvzp87fhvpydh5nwzq7qxeuem8wqpz4mhxue69uhhyetvv9ujuerfw36x7tnsw43q0qhfp5

Publishing to

relay.ditto.pub

Author Public Key

npub1yfr5d05qj3qzz84kntze3c8accyz0unwczgm0fxuypuqdnenkwuqew6g6v

Show more details

Last Notes

2026-02-06 18:04:05 UTC

- reply

@nprofile…twh5 hah I have not thought that far ahead yet...
But as long as the data can map to the same schema we end up with from parsing RSS, I imagine there's no reason we can't support AP or any other data format.
I do however lean towards RSS being a good enough data standard for podcasts/music/etc. What does AP offer in this context that RSS does not?

2026-02-06 17:50:27 UTC

This is a test thread for <podcast:socialInteract> tag testing.
(I need an activity pub post url for us with test RSS feeds.)

2026-01-29 22:53:53 UTC

- reply

@nprofile…sk9g be careful not to get hacked. Seems like there are a lot of attack vectors to moltbot right now. Don’t put anything on it that would cause too much grief for you if it gets hacked.

2026-01-29 20:23:48 UTC

Podverse Alpha update 🛠️
We are nearly ready to implement V4V in Podverse Alpha web...and I am completely out-of-date in terms of what the current practices are.
Can anyone direct me to resources I should follow for a web-based Bitcoin V4V implementation? Is the standard still keysend? Are there multiple types of implementations we should support by now?
#podverse #pv2

2026-01-26 16:30:37 UTC

- reply

@nprofile…x5pc ok, I think I get it now. The Mostr bridge is silently mapping that npub it assigned to me to [email protected] in the API requests it makes with the fediverse. I imagine if we could see the Mostr API requests we would see our original handles in there along with the npub it makes visible in the message text.

2026-01-26 16:07:12 UTC

- reply

@nprofile…x5pc ah thanks for the info.
Do you know if that is a setting that the podcastindex social instance has enabled? Or does the mastodon software issue npub keys to every account created? I'm trying to make sense of how I'm being contacted in the fediverse through an address I didn't create (or don't remember creating).

2026-01-26 15:31:14 UTC

- reply

@nprofile…x5pc looking at hellthread website, I don’t think my Fediverse account even has a nostr npub id? Everyone else on the thread does though. Don’t understand how or where I’m tagged on this stuff.

2026-01-26 15:28:44 UTC

- reply

@nprofile…x5pc what’s weird is I don’t even see myself tagged in the threads I’m getting notified for. They haven’t happened again since last night so ig I’ll just ignore it for now.

2026-01-26 02:18:25 UTC

I'm getting notifications containing weird spam that makes no sense on here. Anyone else having that? How do you stop it?

2026-01-23 21:30:47 UTC

- reply

but I'm also thinking it may be easier for open source contributors to create PRs for Podverse if they can just create 1 PR per feature. Right now, a typical feature update will require changes to multiple repos. It's a clean process imo, but it does add significant overhead to PR creation and reviews.
Anyone have experience and opinions on monorepos? Worth the switch in our case, or could it create more problems than it is worth?

2026-01-23 21:28:29 UTC

As the Podverse infrastructure has grown to 10+ repos, and given how LLM seems to perform better with all its context in one place, and it would be easier to save llm prompt history in one place...I'm wondering if we should make a major switch to a monorepo 😣
I've never done a monorepo before. If it were just LLM, then we may not need it (we could create yet another repo...podverse-llm or something) (cont)

2026-01-23 05:02:04 UTC

- reply

@nprofile…td70 (restoring state and scroll position on a React website on back navigation). Opus knocked it out in ~10 minutes, with a little debugging.
I have all these devops convenience scripts now too that always seemed too tedious to write before, but are huge time savers and make my workflow less error prone now. The whole day cost me ~$25 in Anthropic credits. Not cheap for a FOSS project that is not yet profitable lol, but well worth it. Truly amazed with the state of LLM software development.

2026-01-23 04:58:26 UTC

- reply

@nprofile…td70 I still haven't tried Claude Code. Curious to hear how it goes.
My workflow has been Cursor using Auto mode ~95% of the time, and then for big or difficult tasks, I switch it to Opus 4.5, and in the rare case that doesn't work, I switch it to
I think I literally got 2 weeks of tedious refactoring, validation scripts, and documentation done today over 8 hours. Plus a feature on the website that I've been wondering how to do cleanly for like a decade (cont)

2026-01-22 18:43:29 UTC

- reply

@nprofile…cyex awesome. makes sense to me. thanks!

2026-01-21 07:12:53 UTC

- reply

@nprofile…cyex API is for the main http data router backend.
We also have workers that make requests on the backend from outside the API router app. Should we distinguish those from the router API?
Do you think specifying environments is not worthwhile?
Could you explain a bit more what you mean by distinct?
Haven’t started the mobile app/s yet but I’ll add at least as much info as the Spotify example.
Sorry, barrage of questions 😅 you’re definitely an expert on this.

2026-01-20 19:28:43 UTC

I have been thinking of using a user agent pattern like this for PV nextgen:
Podverse/Local/2/API
Podverse/Local/2/iOS
Podverse/Local/2/Android
Podverse/Alpha/2/API
Podverse/Alpha/2/iOS
Podverse/Alpha/2/Android
Podverse/Prod/2/API
Podverse/Prod/2/iOS
Podverse/Prod/2/Android
1st param = owner
2nd param = environment
3rd param = major version
4th param = device
Does that order look good, or does anyone recommend a better hierarchy?

2026-01-18 05:11:40 UTC

- reply

@nprofile…tsmm yeah that sounds right. The exploit in that case is limited to only that specific browser though, since the server has to grant the httponly status to that device. Without httponly, the JWT value could be extracted and set in a different client/device and used freely from there.

2026-01-17 16:14:59 UTC

- reply

@nprofile…tsmm no, a web browser can't set an httponly cookie. What the LLM did was modify the existing login endpoint in our router API so that it would return the JWT in a response body, instead of setting it as a cookie.
The browser then received the response body, and set it as a non httponly cookie in the browser.
And yes, httponly is needed on cookies to prevent XSS attack vectors. When httponly is true, browsers prevent the JavaScript runtime environment from being able to read it.

2026-01-17 13:32:11 UTC

- reply

This is an egregious and easily avoidable security vulnerability. JWTs should not be set client-side in a browser or they can be scraped by 3rd parties.
The actual solution was to simply add ‘withCredentials: true’ in the request helper’s options. If we were flying in blind without knowing that, it would have appeared like the LLM gave us a valid solution that worked, but in reality it was a critical mistake.
I think the models will probably get smart enough to avoid this particular error tho

2026-01-17 13:27:23 UTC

I was showing my brother how to do some agentic development, and it went great, except for one notable problem.
We were adding a Login page to a website. The auth endpoints were already written. It generated the form, and made the request to the endpoint, and logged-in…
But the HttpOnly cookie wasn’t set in the browser. So we asked LLM for help debugging.
The solution it wrote was to return the JWT as a token in the browser, then set it in a cookie on the client-side ☠️

2026-01-17 13:23:49 UTC

- reply

@nprofile…x5pc I did find an example of a guy who had a workflow setup where it is basically a issue tracker board, and they define the ticket requirements, then when they move it to “In Progress” it automatically kicks off an agent to build the feature. In that case many tickets could be defined first then you could have multiple agents working on in progress tickets (although probably not on the same code base at the same time or get possible collisions).
It seems kind of risky and messy to me.

2026-01-17 12:20:27 UTC

- reply

Here’s a silly example but this person made a gamified visualization program for their agent management. I get the game UI is supposed to be a representation of processes that can be done with traditional text UI, but still, does anyone know wtf is going on? lol
Are people doing useful work with all these agents operating at once?
https://x.com/nearcyan/status/2011897629987520526

2026-01-17 12:11:25 UTC

Guys…I have no idea what people mean when they talk about multi-agent LLM development workflows.
I’ve been using Cursor with “Auto” mode, basically prompting one feature at a time, and I’ve been thrilled with the results.
But I keep seeing ppl talk about multi-agent and I can’t picture what they’re working on. The LLM takes like 10-30 seconds to complete my prompts. Where is there even time to type more prompts before it completes? I can’t tell how much is larp and how much is my own ignorance

2026-01-14 16:45:02 UTC

- reply

@nprofile…05ue it is! Web only though. I would say it’s about 80% finished.
I’m hoping that leaning into LLMs will help me finish the mobile app faster than I expected…we can’t fully release nextgen PV until mobile is finished since we need a major data migration. Trying to make the website and infrastructure a standalone open source project before then tho.
https://alpha.podverse.fm

2026-01-14 14:43:53 UTC

- reply

@nprofile…05ue the new Podverse we are working on has alternate enclosure support built-in from the ground up. We will try to provide the best alternate enclosure functionality we can for your feeds.

2026-01-14 09:24:27 UTC

- reply

I still don’t understand all the common terminology with LLM development I’m seeing thrown around though. Problem could be my own lack of knowledge.

2026-01-14 09:17:50 UTC

I’m finding trying to keep up with LLM advances for coding pretty disorienting…on one hand there are some real and magnificent breakthroughs, and on the other there is a lot of useless slop masquerading as breakthroughs.
Improved LLM models are already accelerating my productivity by several times, but I’m still seeing enough errors and anti-patterns, that I have a hard time believing the “you can do it all with prompts” and “you don’t need to review every line” camp isn’t being naive.

2026-01-12 05:48:23 UTC

LLM experiment update...
GLM 4.7 leaves a LOT to be desired, compared to Opus 4.5.
GLM 4.7 seems to make errors and be completely unable to see them even when I point them out sometimes. Opus 4.5...just seems to work every time.
Hopefully open source LLM models for programming catch up to the Opus 4.5 level of doing things sooner than later.

2026-01-12 05:45:53 UTC

- reply

@nprofile…k8p7 awesome. Thank you!

2026-01-10 23:53:22 UTC

- reply

@nprofile…peg3 do you use a full CLI approach? I haven’t made that leap yet, and with how I’m using LLMs at least, it doesn’t seem like they’ve eliminated the need for manual edits in all cases.

2026-01-10 06:29:02 UTC

- reply

@nprofile…nh7l and yeah cost is a big concern. You can set budget limits with the LLM APIs though to prevent surprise overspending. Planning the prompts and trying to write the most efficient way to get the work done is important though when cost is a constraint.
So far GLM 4.7 seems pretty good. Not Opus level but reasonably close, and 10% the cost is a huge selling point. Only been messing with it for a couple hours but I'm happy with it so far.

2026-01-10 06:23:53 UTC

- reply

@nprofile…nh7l yeah, and it's important to be cautious. I still reviewed all the code, and there were some issues with it, and some choices it made that were not my preference...but even after the manual changes I needed to make, it saved me so much time, it's crazy.

2026-01-10 04:41:10 UTC

- reply

The cost of using Opus 4.5 is substantial. A long 12 hour day of work cost me almost $50 in Anthropic API credits.
Since we haven't turned a profit yet with Podverse, that comes out of my pocket and is not easy to sustain, but the time and energy savings make it worth it.
I'm currently trying to replace Opus 4.5 in my setup with GLM 4.7, with is the highest rated open source / MIT LLM for coding available today. API credits are 10% the cost of Anthropic's, so hopefully it will be good enough.

2026-01-10 04:36:51 UTC

I don't know if there is anyone left who is skeptical...but the LLM updates over the past couple months for programming are a giant leap forward, and it seems like we've crossed an inflection point for the industry.
Using Opus 4.5, I was able to get what would have been 5 days of work for me finished in 1 day.
One stage of my work included writing 8 paragraphs of instructions for Opus, and it wrote hundreds of lines of code across 7 repos, and was ~99% accurate and finished after a few minutes

2026-01-10 04:28:49 UTC

- reply

@nprofile…k8p7 I don't see an artist RSS feed for Hartlight in Podcast Index:
https://podcastindex.org/search?q=hartlight&type=all
Our search directly queries PI API, so if a feed is not discoverable there, it won't be discoverable in PV.
I see that the album feed has a publisher tag that is pointing to the artist feed, which is great, but we have as a requirement that only feed that are in PI can be added to our database.
Does Fountain typically add artist feeds to PI? Or does PI not seek to include them?

2026-01-09 09:00:35 UTC

- reply

@nprofile…td70 @nprofile…7td6 @nprofile…0fv8 @nprofile…peg3 we don’t have an advanced log implementation at all right now. It’s just STDOUT to files from docker containers onto the host machine. I run all the log statements through Winston helpers tho with the intention of doing something more useful with them later. I don’t know what a good logging system looks like yet tbh

2026-01-09 06:02:46 UTC

- reply

@nprofile…w6kx ah I see. I’m a full stack JS dev but LLM is making it easier to jump into other languages.
Maybe you already have a solution, but if you deploy with docker containers, it’s easy to setup log rotation. Two parameters basically. Max size (ex. 50MB) and max file (ex. 3) and the system will save up to 3 50MB files, and when it tries to write a 4th, it’ll delete the oldest one.

2026-01-09 05:25:06 UTC

- reply

@nprofile…w6kx I've rebuilt the Podverse website going on 5 times now over almost a decade lol...so I know the feeling. Just have to keep learning from mistakes until it makes sense what went wrong.
Half the battle might be improving logging to be able to identify where these data mismatches are coming from?
If you ever decide to go open source, I'm happy to review some code or PRs for you and leave comments if I see anything.

2026-01-09 04:48:24 UTC

- reply

@nprofile…td70 @nprofile…7td6 @nprofile…0fv8 yikes 😬
I guess there must have been an error that caused it to abort the parses instantly and bombard you with requests.
Maybe I'll add I short timeout at the beginning of parse operations to make sure you don't get spammed that badly if something goes wrong.

2026-01-09 00:37:41 UTC

- reply

@nprofile…w6kx to start within Y minutes, and send a "Livestream starting soon" notification.
PV Alpha should render date and time in the user's local datetime. If it's not, then that's a bug.

2026-01-09 00:36:13 UTC

- reply

@nprofile…w6kx no need to disparage your work! I just checked out bouncepad.live and it's cool to see a livestream-centric UX. There are more active music livestreams than I realized.
All the feature updates I'm announcing are for Podverse Alpha, which is not live yet. The legacy PV does not support "Livestream Scheduled" notifications, but nextgen PV will.
A "live in 5 minutes" warning is an interesting idea...I could write a service for alpha that checks every X minutes if a livestream is scheduled

2026-01-09 00:31:29 UTC

- reply

@nprofile…td70 @nprofile…7td6 @nprofile…0fv8 the user agent should be:
Podverse/Alpha/2/Worker
Maybe there was a high volume of requests. I had "recursively parse remote items" enabled in our parser, which not only parses one feed, but also parses the remote items from that feed.
I could add some form of rate limiting to our parser to not barrage PI too quickly.

2026-01-08 22:12:29 UTC

- reply

The word "triggered" in the ntfy example is set by ntfy, not Podverse. Apparently we can replace that text. Not sure what to put there. I'm thinking of changing it to "Podverse" aka process.env.BRAND_NAME.

2026-01-08 22:11:08 UTC

Podverse Alpha update 🛠️
Big notification updates!
We completely replaced Google Firebase on the Alpha website with our own Web Push system.
This means the website has no dependency on a 3rd party service, like Google, for listeners to receive notifications via the website.
Also, we have added Unified Push support to the Alpha website. I have tested it using ntfy.sh and it appears to be working.
https://cdn.masto.host/podcastindexsocial/media_attachments/files/115/861/733/834/591/292/original/09fcad86d05e4f84.png
https://cdn.masto.host/podcastindexsocial/media_attachments/files/115/861/751/970/542/874/original/fc11a45f7f777335.png
https://cdn.masto.host/podcastindexsocial/media_attachments/files/115/861/759/006/483/906/original/783e952dd4483951.png

2026-01-08 21:51:09 UTC

- reply

@nprofile…td70 @nprofile…7td6 @nprofile…0fv8 after an hour, it appears that the Cloudflare restriction was lifted.
However, we could really use some way to ensure we don't get blocked by Cloudflare. I would think a valid PI API key should be enough, but I guess the Cloudflare protection must preclude API key validation. Any ideas would be appreciated.

2026-01-08 21:23:57 UTC

@nprofile…td70 @nprofile…7td6 how can we avoid Cloudflare breaking our connection to PI API?
I was able to make 3 successful requests to PI API from Alpha server, and I didn't change or redeploy anything...and since then, my 4 attempts to make a request to PI API have failed. The error message is 429, and it is an HTML respnse that includes:
Access denied | api.podcastindex.org used Cloudflare to restrict access
cc @nprofile…0fv8

2026-01-07 03:35:07 UTC

- reply

@nprofile…peg3 actually...I am just realizing that Unified Push is only compatible with Android. I thought maybe it would work with web also...
Not sure if there would be value in letting people change UP settings through the website. Seems like something that should be configured within the mobile app on each Android device.

2026-01-07 03:11:32 UTC

- reply

@nprofile…peg3 thanks! I'll see how far I can get, and may have some questions for you.

2026-01-07 02:33:40 UTC

- reply

Here's the LLM powered auto-translation code we're using for Podverse Alpha:
translation script:
https://github.com/podverse/podverse-web/blob/v5-develop/scripts/i18n/i18n-llm-translations.ts
compiling script:
https://github.com/podverse/podverse-web/blob/v5-develop/scripts/i18n/i18n-compile.ts
translation files:
https://github.com/podverse/podverse-web/tree/v5-develop/i18n

2026-01-07 02:28:37 UTC

Podverse Alpha Update 🛠️
Added multiple Language support to the website.
This is one of my favorite features in Nextgen so far...we write a root language file in English, then before deploying, we use LLM to automatically translate the file into other languages, so it's effortless to add support for as many languages as we want.
We also have an "overrides" file for each language, so if LLM gives us funky translations, we can replace them as needed.
https://alpha.podverse.fm
#podverse #pv2
https://cdn.masto.host/podcastindexsocial/media_attachments/files/115/851/436/288/875/362/original/4c0de008260603f5.mp4

2026-01-07 02:21:05 UTC

- reply

Next, I will try to add Unified Push notification support, so listeners can use their own open-source, self-hosted push notification systems.

2026-01-07 02:20:13 UTC

Podverse Alpha Update 🛠️
Push notifications have been added to Podverse Alpha web.
Listeners can configure 3 different types of notifications:
- New item (episode or music track)
- Livestream scheduled
- Livestream started
They can also set different settings for different feeds.
This was a lot of work to add. The code includes not only support for web, but an orchestrator to send notifications to also iOS and Android (when the mobile app is ready).
https://alpha.podverse.fm
#podverse #pv2
https://cdn.masto.host/podcastindexsocial/media_attachments/files/115/851/379/637/421/016/original/a908a9d29f10b179.png
https://cdn.masto.host/podcastindexsocial/media_attachments/files/115/851/381/592/141/112/original/78f15838e0804554.png
https://cdn.masto.host/podcastindexsocial/media_attachments/files/115/851/383/779/817/232/original/b4031c4fda251068.png

2026-01-03 20:03:51 UTC

- reply

@nprofile…7td6 @nprofile…8zna thanks! Sounds like it should be a simple fix. We just need to prioritize the artwork field over the image field from PI API?

2026-01-03 03:15:27 UTC

- reply

@nprofile…km5y @nprofile…8zna it looks like the code is already setup to prefer the itunes:image over the image. If you can provide an example feed with the issue in Podverse Alpha that could help with debugging.
https://github.com/podverse/podverse-parser/blob/v5-develop/src/lib/compat/partytime/channel.ts#L97-L107

2026-01-02 22:48:44 UTC

- reply

@nprofile…km5y @nprofile…8zna can you link me to podcast you identified the issue with?

2026-01-02 21:10:47 UTC

- reply

@nprofile…km5y @nprofile…8zna thanks! I’ll look into it.
We have a helper function that should use podcast:image, itunes:image, image in that priority order. I must need to fix that.

2025-12-30 21:40:12 UTC

- reply

@nprofile…5yxp looks good 👍 should work fine

2025-12-30 21:37:48 UTC

- reply

@nprofile…8nsk @nprofile…5yxp @nprofile…y5nd ok, as long as it has to be a UUID, then that won't require a lot of changes on our side.
right now our channel.podcast_guid column in postgres is a UUID data type. i have some extra validation before updates are saved to confirm the UUID is UUIDv5, but i can replace that with a check for any valid UUID type.

2025-12-30 21:04:18 UTC

- reply

@nprofile…8nsk @nprofile…5yxp @nprofile…y5nd hmm I designed our data schema to match the PI namespace docs, and it seems to specifically say UUIDv5 for <podcast:guid>. If other UUID types are valid for <podcast:guid> I need to make substantial changes.
https://github.com/Podcastindex-org/podcast-namespace/blob/main/docs/tags/guid.md

2025-12-30 01:17:02 UTC

- reply

@nprofile…5yxp I believe an <item> tag's <guid> tag can be any type of string potentially. So itemGuid in your example is correct.
But the feedGuid attribute is supposed to correspond with the <podcast:guid> tag, which is a UUIDv5 according to P2.0 spec.
https://github.com/Podcastindex-org/podcast-namespace/blob/main/docs/tags/guid.md
For a non-guid remoteItem feed identifier, the feedUrl attribute exists for that purpose:
https://github.com/Podcastindex-org/podcast-namespace/blob/main/docs/tags/remote-item.md

2025-12-29 15:29:23 UTC

- reply

@nprofile…5yxp added uuidv5 filtering and validation into podverse alpha. in local testing this will let the parser successfully parse the rest of your items.
it's not deployed yet, but should be tonight. i'm working on the Artists and possibly Lists pages later today.
https://github.com/podverse/podverse-helpers/commit/058036cdec259eaa2b58afb89c3010e8e4aab722
https://github.com/podverse/podverse-orm/commit/cb52ae7b87daefc1f4096b469aee86a039792820
https://github.com/podverse/podverse-parser/commit/bba6825c964d9e5c414a582b8a96a78e1e568a7e

2025-12-29 12:11:43 UTC

- reply

@nprofile…5yxp now that I think of it, I can just make our partytime fork skip parsing remoteItems with invalid guids before they reach the database.
Sorry for the walls of text over 1 character missing 😅 an invalid remoteItem shouldn’t stop the others from being saved. I’ll update partytime validation.

2025-12-29 12:04:17 UTC

- reply

@nprofile…5yxp but I understand that people who are early adopters of a tag like remoteItem may be manually creating these feeds by hand, which makes it possible for something like copying 35 characters of a podcast guid when a podcast guid needs 36 characters in UUIDv5 format to adhere to the P2.0 spec. This is super easy for an RSS generator to programmatically validate and make sure never happens…but not easy when people may be typing or copy/pasting fields like this manually by hand.

2025-12-29 12:00:15 UTC

- reply

@nprofile…5yxp I will try parsing your other list feeds later this morning.
I also may be able to add filtering before we do a bulk save Postgres transaction, to remove remote items with invalid guids so the error will be avoided.
Also, new Podverse can handle if a remoteItem or value tag points to a feed that does not exist, but trying to save “invalid data” like this edge case (an invalid podcast guid structure) errors the parser. imo it should be the RSS generator’s job to not publish invalid guids

2025-12-29 06:14:23 UTC

Podverse has started maintaining a fork of the Partytime RSS Feed parser library.
Partytime is a comprehensive Node.js RSS podcast feed parser, with support for all Podcasting 2.0 namespaces.
Thank you to @nprofile…0lgv and @nprofile…y5nd for creating the library! The code is easy to work with, and the unit tests are appreciated 😄
https://github.com/podverse/podverse-partytime
#podverse #pv2

2025-12-29 02:28:31 UTC

- reply

@nprofile…5yxp it's both. A feed with that GUID does not exist in PI (I used the PI API to query for it), possibly because the GUID is invalid, and PI rejected it.
26d34388-83a3-5fe3-94eb-be4d0008449
That invalid UUIDv5 is 35 characters, and 8-4-4-4-11 characters.
A valid UUIDv5 is 36 characters, using 8-4-4-4-12 format.
If the feed is manually edited, it could just be a typo, hitting a backspace unintentionally, or bad copy/paste.

2025-12-29 02:02:30 UTC

- reply

@nprofile…5yxp strict on valid data handling. For now at least, our parser rules are going to be strict bc it helps everything overall, but if it's not practical we may end up removing some of the strictness by the time PV2 goes live.

2025-12-29 02:01:38 UTC

- reply

@nprofile…5yxp Ideally, whatever RSS feed editor Homegrown Hits is using would automatically insert a valid UUIDv5 guid, so invalid guids can't get in there.
I *could* remove the transaction, and save all the rows as individual queries, OR make the transaction save a rollback state and gracefully fallback and move onto the next one when one row within the transaction fails...but that will multiply the number of queries needed to save the feed data by a lot.
Anyway, it's experimental for us to be so

2025-12-29 01:59:03 UTC

- reply

@nprofile…5yxp so PV2 is attempting to save the channel remoteItems in a transaction, which is a batch of 100 rows at a time. if 1 of those rows fails to be saved, then the whole batch of 100 fails (the transaction "rolls back"). We do this because transactions are more efficient database operations, and it helps us scale affordably.
If your feed is just cloning data found in Homegrown Hits, then yes, issue is in their source feed, and Homegrown Hits may not parse successfully.

2025-12-29 01:39:10 UTC

- reply

@nprofile…5yxp I should clarify... PV 2.0 is strict on enforcing UUIDv5 guids for the feedGuid (which is the problem here, and corresponds with the <podcast:guid> tag), but itemGuid can be any string (like a url), and not just UUIDv5.
From the spec: https://github.com/Podcastindex-org/podcast-namespace/blob/main/docs/tags/guid.md

2025-12-29 01:38:54 UTC

- reply

@nprofile…5yxp I should clarify... PV 2.0 is strict on enforcing UUIDv5 guids for the feedGuid (which corresponds with the <podcast:guid> tag), but itemGuid can be any string (like a url), and not just UUIDv5.
From the spec: https://github.com/Podcastindex-org/podcast-namespace/blob/main/docs/tags/guid.md

2025-12-29 01:35:00 UTC

@nprofile…5yxp it appears there is an invalid guid in your Homegrown Hits playlist
<podcast:remoteItem feedGuid="26d34388-83a3-5fe3-94eb-be4d0008449" itemGuid="bd15f03c-8e7d-5e38-bddd-249e0b5760c7"/>
26d34388-83a3-5fe3-94eb-be4d0008449 is not a valid UUIDv5
Podverse 2.0 is strict on enforcing valid guids, so this causes our parser to not parse the entire feed.
I looked up 26d34388-83a3-5fe3-94eb-be4d0008449 in Podcast Index, and it's not there either.
Maybe a typo?
https://raw.githubusercontent.com/ChadFarrow/chadf-musicl-playlists/refs/heads/main/docs/HGH-music-playlist.xml

2025-12-29 01:05:56 UTC

- reply

@nprofile…0lgv I think I figured out the correct config for v3 to v4 fast-xml-parser. I shouldn't be bothering you further 😂 thanks again for all the work you've done.
https://github.com/podverse/podverse-partytime/pull/1/files#diff-e678d75e81c43e5f6be2f2c1675c119f8aa3870a7346cdd1b6a6ff3641e873c7L6-L21

2025-12-28 19:44:58 UTC

- reply

@nprofile…0lgv I think I’ve made progress on this. The he.decode line in the parser config was stumping me for a while. Things seem to be working when I remove it. Now I gotta figure out what problems removing that could cause…

2025-12-28 08:40:55 UTC

- reply

@nprofile…0lgv I tried to make progress with an upgrade from v3 to v4, but I have not been able to figure out how the new validated and parsed output format is supposed to work.
I imagine many helper methods that you wrote would need to change. I'm open to doing this work, but I don't have a clear direction on how to proceed yet. If there's any insight you can offer I would appreciate it, but if you're too busy no problem.

2025-12-28 08:40:28 UTC

- reply

@nprofile…0lgv I tried to make progress with an upgrade from v3 to v4, but I have not been able to figure out how the new output format is supposed to work.
I imagine many helper methods that you wrote would need to change. I'm open to doing this work, but I don't have a clear direction on how to proceed yet. If there's any insight you can offer I would appreciate it, but if you're too busy no problem.

2025-12-28 08:37:30 UTC

Hey @nprofile…0lgv I started working on a fork of partytime. The codebase and dev setup you created is great! I was able to add the "medium list" types, as a first feature addition.
I am wondering if you have any insight on how upgrading v3 fast-xml-parser to v4 fast-xml-parser would work.
There is a moderate security vulnerability with v3, and I tried upgrading, but apparently the output of the parser is different with v4, and your validation methods are incompatible.
https://github.com/podverse/podverse-partytime/pull/1/files

2025-12-27 00:17:34 UTC

- reply

@nprofile…dzgg @nprofile…y5nd we should be able to integrate with the Rust parser outputs. I think we may want a layer of Typescript types to go with the output. That’s just a data mapping exercise tho, and not strictly necessary.

2025-12-26 18:38:16 UTC

- reply

@nprofile…dzgg but also...partytime just seems really stable for our purposes, so I'm not sure if it is worth it for us to break from our full-stack JS pattern to incorporate feedparser/Rust. Also partytime uses Typescript types which is good for our builds and workflow.
Anyway, if no one else is eager to maintain partytime, I think I will fork it tomorrow, make the changes needed, and maybe longer term start a migration to feedparser. I see you made some PRs that are not merged and I can take a look.

2025-12-26 18:35:06 UTC

- reply

@nprofile…dzgg I have seen it. I'm open to using it at some point, but podcast-partytime works well with our full-stack JavaScript infrastructure.
I'm thinking it would take me a day of work to add the missing pieces we need to podcast-partytime, but switching to feedparser which is written in Rust would be a bigger initiative.
partytime logic is only used in podverse-parser/src/lib/compat and podverse-parser/src/lib/rss, so swapping it out doesn't seem crazy difficult
https://github.com/podverse/podverse-parser/blob/v5-develop/src/lib/rss/channel/channel.ts

2025-12-26 17:36:24 UTC

- reply

@nprofile…0lgv @nprofile…5yxp @nprofile…dzgg @nprofile…tmcv @nprofile…ezja @nprofile…y5nd @nprofile…wsp3
So I'm wondering, Ryan would you like to continue maintaining Partytime, or should we move forward with a fork?
And if we do fork it, does anyone else want to be the maintainer?
I can bring it under the Podverse org umbrella, and I don't think it would be overwhelming bc Partytime seems to do 99% of what it needs to already, but if someone else would like to be the maintainer, we'd certainly appreciate it.

2025-12-26 17:32:38 UTC

The Podcast Partytime JS library that @nprofile…0lgv made has been working great for us, but it needs a few (I think light-weight) updates to get us past some blocking limitations.
One of the issues is that Partytime doesn't parse the "list" medium types, which is an issue for list feeds that people like @nprofile…5yxp make.
https://github.com/RyanHirsch/partytime/issues/61
Ryan has done amazing work but I'm not sure if he is still able to maintain it.
https://github.com/RyanHirsch/partytime/issues
cc @nprofile…dzgg @nprofile…tmcv @nprofile…ezja @nprofile…y5nd @nprofile…wsp3

2025-12-24 08:48:09 UTC

- reply

Playlists also work with clips. If you have clips in a playlist, then press the "next track" button, the player will skip to the next clip in the playlist.
The "previous track" button should also work.
Trying to make the auto-queue experience as smooth as possible.

2025-12-24 08:46:18 UTC

Podverse Alpha Web update 🛠️
[5.1.11] - 2025-12-24
Added playlist auto-queue support. When you press the play button on a playlist item, then after the item finishes playing, the next item from the playlist will play.
Enabled repeat and shuffle mode for playlist playback.
Fixed a variety of issues with the queue and auto-queue when pressing the previous track and next track buttons.
Sample playlist: https://alpha.podverse.fm/playlist/4DNEp87yRs
Merry Christmas! 🎄
#podverse #pv2
https://cdn.masto.host/podcastindexsocial/media_attachments/files/115/773/666/318/852/461/original/55e5509f122118c9.png

2025-12-24 08:45:27 UTC

2025-12-22 07:51:23 UTC

Podverse Alpha Web update 🛠️
Added the Music Livestreams and Music Livestream pages.
Also added the Repeat and Shuffle buttons to the player.
Reduced some font/image sizes a bit, and applied some fixes for mobile layouts.
Coming soon: auto-play next behavior for Playlists, remember last repeat/shuffle setting on page load, and Artist pages.
https://alpha.podverse.fm
#podverse #pv2
https://cdn.masto.host/podcastindexsocial/media_attachments/files/115/762/039/006/463/669/original/c18b85e53f678d94.png

2025-12-20 02:28:40 UTC

- reply

@nprofile…hs8h @nprofile…efpk I love the idea. Do you have any sample, usable RSS feeds online already?
No rush if not, but when you do have some content we can parse, we can see how Podverse handles it.

2025-12-20 02:21:41 UTC

- reply

@nprofile…efpk @nprofile…hs8h looks interesting! Will the RSS feeds support any Podcasting 2.0 namespaces?
A simple one could be the `<podcast:medium>music</podcast:medium>` within the `<channel>` for every feed. This signals to apps that a feed should be handled as a "music experience" instead of a "podcast experience".
https://github.com/Podcastindex-org/podcast-namespace/blob/main/docs/tags/medium.md
The new Podverse (open source, Podcasting 2.0 compatible app) has a heavy RSS music focus. A live alpha demo can be found here https://alpha.podverse.fm/album/YX0N5AHegy

2025-12-20 02:16:12 UTC

- reply

It's a big, very modularized infrastructure...but the path to getting it running should be fairly easy (ideally), with the helper scripts we have created.

2025-12-20 02:15:01 UTC

I just updated the "Getting Started" documentation for Podverse Alpha local development.
If there are any developers interested in trying to get the whole infrastructure running on your machine, we'd appreciate help verifying our documentation, and suggesting improvements if needed.
https://github.com/podverse/podverse-ops/blob/v5-develop/dev/local-dev-setup.md
I also think I am going to close / archive all of the Github issues that pertain to the previous version of Podverse, since we are only working on the next generation version going forward.

2025-12-18 20:31:50 UTC

- reply

@nprofile…5yxp sorry, too early. Anything on Alpha could be deleted later. There will be a complex migration process from PV1 to PV2 when it finally goes live, and the alpha data won’t be part of that (hence the disclaimer and fact we’re not charging for alpha memberships).

2025-12-18 20:04:42 UTC

- reply

@nprofile…5yxp well I’m happy to refine it to get it to where it needs to be.
The issue of not having everything possible in our database (until people manually add it) may not have an easy answer…our new infra is designed like this to prevent spam and database bloat.
However, if there aren’t too many music feeds today (<10000?), we could potentially just add every music feed that is in the index to our directory.
A V4V only filter could be nice to highlight those feeds.

2025-12-18 17:29:04 UTC

- reply

@nprofile…y5nd @nprofile…wsn5 no pressure but I noticed before the PI search endpoint apparently lets you filter by music but not by video.
It’s not critical but if PI search ever supports filtering by any medium, we could add radio buttons below the search bar so people could refine their search.
If it takes a lot of work, it’s not worth it, but if it’s trivial to add, could be nice.
https://cdn.masto.host/podcastindexsocial/media_attachments/files/115/741/745/889/288/115/original/f738f22eff951e79.png

2025-12-18 16:59:48 UTC

- reply

How about fighting LLM with LLM? Adding an LLM service layer all feeds must pass through before they are added to PI directory? The service flags suspicious feeds and saves references to them in a separate table.
Feeds marked as spam incorrectly would still need to be corrected somehow…maybe via a privileged endpoint RSS hosting companies support teams can interact with, and/or crowdsourced maintenance by trusted community members.

2025-12-18 16:43:06 UTC

- reply

@nprofile…5yxp that way. Share is common enough maybe we add it but I am concerned about “choice paralysis” if the dropdown menus have too many options.
Thinking of adding a Like button to quickly add items to a Favorite Podcasts or Favorite Music playlist, similar to the heart button in PV1. Infrastructure already there, would just need to add buttons for it on the page.

2025-12-18 16:40:15 UTC

- reply

@nprofile…5yxp yeah auto may need to allow customization but presumption is listener wants to hear the album OR the playlist. Playlist autoqueue may be incomplete, not sure.
What tracks are missing from global? It’s “global” in the sense they’ve been added to our directory by someone searching for it and pressing add feed.
“Top” filter not implemented yet until stats tracking is enabled.
Shuffle and repeat coming soon.
Share button for track can be added although you can Go To Track and copy url

2025-12-18 16:34:21 UTC

- reply

@nprofile…5yxp @nprofile…8nsk reorganizing your queue with thousands of items a huge operation bc every rows position in your queue would need to be updated when 1 row changes 🤦‍♂️