Sir Alecks Gates

https://podcastindex.social/@alecksgates

Podcasting 2.0 consultant
Master of the protocols
Knight of Podcasting 2.0
Inventor

INTJ-A

#Wisconsin -> #Tennessee

Essays: https://write.agates.io
GitHub: https://github.com/agates
GitLab: https://gitlab.com/agates
XMPP: xmpp:[email protected]

Public Key

npub1j0hf3w389k69mc6s0etpasn5658e5pd2zl03sj5fs3s55geem3dq94u952

NIP-05 Address

[email protected]

Profile Code

nprofile1qqsf8m5chgnjmdzaudg8u4s7cf6d2ru6qk4p0hccf2ycgc22yvuacksprpmhxue69uhhyetvv9ujumt0d4hhxarj9ecxjmntqyxhwumn8ghj7mn0wvhxcmmvqyg8wumn8ghj7mn0wd68ytnhd9hx2qg5waehxw309aex2mrp0yhxgctdw4eju6t0rwxrhr

Publishing to

Author Public Key

npub1j0hf3w389k69mc6s0etpasn5658e5pd2zl03sj5fs3s55geem3dq94u952

Show more details

Last Notes

2026-08-07 22:26:09 CEST

- reply

I will be running my own iroh relay soon to be testing my app, so you'll be in good company.
I've done a *huge* amount of testing on devices behind double NAT.

2026-08-07 20:46:37 CEST

- reply

should I dig out an old raspberry pi zero to do some armv6 testing?

2026-08-07 19:02:46 CEST

- reply

here I am getting ready to test video on Android TV lol

2026-08-07 16:50:01 CEST

- reply

my attention is not for sale :)
If attention-dependent media dies out, so be it.

2026-07-29 19:27:59 CEST

- reply

Right. As long as we can figure out a compromise for splits/fees, I'm okay with it. Until then I'm not interested. I think we underestimate the lengths people are willing to go for support if they truly want to.
I realize it's probably controversial, but I think the boost button was always an unhealthy anti-feature. Really cool, no argument about that. But gamification in general is not good, even if it's "open.". I feel the same way about YouTube/twitch gamification.

2026-07-29 18:56:25 CEST

- reply

Yeah, I have zero plans to support lightning. But y'all have known my stance from the start.
Playlists are on my list of course, but what prevents them from being worth the time?

2026-07-29 18:54:12 CEST

- reply

Man, even the big music apps struggle with this. I almost wonder if we should crowdsource it and hook up some of the models Dave was working on.
Personally, I think we still need genre tagging. I may just implement a draft spec and tell some of the musicians/curators I support it.

2026-07-29 18:35:52 CEST

- reply

I'm trying to focus on music in my app. Seems like we still have a lot of work to do, but at least we have the basics.
Do you have any ideas for improvement of music in pc20?

2026-07-23 17:27:33 CEST

- reply

My app is explicitly anti "algorithm." It's surprising how many of the everyday design decisions we're accustomed to are unhealthy for us.
Not that discovery is inherently bad. I just think it needs to be intentional and neutral.
But I also have nothing to gain from attention.

2026-07-18 04:30:29 CEST

- reply

A sign from the Boardroom Gods.

2026-07-17 19:26:16 CEST

- reply

Every time I have to use the iOS keyboard I feel like all of the Apple users have Stockholm Syndrome.

2026-07-10 18:53:01 CEST

- reply

Some things are genuinely hard to interpret without explicitly being told. They are easy signals, but provide crucial context.
Others are additive and improve user experience, but can usually be derived with confidence. They fundamentally come down to a caching problem or opinionated output. Nice to have, not critical. And everyone's opinion is different.

2026-07-10 18:51:36 CEST

- reply

From a pc20 specification perspective, standards generally fall into "soft" and "hard" problems.
I've always liked to say "don't make the app guess."

2026-04-07 19:22:02 CEST

- reply

Memory safe doesn't make logic errors disappear

2026-03-27 15:46:38 CET

- reply

with plexor I attach a trace context to every event for this very reason
https://gitlab.com/plexo/plexor/-/blob/alpha-release/plexor-core/src/payload.rs
https://gitlab.com/plexo/plexor/-/blob/alpha-release/plexor-core/src/logging.rs

2026-03-27 04:30:25 CET

- reply

Strange that a peer discovery process would hang a publishing component.

2026-03-27 03:25:31 CET

- reply

I might have to finally spend some time with it....

2026-03-25 16:32:23 CET

- reply

I was using "glow" for a few things. Might try this.
https://github.com/charmbracelet/glow

2026-03-23 18:59:49 CET

- reply

I can't speak for Dave, but I think this experiment exists in this state in the event of an immediate Hive disaster.

2026-03-23 18:59:02 CET

- reply

As far as I'm concerned, it's Podping just on an additional messaging layer.
I have yet to get my hands on any of it due to more important obligations, although Dave and I talked through the implementation quite a bit.
Personally, when I can get some time to work on it, I am going to try to add both gossip listener and writer to podpingd which will watch both Hive (which it already does) and gossip.

2026-02-20 22:52:08 CET

- reply

#careergoals

2026-02-20 22:29:00 CET

- reply

It's all good, I'm not waiting on you :)

2026-02-20 17:28:44 CET

- reply

We think it's just the list of endpoints. It hasn't been updated in 2 years. So I'm going to update the list and make it configurable.
Maybe have dave provide an endpoint of known good endpoints the user can opt-in to using.

2026-02-20 17:21:47 CET

- reply

I'm going to try to get the hive writer changes done tonight to help mitigate this for podping.cloud.
Ultimately we are going to try to get a non-Hive solution in place. I have a decent plan but I want to hear what @npub1rcn…dw4h thinks about it first.
Unfortunately, the next few months are going to be very busy for me between work and moving states. And I'm just a volunteer with all this.

2026-02-20 16:49:38 CET

- reply

How are you submitting to Hive?

2026-02-20 02:05:17 CET

- reply

> @agates can’t even be bothered to contribute
What? Sorry, James, that's just disingenuous. And I absolutely have no ill-will against you in this matter.
I've merely stopped caring.
If I need to resurface the issue for something in the future, I will.
However, I imagine it will resurface on its own. It always does.

2026-02-18 04:20:52 CET

- reply

The industry never got to the point of figuring out how to solve the same amount of problems with less work.
Now the opposite is happening.
The ratio of software developers to available work was already a bottleneck.
If you consider it as a problem of energy efficiency, this was inevitable. As far as I can tell, there is no increase in total energy efficiency for the output of quality work.
And that's what makes it different from other technological revolutions.

2026-02-14 17:35:24 CET

- reply

Oh I don't know, maybe sending boost metadata to the podcaster's host directly

2026-02-14 17:27:23 CET

- reply

I know.
But the URLs are in the wallet metadata which is the only way the podcaster knows about it without the metadata being controlled by the podcaster.
It's not the best we have and you know better.

2026-02-14 05:12:34 CET

Wallet-hosted value metadata URLs makes anything other than lightning impossible. I will never support it for that reason.

2026-02-11 17:34:49 CET

- reply

If only hosts that use podping would put usesPodping=true in their feed :)

2026-02-09 15:54:34 CET

- reply

i'm all about traces these days, probably because i'm using so much async/multithreaded code.

2026-01-30 18:37:01 CET

- reply

This "standard" leaves out any other apps that want to view them because the metadata is not stored somewhere defined in the feed.
It's not particularly impressive.

2026-01-28 17:21:52 CET

- reply

I'm really worried about current junior developers. They can't learn like this.
Maybe that won't matter in the short-medium term. But when my generation starts to retire, it's going to be a problem if it keeps up like this.
I suppose that's the usual sci-fi story we see, where the builders die off and the automated systems keeping everything going start to fail.

2026-01-28 16:58:49 CET

- reply

right, I suppose I have the experience to ask all this stuff ahead of time. That's basically my day job lol.
(The brainstorming thing, not C++)
I can see why formalizing that would be beneficial to a lot of people.

2026-01-28 16:54:45 CET

- reply

yeah with Gemini I just do all that manually, haven't had any issues.
I guess that speaks to where I draw the line. I am not even remotely vibe coding. I am using the agent to help the speed of implementation keep up with my brain. I am reviewing just about everything it does.
Maybe I'll expand upon that eventually? But the work I'm doing is to reduce stupid development boilerplate via a protocol in the first place, and I started that long before AI came into the picture.

2026-01-28 02:22:57 CET

- reply

I'm over here not using any skills or plugins and doing just fine lol

2026-01-11 23:05:45 CET

- reply

I honestly think we can gamify it

2026-01-11 04:59:45 CET

- reply

I was sticking with Junie in the beginning, but I ended not using all the credits i bought ahead of time because I was busy with work. So I ended up cancelling.
And I already had a Google AI Pro subscription from a promotion from my Pixel 10 Pro purchase, so I figured why not try it.
It turns out, it's really helpful for non-code projects too like capnproto config or system configuration. And I find it easier to manage across multiple terminal sessions + it crashes less.

2026-01-11 03:05:05 CET

- reply

it sort of depends on how much experimentation I need to do, but even if I'm using the CLI, I'll usually open a CLI editor to the side.
Nothing beats an IDE for exploring code though and I still perform reviews.
Very similar to just having a junior dev to check in on tbh. Just a lot cheaper.
I opted for the Google AI Ultra subscription because I kept hitting the daily limit on the pro plan. And they have a promotion for $150/3 months... not sure if I'll keep it yet.

2026-01-10 17:59:40 CET

- reply

I'm using Gemini CLI. It's been great. Gemini 3 especially.

2026-01-04 18:10:26 CET

- reply

Author never heard of type hints in python (AI loves them)
@npub1fmx…wp74 how's your type hint game? :)

2026-01-01 00:29:46 CET

- reply

I'd love to see UnifiedPush in Vanadium, as I'm aware of the security tradeoff from using a Firefox-based browser on Android. Implementing UnifiedPush is relatively straightforward.

2026-01-01 00:23:28 CET

PSA: IronFox on Android supports UnifiedPush.
A great feature if you use any sites/PWAs with browser notifications and want to avoid metadata leakage through Google's push notification servers.
If you're on GrapheneOS, it's available in Accrescent. You have to manually enable UP in the browser settings

2025-12-31 23:48:19 CET

- reply

I made a proposal for this.
https://github.com/Podcastindex-org/podcast-namespace/discussions/620

2025-12-17 07:11:37 CET

- reply

Add to that people don't even agree on the definitions of a lot of genres. So expecting anyone to agree is basically pointless.
The owner can tag it what they want, but should also expect to follow some basic guidelines if they want it to be useful for discovery.

2025-12-17 06:51:33 CET

- reply

I think it would be rather straightforward for the community to maintain a slug list like with anything else.
That slug list could be internationalized

2025-12-17 06:50:07 CET

- reply

Just need some music feeds to implement it.

2025-12-17 04:31:09 CET

- reply

We should just start using this :)

2025-12-13 18:34:36 CET

- reply

also it's funny how I'm also writing my own rust message queue system 😂

2025-12-13 18:33:56 CET

- reply

Interesting. Do you want persistence?
During a planned maintenance window you could just turn off the ingest and wait for jobs to finish and the queue to flush, but that's sort of annoying.
In an unexpected restart you would lose queue metadata and have to rebuild from object storage on startup.
I feel like redis is the path of least resistance here because of the persistence options and it being very well documented.
(Or postgresql unless you surpass a certain # of items per second)

2025-12-13 03:18:39 CET

- reply

hmm shouldn't there be a parse queue in front of the parser?
Also how much of a bottleneck is it to generate SQL statements? In my experience, if we ingest JSON strings as serde structs, it's pretty trivial to use those structs directly in an sqlx insert.

2025-12-09 17:53:22 CET

- reply

I wouldn't even use the same API key set or database. It's a different thing, and the point of the initiative is to make it cleaner and easier to maintain.

2025-12-09 17:35:40 CET

- reply

Define publisher permissioned keys?
But yes if there is an API in front of the queue it would be trivial. You would want regular health checks and a dashboard to report the status of distributed workers.

2025-11-07 18:52:27 CET

- reply

You just said we should continue to host metadata on wallets and now you're coming back with this, which appears to be an entirely different thing.
This is going in circles. One seems to be a specification and the other is an implementation detail.
I don't really care how you implement it. As long as the standard to host metadata is payment agnostic and determined by the podcast/host in the feed. A link to that metadata in the payment is fine.

2025-11-07 18:43:13 CET

- reply

What about other payment methods?

2025-11-07 18:33:30 CET

- reply

This doesn't make any sense. They could just host their own metadata.

2025-11-07 18:19:12 CET

- reply

I think I'm missing something. What does the wallet have to do with it? Why does it need to be aware of the metadata outside of a link to it?
I'm replying during my day job (ask Adam/Dave how hectic it is) from a mentioned notification, so maybe I overlooked something obvious.

2025-11-07 17:42:19 CET

- reply

Sorry what exactly is the problem with an API in the feed managed by the podcast host? I haven't seen a legitimate criticism.

2025-11-07 17:41:08 CET

- reply

The solution needs to be payment provided agnostic regardless. Bitcoin or not.
Everything else in the feed is from the "host" -- the source of truth. Why would we change that now?

2025-11-07 17:35:29 CET

- reply

we should deleted all of the noagendatube feeds from the index :)
I guess I can do that from the admin page right?

2025-10-19 03:18:05 CEST

- reply

I think a growth mindset is important to have, but it's a lot more specialized than we're led to believe.
It's not healthy to outright reject such specialization, either. Some things *do* need to scale, even if one can't comprehend why. We all have our own problem domains.
Desire for profit is another matter. People will do anything to retain their social status and power, and I don't see that stopping any time soon.
Maybe the cycle of empires is inevitable?

2025-10-14 03:02:10 CEST

- reply

And now we live in a world where our most valuable asset is our attention and the average person's spending power is almost irrelevant to the economy.
Technofeudalism, perhaps, for lack of a better word.

2025-10-09 18:57:08 CEST

- reply

Faker is a great library.

2025-10-09 05:52:25 CEST

- reply

I like them for basic things on Kagi.
I guess there's a threshold where I have my own confidence level of the response vs the criticality of the information.
I only really use Kagi though.

2025-10-07 20:21:33 CEST

- reply

would be nice to see playlists like this show up in other apps :)

2025-10-06 13:49:25 CEST

- reply

I will have something in the works but there's no major changes, just some updates of some of my libraries.
I need to get a new reason code (newItem) and the radio medium added, haven't been in a rush though.
(Behind the scenes I wrote my own event handling library/protocol called "plexo" which I use in podping-hivewriter. I'm making a rust version to use in podping.cloud and podpingd. It's been very educational.)

2025-10-06 13:39:27 CEST

- reply

updated with what?

2025-10-05 02:00:35 CEST

- reply

The playlist medium should be podcastL, no?
Title could be "Golden Globes Nominees 2025" or something, as an indicator.
https://cdn.masto.host/podcastindexsocial/media_attachments/files/115/318/607/536/836/681/original/259980f9c14f5f6e.png

2025-10-03 02:42:53 CEST

- reply

everybody panic!!

2025-10-01 22:47:52 CEST

- reply

If it's Podping.cloud you can and should have a blocklist of abusers. If they want to spam Podping they can do it on their own dime.

2025-09-28 02:46:22 CEST

- reply

I'll have to spend a while doing my own research to come to a conclusion.
My initial instinct is that we would still want to keep the metadata separate for flexibility.
And what happens to "splits"? Both for guests and payment activated services like your boost chapter images.
(didn't see this at first, my long-deleted AP username is still cached here LOL)

2025-09-27 15:49:32 CEST

- reply

Sorry I didn't catch it. Firewall troubleshooting session during the board meeting :/

2025-09-18 05:18:35 CEST

- reply

I am disappointed that alternative, self-hostable metadata solutions were not discussed.

2025-08-09 17:41:10 CEST

- reply

Another FYI, the peertube live chat plugin adds a chat tag with XMPP for live streams. Should help with your testing when you get to that point.

2025-08-08 16:55:40 CEST

- reply

Yes, the reason it wasn't usable in antenna pod until recently is there was an issue with the video format for playback. So I had it default to HLS.
(I still don't know why they never added support for HLS streaming, seems like an obvious win, but hey it's not my project)

2025-08-08 16:18:36 CEST

- reply

https://github.com/Chocobozzz/PeerTube/pull/5487

2025-07-20 22:17:21 CEST

- reply

I'm also working on a unified API for Podping push notifications, if you're interested. Might explore offering b2b support if there is demand.

2025-07-20 22:13:58 CEST

- reply

that's Steven Crader's service. I maintain the python one. But also have a rust version I'm working on (podpingd).
The python/JavaScript ones shouldn't be directly affected, but anything not expecting the new mediums or reason codes will need to be updated.

2025-07-20 21:57:26 CEST

- reply

how do you watch and push Podpings?

2025-07-20 17:28:09 CEST

- reply

If I don't get any feedback on this, I'm probably just going to do it :).
Need to get radio mediums added to Podping as well for @npub1thg…zq23

2025-07-10 05:29:35 CEST

Hi.
I'm thinking about adding a new reason code to Podping to indicate that an RSS feed has added a new item.
This would primarily serve as a means to allow consumers to prioritize parsing of new episodes vs regular updates -- similar to "live" and "liveEnd" but for standard items.
It would particularly be useful for notification systems on the client side. But we also use different reasons to prioritize Podpings in podping-hivewriter to (so "live" is sent before "update").
Pros/cons?