Comment by jauntywundrkind | Hacker News Reader

Comment by jauntywundrkind | Hacker News Reader

Parent Full thread

jauntywundrkind

That's quite unspecific and unhelpful.

Personally I find XMPP much makes sense. Sure it's this weird streaming XML thing, but there's a request/response pattern (IQ) inside that seems fine. I love how nicely it composes: accounts have a tree of nodes they can define for whatever content they want, which is a sensible & flexible base. Then there's pubsub, and ACL capability specs on top of that. Everything stacks relatively sensibly. The past decade has seen some good XMPP Enhancement Protocols (XEP) to create best practices & recommended feature sets.

It was such a small jump to create a full "everything app" atop XMPP baseline capabilities:

> Libervia [ed: nee Salut-a-Too] is a all-in-one tool to manage all your communications needs: instant messaging, (micro)blogging, file sharing, photo albums, events, forums, tasks, etc.

https://libervia.org/

It's unfortunate that the top rated comment is uncontestable blanket desparagement. Can we raise this from a low criticism to something respect-worthy?

XMPP missed the boat largely because it couldn't handle multiple clients correctly for years - the default is to deliver messages to one of your clients, you need an extension to do the sensible thing, and that extension spent years in bikeshed limbo right as smartphones were taking off and people started wanting to use the same messenger on their phone and computer at the same time. (I've heard that performance/battery issues from XML validation didn't help either)

Personal speculation but I blame the "everything is an extension" model - it was meant to reduce fragmentation and allow clients with different featuresets to interoperate, but in practice adding a new XEP seems to have all the downsides of making a change to a non-extension-based standard (you still have to get all the clients to agree) and none of the upsides.

This kind of checks out for me. But also, there have been decent protocols around this for a long time, that many clients & servers have implemented. From Multi-device in the excellent Modern XMPP:*

> XEP-0280: Message Carbons - for "live" synchronization of conversations between online devices. XEP-0313: Message Archive Management - for "catch-up" of messages that were exchanged while a device was offline

https://docs.modernxmpp.org/client/protocol/

The XEP-0313 spec dates back to 2012 which is less old than I expected, and that's only the 0.1. So, very fair point.

More generally, the above complaint was about the development experience of XMPP. I feel somewhat like complaining about XMPP being failed is well tread from a why consumers didnt adopt it view, that negative sentiment abounds & everyone is more than happy to cast blame as to why. I've seen a lot less complaints about the development experience, and felt like maybe there was some novel fruitful grounds that a more developer-centric view might have been able to open up.

> but I blame the "everything is an extension" model ...

100%. If you're not keen on self-hosting and want to use any one of the many, many public servers this becomes such a pain, and leads to the same "decision deadlock" trying to get friends to join Mastodon or Lemmy ("which one do I want? there are so many, how do I know if it's good"). Because this is a thing:

    https://compliance.conversations.im/

"Hmmm, does xyz.com support XEP-1234 for message archiving?" or whatever it is; there's a real uphill social battle unless you make those choices for your friends to get started. While Signal is not perfect, it's easy onboarding without confusing XEP which-server choices for the average human. $0.02 I struggle to get people on Signal as it is.

Edit: found it, it's XEP-0313 and only 92% of public servers support it. Only 86% support Push Notifications, 94% OMEMO. One can argue server operators have disabled them but it points back to decision deadlock.

There is also Quicksy.im which offers Signal style onboarding and contact discovery but you also get a routable XMPP address.

95% XEP feature coverage. https://compliance.conversations.im/server/Quicksy.im/

Quicksy is run by the same guy who developed Conversations and this compliance test btw. It will work for its target audience. If you know what a XEP is, you don't need Quicksy.

> Personal speculation but I blame the "everything is an extension" model - it was meant to reduce fragmentation and allow clients with different featuresets to interoperate

I could be wrong, but that reads like you suggest that there is an alternative to the "extension model".

However, any solution where standardization and implementations are independent entities, and thereby experience a sufficient degree of freedom, will have a trajectory to a situation where you have a robust core specification and optional extensions.

Think about protocols like SMTP and DNS—each has a foundational core that’s been expanded upon by numerous optional features.

> any solution where standardization and implementations are independent entities, and thereby experience a sufficient degree of freedom, will have a trajectory to a situation where you have a robust core specification and optional extensions.

You can call the kind of optionality that those kind of protocols have "extensions" if you want, but it's a lot more lightweight than the kind of extensibility that XMPP was designed around, which is the thing that I'm arguing did more harm than good.

Optional features is something different than uncoordinated extensions which might conflict with each other.

I am not sure if I would phrase it that way.

(Seemingly) conflicting extensions are another consequence of the loosely coupling between standardization and implementations. In addition, the emergence of several functionally overlapping extensions is stimulated by the freely accessible standardization process.

Especially in the early phase of an extension, you want to encourage experimentation with different approaches. Early selection would be disadvantageous.

> Especially in the early phase of an extension, you want to encourage experimentation with different approaches. Early selection would be disadvantageous.

With any standard you can experiment what you want, nobody* even can prevent you from doing it no matter how inaccessible the standardization process is.

The standardization process comes into play when you think you have found a good solution, which should be adopted by THE standard respectively the ecosystem.

What matters is what the standard itself looks like, do you have a coherent specification which specifies the current way of doing things, including optional components?

Or do you have a set of independent ways of doing it, because the standardization process doesn't actually decide what is the correct way of doing something (e.g. managing a group chat)

*okay technically not correct. Law can e.g. decide making e2ee illegal technology and criminalize even playing around with it.

> The standardization process comes into play when you think you have found a good solution, which should be adopted by THE standard respectively the ecosystem.

Na, the standardization process starts much earlier. Using the example of the IETF process, after which XMPP standardization process is largely modeled: standardization starts when you submit an I-D to IETF and/or approach an IETF WG.

> What matters is what the standard itself looks like, do you have a coherent specification which specifies the current way of doing things, including optional components? Or do you have a set of independent ways of doing it, because the standardization process doesn't actually decide what is the correct way of doing something (e.g. managing a group chat)

Well put and I totally agree (I think no one would have a reason to disagree with that statement).

Here's a specific complaint about XMPP, and possible explanation why nobody uses it any more. (I worked on a large-scale XMPP implementation back in the day.)

Presence. That's the colored dots indicating "is somebody online or not". The traffic needed to maintain presence scales by N^2, and in any large-scale implementation, the traffic to maintain presence data completely dominates anything useful.

Not to mention that for the past 15 years or so (ever since everybody has a connected cell-phone all the time) the whole idea of presence (am I online or not?) is either meaningless or just badly modeled.

So the result is a protocol which spends tons of bandwidth and battery maintaining metadata that is functionally useless. That's why the real world has run away from it as fast as possible.

>The traffic needed to maintain presence scales by N^2

Only true if we assume the average number of contacts scales linearly with the total number of users, right? But then, we could also assume that the average number of messages that a given user sends scales linearly with the total number of users, in which case the amount of traffic needed to transmit messages also scales by N^2.

A couple of easy fixes: cap the number of contacts at some large constant (as many services do), or just disable presence information altogether in your implementation. I'm skeptical that this played a major role in XMPP's lack of popularity, especially because e.g. WhatsApp and Facebook Messenger have presence information, and are still popular.

I wasn't trying to be specific, just an opinion on what I experienced 10+ years ago. Others are welcome to work with the protocol and develop their own opinions.

If your opinion is based on concrete experience, you could help people understand your position by sharing the specific aspects of XMPP that you dislike. An opinion without evidence or reasoning is not a valuable contribution to the conversation.

It's been well over 10 years since I worked with, almost 15. I remember issues with keeping multiple devices in sync, syncing them back up when a user comes back online, especially with multiuser chats. I understand that is probably better now, with carbons and archiving XEPs.

In general, it felt like, XMPP has too many "optional" features. The core protocol is tiny, but everything you need and want to make it useful is optional.

My opinion: XMPP is too little like Matrix (e.g. decentralized rooms, people as verification targets, messages (incl. e2ee ones) easily synced to all sessions) while Matrix is in a sub optimal state due to the Element/Element X client split

Don't worry, every single one of these discussion is a dozen fanboys trying to convince us that our problems back then weren't real :/

(JFTR, was a relatively happy user amongst fellow nerds & family until everyone just stopped using it, and also usage on mobile was terrible on early smart phones and fixed much too late)

In today's world of containerization and AI powered UI automation, perhaps a single user-facing client could be viable again, powered by hidden per-service clients under the hood. Where each services' UI state is continually monitored and interacted with by an AI directed by user interactions in the visible interface. That would be against the services' ToS probably, but it could work I think.

Who needs APIs when a computer can exploit the analog hole and use the same affordances as a human?

I don't think having a single user-facing client has ever really been held back by the technology. It's always the services being intentionally proprietary, intentionally breaking 3rd party clients, and ToSs making it risky to do.

I mean, different platforms exist just the same as back then. Windows, Linux, Android, iOS (and let's say some nerds will make it work on osx and the BSDs from the linux version).

That was a problem back then and it's a problem right now

There have always been approaches to multi platform apps, the limitations attempts at this (both in the past and currently) have repeatedly been a great UI with hostile 3rd party services.

A "User-Agent" if you will.

Reply on news.ycombinator.com