5 ms·

XMPP – Myths and Legends

44 points by octosphere 8y ago

rhn_mk1 8y ago
> XMPP is unreliable without a bunch of extensions. and > XMPP typically uses reliable stream transports, so even without extensions it's pretty reliable. are really saying the same thing. TCP doesn't protect from losing data that's been sent but never received due to dropped connections, which in practice happens a lot, especially in the mobile world. That even seems to be acknowledged in the explanation. A more honest answer would just point out the extensions. Apart from that, I agree. Hopefully XMPP in the IM protocol incarnation can reunite the pitiful, broken up messenger world.
- Boulth 8y ago
  Oh well I don't know about reuniting as it's been ages since XMPP exists and the IM world has never been as fragmented as now. But for what it is worth at least now we've got user friendly clients (such as Conversations.im and new Gajim) that one can recommended to non technical people.
  - beagle3 8y ago
    It was about as fragmented in 2000 or so: MSN messenger, AOL Messenger, ICQ, Yahoo messenger, POWwow... I counted 40 at the time. There were also things like Odigo and hyperbix/Gooey which were used for messaging even though their main function was “char on page” Now we have FB, WhatsApp, Telegram, Google, Google, Google.
- 51lver 8y ago
  We lose messages alllll the time due to moving between wifi and lte, or when in transit and entering a tunnel. Annoying.
  - seba_dos1 8y ago
    That's fixed by the Stream Management extension.
- SllX 8y ago
  > Hopefully XMPP in the IM protocol incarnation can reunite the pitiful, broken up messenger world. This doesn’t happen in a world where SV sets the standard. First, it was a mistake to use XML. The issues with XML are well known so I won’t rehash them here, but if you really want some reading material, this[0] will suffice. Second, end-to-end encryption by default is a critical necessity for any new standards. Having a central server to handle the key exchange is less than ideal, but still better than a total lack of encryption in most private communications because there is no shortage of entities looking to collect any kind of plain text traveling across the internet. At the very least, we can still make it expensive to process the private communications of private citizens. Third, nobody makes money from IM being an open standard. The only ones with anything to gain from open standards IM are the people sending messages and potential new entrants into a market. Critically this leaves out existing interests with their own lock-in, app and sticker stores and payment processing services. You can’t make something like Animoji or Memoji a selling point for a new phone if you can suddenly use those features from a Samsung or OnePlus. It also leaves out VCs which are more interested in a good ROI than upending the world of messengers to make it a better place. The alignment of interests just isn’t there. Fourth, for any kind of federated messenger system to take off and stay federated, you would need to be able to run your own server to connect to maintain your online presence, and make your own client to access your server. Not everyone would do this, and not everybody would have to, but you would most likely run into the issue of not enough people choosing to, with not enough back ends or front ends being developed and only a handful of the ones that are being worthwhile to actually use. Then once you decided there’s a better way to do X and try to change the standard to reflect this improvement, you would see delays in actually implementing this new method across different servers and clients causing the stack to become fragmented and frayed on both ends. Oh and this still doesn’t really make anybody enough money to actually be worth the effort. In the end XMPP failed to shake up the messenger world because it just wasn’t a great standard, and any commercial interests that brought it to the mass market eventually phased out support because maintining even an XMPP gateway just wasn’t worth the opportunity costs and labour for what amounted to an extra bit of technical debt that hardly anybody even used. It didn’t make money, and it’s only real feature was an anti-feature to interests demanding some kind of lock-in. Strangely enough you probably could use a blockchain to resolve some of the issues I raised, but you would still have to contend with the issue of making money and convincing enough people or coercing enough entities to actually make it into the messenger standard. Why bother when you could just use Slack with its craptastic Electron clients or iMessages or whatever the Google Messenger-of-the-Day is? Or whichever Messenger basically replacesd SMS in the country you live and any supplementary messenger you might need is right there in the App Store? [0] http://harmful.cat-v.org/software/xml/s-exp_vs_XML http://harmful.cat-v.org/software/xml/s-exp_vs_XML
edhelas 8y ago
Happy to see that link on HN :) I've been pretty much pushing XMPP for past decade, especially to show that it can also be used as a really decent base to do social networking (we didn't waited for ActivityPub to do federated social protocols, everything is pretty much there with Pubsub + Atom for more than 15 years). I am working on Movim (https://movim.eu/ https://movim.eu/) to bring the "modern XMPP" in a web platform for all with IM and social features built in :)

44 points by octosphere 8y ago

> XMPP is unreliable without a bunch of extensions. and > XMPP typically uses reliable stream transports, so even without extensions it's pretty reliable. are really saying the same thing. TCP doesn't protect from losing data that's been sent but never received due to dropped connections, which in practice happens a lot, especially in the mobile world. That even seems to be acknowledged in the explanation. A more honest answer would just point out the extensions. Apart from that, I agree. Hopefully XMPP in the IM protocol incarnation can reunite the pitiful, broken up messenger world.

Boulth 8y ago
Oh well I don't know about reuniting as it's been ages since XMPP exists and the IM world has never been as fragmented as now. But for what it is worth at least now we've got user friendly clients (such as Conversations.im and new Gajim) that one can recommended to non technical people.
- beagle3 8y ago
  It was about as fragmented in 2000 or so: MSN messenger, AOL Messenger, ICQ, Yahoo messenger, POWwow... I counted 40 at the time. There were also things like Odigo and hyperbix/Gooey which were used for messaging even though their main function was “char on page” Now we have FB, WhatsApp, Telegram, Google, Google, Google.
51lver 8y ago
We lose messages alllll the time due to moving between wifi and lte, or when in transit and entering a tunnel. Annoying.
- seba_dos1 8y ago
  That's fixed by the Stream Management extension.
SllX 8y ago
> Hopefully XMPP in the IM protocol incarnation can reunite the pitiful, broken up messenger world. This doesn’t happen in a world where SV sets the standard. First, it was a mistake to use XML. The issues with XML are well known so I won’t rehash them here, but if you really want some reading material, this[0] will suffice. Second, end-to-end encryption by default is a critical necessity for any new standards. Having a central server to handle the key exchange is less than ideal, but still better than a total lack of encryption in most private communications because there is no shortage of entities looking to collect any kind of plain text traveling across the internet. At the very least, we can still make it expensive to process the private communications of private citizens. Third, nobody makes money from IM being an open standard. The only ones with anything to gain from open standards IM are the people sending messages and potential new entrants into a market. Critically this leaves out existing interests with their own lock-in, app and sticker stores and payment processing services. You can’t make something like Animoji or Memoji a selling point for a new phone if you can suddenly use those features from a Samsung or OnePlus. It also leaves out VCs which are more interested in a good ROI than upending the world of messengers to make it a better place. The alignment of interests just isn’t there. Fourth, for any kind of federated messenger system to take off and stay federated, you would need to be able to run your own server to connect to maintain your online presence, and make your own client to access your server. Not everyone would do this, and not everybody would have to, but you would most likely run into the issue of not enough people choosing to, with not enough back ends or front ends being developed and only a handful of the ones that are being worthwhile to actually use. Then once you decided there’s a better way to do X and try to change the standard to reflect this improvement, you would see delays in actually implementing this new method across different servers and clients causing the stack to become fragmented and frayed on both ends. Oh and this still doesn’t really make anybody enough money to actually be worth the effort. In the end XMPP failed to shake up the messenger world because it just wasn’t a great standard, and any commercial interests that brought it to the mass market eventually phased out support because maintining even an XMPP gateway just wasn’t worth the opportunity costs and labour for what amounted to an extra bit of technical debt that hardly anybody even used. It didn’t make money, and it’s only real feature was an anti-feature to interests demanding some kind of lock-in. Strangely enough you probably could use a blockchain to resolve some of the issues I raised, but you would still have to contend with the issue of making money and convincing enough people or coercing enough entities to actually make it into the messenger standard. Why bother when you could just use Slack with its craptastic Electron clients or iMessages or whatever the Google Messenger-of-the-Day is? Or whichever Messenger basically replacesd SMS in the country you live and any supplementary messenger you might need is right there in the App Store? [0] http://harmful.cat-v.org/software/xml/s-exp_vs_XML http://harmful.cat-v.org/software/xml/s-exp_vs_XML

edhelas 8y ago

Happy to see that link on HN :) I've been pretty much pushing XMPP for past decade, especially to show that it can also be used as a really decent base to do social networking (we didn't waited for ActivityPub to do federated social protocols, everything is pretty much there with Pubsub + Atom for more than 15 years). I am working on Movim (https://movim.eu/ https://movim.eu/) to bring the "modern XMPP" in a web platform for all with IM and social features built in :)