XSF Discussion - 2021-10-22


  1. ralphm

    https://fosdem.org/2022/news/2021-10-22-fosdem-online-2022/

  2. Zash

    So no FOSDEM this year either :-(

  3. flow

    given that some european countries show an increased incidence rate, it is probably the right decission

  4. flow

    will sure to fun attempting to recognize people again at fosdem 2023

  5. emus

    But do we want to participate anyway?

  6. MattJ

    Sure. Hopefully better than we participated last year.

  7. vanitasvitae

    flow: hehe, at that point the "beard percentage" surely has gone up :D

  8. deuill

    Not sure if this is off-topic here, but I've been reading through the blurb for the Element/Matrix Managed Bridges (EMS) service, and have been wondering if they somehow ensure/protect users from being kicked off third-party platforms for ToS violations.

  9. Holger

    I've been wondering the same, but it's certainly off-topic in here, yes 🙂

  10. deuill

    There's nothing specific written up, but they do mention the EU's Digital Markets Act as being a sort of change in what was traditionally a rather risky endeavour. Do we know if, legally, bridges/gateways are more viable these days?

  11. deuill

    I guess I'm mostly wondering in terms of whether this might lead to fresh efforts on the XMPP side -- I get the feeling these haven't been viable exactly because third parties tend to be rather beligerent to users that access their platforms via third-party means.

  12. deuill

    (There probably isn't a great answer to this until someone gets banned for a ToS violation and raises this with the relevant EU authorities)

  13. Zash

    Rant on the topic of bridges: https://mail.gnome.org/archives/desktop-devel-list/2017-September/msg00047.html

  14. Sam

    This is a good explanation; thanks. Someone recently asked me if I would be interested in splitting up a client into a local client/server archetecture over D-Bus and I was having trouble figuring out how to explain why I thought this was a bad idea. This gives a good example.

  15. Kev

    Microservices are good except when they aren't? :)

  16. Zash

    True fact.

  17. deuill

    Oh man I remember reading this, really good write-up. Though AFAIR, this was mostly aimed at issues with abstracting against protocols at the UX level; this is hard because you essentially have to build out a common denominator that covers an unknown and moving surface area.

  18. Sam

    Microservices are fine when they're running on a backend with a team to develop each one and lots of money to throw at coming up with the protocol that they all speak to eachother with. Not so much when you're trying to run something on someones desktop.

  19. Kev

    I'm certainly not a microservices expert, but it seems to me they provide value when they ... well, provide value. e.g. Language disconnect, horizontal scaling, team independence, etc.

  20. deuill

    That is to say, the Telepathy D-BUS API becomes its own protocol, essentially, complete with opinions on how the UX itself is presented.

  21. Sam

    Exactly; things you don't need in a small desktop application :)

  22. Zash

    Sam, microservices are fine when you want to build a mirror of your organizational structure? 🙂

  23. Kev

    And however much value they provide, they *always* have a non-trivial cost.

  24. Sam

    This ⤴️

  25. Kev

    (I'm actually playing with such things and trying to benchmark to see if it's viable within an XMPP service without breaking everything)

  26. deuill

    This is perhaps different to bridges/gateways which attempt to match from one fairly stable set of semantics (say, XMPP), to a perhaps moving set of separate semantics (whatever third parties are involved).

  27. Zash

    deuill, but you were thinking more of the legal ramifications of bridges, rather than the technical and UX?

  28. Kev

    deuill - yes, I think you're right, that is mostly about the telepathy model, rather than gateways in general.

  29. deuill

    Yeah, whether bridges are more viable these days simply because platform providers cannot as easily boot people off for using them (and may be expected to provide interop APIs).

  30. Kev

    Or, well, bridges I guess. Gateways are dreadful to write, and bridges are worse.

  31. Zash

    Kev, and you know the difference between those???

  32. Sam

    I was about to ask…

  33. Kev

    I'm probably misusing terminology.

  34. Zash

    The terminology is confusing and overlapping, even before Matrix came along and invented their own set of terms for everything.

  35. Kev

    Thinking of IRC gateways, I tend to think of the traditional model (you join a magic room on the gateway) as gatewaying, and what M-Link does (you join a normal XMPP MUC, or normal IRC channel, and just appear in both) as gatewaying (although we still call it an IRC gateway). I think I might have invented the distinction, though.

  36. Kev

    Thinking of IRC gateways, I tend to think of the traditional model (you join a magic room on the gateway) as gatewaying, and what M-Link does (you join a normal XMPP MUC, or normal IRC channel, and just appear in both) as bridging (although we still call it an IRC gateway). I think I might have invented the distinction, though.