- From: Dmitri Zagidulin <dzagidulin@gmail.com>
- Date: Thu, 2 May 2024 21:50:00 -0400
- To: aaronngray@gmail.com
- Cc: Lisa Dusseault <lisa@dtinit.org>, Evan Prodromou <evan@prodromou.name>, public-swicg@w3c.org
- Message-ID: <CANnQ-L48Z24oSCoOdcEBHtGnK50GdYzXRf0ztW84P7HuHuJZ5g@mail.gmail.com>
It is, yep! Calendar entry: https://www.w3.org/events/meetings/6b91f267-6cb4-41ca-9a67-5035cee827dd/20240503T130000/ On Thu, May 2, 2024 at 1:02 PM Aaron Gray <aaronngray@gmail.com> wrote: > Is the meeting at the usual place :- https://meet.jit.si/social-web-cg > > > On Thu, 2 May 2024 at 17:45, Lisa Dusseault <lisa@dtinit.org> wrote: > >> If we have time for it, I'd like to kick off a conversation about >> tradeoffs between different approaches in copying content in a data >> transfer. Here's some of the context. And if we don't have time, I plan to >> schedule the next Portability TF call anyway & work on it. >> >> We have significant tradeoffs in portability between two approaches: >> 1. Trying to reproduce a change log >> 2. Trying to reproduce a content repository. >> >> Assume in either of these cases we'd figure out how to transfer likes, >> replies, listens, read etc. The question remains still how we treat >> content objects. >> >> To *reproduce a change log*, we'd specify that the destination AP server >> copies over Create activities holding Note objects - but also Update and >> Delete activities. Consequences: >> >> - Destination servers would have to understand, parse, and know how >> to transform the envelopes around content as well as the content items. >> - I believe this would automatically result in 3rd parties notifying >> their users of "new" posts even though the users had seen them before. >> - It wouldn't be a simple copy of changelog intact. Every changelog >> entry would have to be parsed and transformed for its new context. >> >> To *reproduce a content repository*, we'd specify that the destination >> AP server copies over "Note" objects (and other content objects too of >> course) >> >> - The destination server can now put content objects in an Outbox >> with an activity wrapper if it decides to. >> - It may be possible to put the Note in the outbox without an >> activity wrapper ( I think some implementations do this) - I would love to >> know more from implementers how this works in the wild >> - The destination server can also copy the content somewhere else >> entirely, not just to an outbox. This approach could also more easily be >> used by systems other than AP servers -- it would have greater generality >> than the change-log copy approach. >> >> I'd especially like to know more about the amount of implementation work >> for each of these. It seems at the surface that it's cheaper to transfer >> the outbox, but as I dig into the details of that I don't think it's really >> as simple as it seems to copy the outbox. Many kinds of events in the >> outbox, the destination server would not want to copy over at all, or can >> get in another way that might be easier (the Likes collection, for example, >> on an Actor). >> >> Related, I'd also love to know what implementations do to present the >> "back catalog" of content. Do implementations do anything besides allowing >> friendly servers to page back through outbox history? >> >> Lisa >> >> >> >> On Wed, May 1, 2024 at 8:03 PM Evan Prodromou <evan@prodromou.name> >> wrote: >> >>> I have three agenda items for Friday. >>> >>> 1. I'd like to discuss the upcoming work I'll be doing for Summer of >>> Protocols on implementing E2EE direct messages in AP. >>> >>> 2. I'd like to collect feedback on the Miscellaneous Terms document, >>> solicit implementations, and consider moving it forward as a CG draft. >>> >>> https://swicg.github.io/miscellany/ >>> >>> 3. I'd like to continue the conversation about adding publicKey, >>> publicKeyPem, owner and Key to the Activity Streams 2.0 context document. >>> >>> We have a lot of other projects (Webfinger, HTTP Signatures, >>> Forums/Threaded, Data Portability) to report out on, but I'd love to get >>> these three items covered. >>> >>> Evan >>> >> > > -- > Aaron Gray - @AaronNGray@fosstodon.org > > Independent Open Source Software Engineer, Computer Language Researcher, > Information Theorist, and Computer Scientist. > >
Received on Friday, 3 May 2024 01:50:21 UTC