4 ms·
I've been running Matrix since ~2017. Nothing has compared. It's been frustrating as cycles have crested, and crashed. Riot was incredible until it wasn't. The
by COGlory 3y ago
I've been running Matrix since ~2017. Nothing has compared.
It's been frustrating as cycles have crested, and crashed. Riot was incredible until it wasn't. Then RiotX/Element was incredible... until it wasn't. And now we're on to Element X. What I will say is that you have all seemed to make the right decisions ultimately, even if they are painful. It's easy to see why Riot had to be abandoned - it was a proof of concept. And it's easy to see why Element needed to be sunsetted. It was designed to solve an immediate problem - not scale to the point Matrix has. You've clearly learned your lessons about future-proofing and scope, and while I think you may have misjudged the needed scope with Element (albeit only slightly, it worked well enough at first) and while Element X has been paaainfully slow to arrive (please for the love of god add a system share menu integration), it would seem that's because you're doing it in a thorough, sustainable, future-proof way.
While it's clear that Element X is very close, and I 100% believe it will get there and be an absolutely killer app very soon, I'm now more concerned about the server side. Switching to OIDC is about to add a ton of friction. Sliding sync is even more friction. There's also the state compressor just out there in the wild that frankly needs to be built in. There needs to be a seamless Synapse update that just does those things, and it needs to go off without a hitch. If you can't swing that (and if you can't, I beg you to reconsider), there needs to be dead tight documentation of exactly what I need to do, coming from any possible configuration - binaries, docker, etc. Just when I'm finally putting away my list of caveats with Element, we're getting them with Synapse. I understand there may be engineering reasons why this can't be seamless, but from a product standpoint... it NEEDS to be seamless. Matrix does not need any more controversy or setback.
Otherwise, I just want to thank and congratulate all the folks at Matrix and Element. It's an incredible product, and it's so many different things at once. It's easily my favorite open source project, and it's one of just two that I donate to. Now, more than ever is the time to buckle down. Execute perfectly or not at all, and I'm confident Matrix will emerge on the other side so much stronger than it's ever been, and this rough patch will all be in the rearview mirror, and things will go back to normal.
- Arathorn 3y agoThe whole cyclical thing is interesting - I wonder if the previous generation app actually regresses while the next gen is coming together, or whether it's just perceptual that the next-gen is so much better than the previous one, the previous one feels stale? It's not like we've been deliberately trying for cresting & crashing cycles, so I wonder what the explanation really is. (Separately, it's obviously a questionable choice that we've rewritten the Android app 3 times now: once from old-school Android 2 (Riot) to using Kotlin+Realm (RiotX/Element), to now using rust-sdk (Element X)). About the only instance I can think of where an app has concretely regressed are the stuck notification problems on Element Web over the last year (caused by trying to fix thread notification reliability - but causing way more problems in the process, which have then been a nightmare to hunt down efficiently). Totally agreed that the transition to supporting Matrix 2.0 needs to be easy and seamless, and we'll do our best. Thanks for holding the faith...