Any caveats, gotchas or hidden complexities to be aware of?
I suppose it'd be better to build a slack clone based on xmpp.
If Slack dies for any of the above reasons, that won't cause people's need for a solution to disappear. I imagine Mattermost would survive just fine. If Mattermost dies, it will be because something came along that's better, similar to how I started out with ICQ, then jumped to MSN Messenger, and now I'm on Whatsapp, WeChat, Kakao Talk, Slack, and Mattermost. Same for Slack. If Slack dies because of better competition, perhaps we'll shift to that away from Mattermost. Who knows?
Gah, how is it possible that the number of chat clients I use keeps growing?
edit: typo
Should be quite trivial to build also, and cheap enough to run as a bit of charity if projects pay for their own storage.
The two I regularly encounter (which shared storage fixes) is either moving to a different device (or trying to remember which ones have which bits of history on) and also integrating new team members.
The latter is especially important if you're doing anything technical/support related; if you've had 2 days of discussions and want to add someone else in, they need the history, not for you to try and cherry pick the important bits of the last few hours to then fail to get them upto speed.
Slack might cut it off, but they gain nothing by doing it.
I sure do hope this is sarcasm. Surely you do know this can go away at any time.
The problem with better chat is that it's better, so you spend more time on it.
There's a lot to like in Discord, and it rapidly became the go to client for all my gaming related friends and activities, but it suffers from the same root pitfalls that Slack always has. It will inevitably run out of VC money (or whatever they're spending now) at some point and start wanting to actually turn a profit.
I see the hosted version is down, though.
Slack recognizes, unfortunately before most of it's customers, that the history is one of the most important assets of a company -- it is the shared institutional knowledge, the thousands of hard-won lessons of all the things that can go wrong that are now fixed, and how they were fixed. This institutional knowledge is not only in the heads of the workers, but also in the papers, and many would need, or at least benefit from, a lookup of the details to re-use that knowledge.
This institutional knowledge is a key competitive advantage of almost every company.
This business model of capturing it and rent-seeking on selling it back to them is a devil's bargain.
"Here's this really slick UI for free, only costs a bit later... nevermind that we'll own the soul of your company and you'll need to pay us tribute in perpetuity..."
It turns out that people don't necessarily want lots of records of everything everywhere.