radical proposal: move IRC to Rocket.Chat

[Christian Loosli]

Am Dienstag, 8. August 2017, 20:17:08 CEST schrieb Cristian Baldi:
> Hey there,

Hello hello, 

> [Various Issues I agree with]

> Rocket.Chat does not have an official mobile client as of today, again
> Ruquola could solve this once it is compiled for Android. Right now the
> official way to use Rocket.Chat on mobile is to use some kind of wrapped
> WebView which does not work well (when I had that installed I did not
> receive notifications or received them randomly).

Same goes for slack and mattermost, and these things are horrible. 
First of all: they are massive battery and memory hogs. 

Same goes for the electron based wrappers that are sometimes used on the 
desktop. 

Also they don't integrate UX wise.

> As Jonathan said Rocket.Chat (but really, any modern messaging system)
> offers tons of features missing from IRC.

Out of interest: what exactly does IRC lack? There are 4 things coming to mind 
for me, all of them with my personal opinion: 

- Lack of emojis and stickers: whilst I think it's great that I can send 
stickers of kitties hugging each other on Telegram, I hardly see a need for 
that in a more "professional" environment. Emojis are UTF-8 and thus 
technically work on IRC and clients can handle them, if they want. 

- Lack of backlog / disconnecting when offline:  Not only a solvable issue 
with things like IRCCloud or bouncers, but actually a solved issue for KDE, 
given there is an official ZNC instance available. 

-  Lack of support for code:  Yes. There I like the old UNIX philosophy of 
"one tool for one purpose". We do have phabricator, which handles diff and 
comments on code (including highlighting) much better, so why place that all 
into a client, which will only make it way less lightweight? 

- Lack of a file drop: Yes, that might be an issue. However, putting that in 
the protocol not only creates more of a burden for infra and people (more on 
that below), technically it can be done in IRC. KMail just recently got a 
feature which puts files too large for e-mail on a file drop  (e.g. owncloud, 
dropbox or whatnot). It would not be too hard to implement that as a feature 
in IRC clients, so it uploads the file to $whatever  (e.g. imgur for images, 
dropbox for documents, ...) and place an URL in chat. 

> A few months ago we also tried Mattermost (similar to Rocket.Chat but it
> seems to have gotten much better).

Mattermost is actually more of slack than rocket (see my other mail), as it 
tries to (and mostly is) compatible with Slack. I recommended it already as a 
FOSS alternative to Slack. 

> I would suggest investigating all the alternatives and going with the one
> that works and feels better, offering the best native experience and having
> the most stable core.

Always also keep in mind what impact it would have on

- The infra. If we have an application that allows files and (endless) 
backlog, consider that this uses memory, disk space and bandwith.

- The community. As you write correctly, it is hard to migrate people over. So 
I do prefer protocols that can be linked, at least. 

- People with less access to decent hardware and bandwith, since KDE 
collaborators work all over the world, not everywhere you have decent up- and 
downloads and devices with endless amounts of RAM and / or huge batteries. 
IRC is super lightweight, both clients (you can just ssh to a box with irssi 
on it, you don't even need a GUI) and servers, both bandwith and memory/cpu 
consumption. 

> Cristian

Kind regards, 

Christian (Fuchs on freenode)

> 
> On Tue, Aug 8, 2017 at 7:08 PM, Luca Beltrame <lbeltrame at kde.org> wrote:
> > Il giorno Tue, 08 Aug 2017 18:16:17 +0200
> > Luigi Toscano <luigi.toscano at tiscali.it>
> > 
> > ha scritto:
> > > So -1 for moving to Rocket.Chat.
> > 
> > -1 as well. As Luigi said, matrix.org is a better replacement because
> > the bridge is already up there. Also, it is federated, and FOSS.
> > 
> > --
> > Luca Beltrame - KDE Forums team
> > KDE Science supporter
> > GPG key ID: 6E1A4E79