Survey for prioritization of requirements for an IM/chat solution for KDE

[Christian Loosli]

Hi Thomas, all, 

First of all: thanks for putting the survey up. 
I think the idea of matching these to protocols in a wiki is also good, and I 
can gladly help with the IRC part. 

As a wiki is more a binary (or trinary, in this case) thing I shall give a bit 
of a longer read on the points here: 

With my KDE hat on: 

The survey is quite what I expected, even though the Telegram like GUI, 
Avatars and Stickers got even less love than I expected <sad fox sticker>. 

>From a protocol side of view, to me it is rather clear that the combination of 
IRC and Matrix will cover the biggest amount of points, so protocol-wise I 
recommend a status quo, using IRC (freenode?) with a decent Matrix bridge. 

Because: 

Both protocols, especially together, fulfil most if not all of the must have 
requirements, from FOSS to free clients, protocol development and type. 
Impact on KDE infra should be minimal, as freenode can gladly continued to be 
used, and whether we want to host our own Matrix server or not is not 
mandatory and to be defined. 

The wide availability of various clients covers most cases, e.g. 
people who want easier file sharing, cross device history (without having to 
use an existing or set up an own bouncer) or a slightly more modern protocol 
can gladly use Matrix.
Those who want a more IRC like client for big / very active channels,
those who want or need a CLI client, those who prefer anonymity  (using Tor, 
being able to use chat without having to provide any personal details or 
registering at all) can use IRC. 
The availability of web clients should also cover availability in most 
countries, universities and the likes, whilst Tor  (little caveat, see below, 
freenode) takes care of the others. 

With this combination I think we can cover the biggest amount of must have, 
inclusion and attractors, and also some of the "nice to have" ones. 

Now of course there is room for improvement. Right now you have to go for some 
compromises if you choose one solution over the other  (important: you have, 
not the other users. So nobody "suffers" because you use either IRC or Matrix) 
If we had an official Matrix client, we would have these features available for 
people whilst not sacrificing integration into the desktop and resource usage. 
If our IRC client would support file sharing (which is technically possible, 
even with drag and drop) and stuff like opt-in image previews, IRC people 
would not miss out on these features either.  So if someone had the time, it 
would be great if we had some improvements in the IRC client  (file sharing, 
image previews, maybe other) and a native Qt/KDE Matrix client (desktop 
integration, resource usage, ...) 

Now with a freenode hat on: 

Still the survey is more or less what I expected. And from a freenode PoV, the 
IRC with matrix bridge solution also sounds good. Whilst Matrix currently is 
not the solution we at freenode are focusing on, we are working closely with 
Matrix devs to improve the current bridge and iron out some of the known 
caveats.  So it's unlikely that the bridge will go away or get worse, it's 
more likely that it will improve  (e.g. stability, being able to deal with 
invexes, removes etc.) 

There are some open points we (freenode) have to work on, and these are known. 
There is e.g. the chicken egg problem of not being able to use Tor without 
registering an account first, as otherwise we'd see even more abuse from Tor. 
Registering an account can't be done via Tor yet. There are also, as per 
above, known issues in the Matrix <> IRC bridge  (which we do not maintain, 
but we have an interest of it improving). We are working on these though. 

*hats off*

tl;dr: 

the status quo of protocols, IRC with Matrix bridge, covers a great amount of 
these points, especially the must have and including ones. 
Client-side there is room for improvement, but that's way easier and better 
than switching around protocols. 


Kind regards, 

Christian