News for hackers. Stuff that matters.
Re:Making IRC better by slightwinder on Wednesday July 06, 2022 @09:56AM (#32000953)
> but why does it have to be IRC?

Because IRC works. It's open, battle tested and has huge support by an endless amount of tools and documentation. There is not much reason to not use if all you wanna do is chat.

> I like being able to paste code into chat and have it syntax highlighted

> I like being able to ping people so that they get a notification on their phone

> Occasionally, I like to send a GIF.

That's the job of your client, not the communication-protocol.

Re:Making IRC better by IshKebab on Wednesday July 06, 2022 @04:17PM (#32006243)
IRC works... if all you want to do is send basic text messages back and forth like it's still 1995.

That's not what I want. I want modern features like text formatting, links, images, file sharing (don't mention DCC), threads, linking to messages, persistent history, and even emoji reactions.

And when I say "modern" I mean "modern relative 1995 which is where IRC proponents are still stuck".

Re:Making IRC better by qu4z-2 on Wednesday July 06, 2022 @10:07PM (#32009187)
Man, I use Mattermost at work and I don't like any of those things. I've managed to hide all the emoji with a CSS rule, inline images are at least collapsed by default, and thankfully no-one is using it for file sharing.

It allows you to disable threads in your UI but that's useless since then you can't reply to people by posting after them since for them it goes to main chat instead of the thread (instead of, y'know, just being the next message).

Also IRC definitely supports links, haha.

Re:Making IRC better by IshKebab on Thursday July 07, 2022 @06:31AM (#32012201)
> I don't like any of those things.

Why though?

> Also IRC definitely supports links, haha.

Oh how do you set the text of a link? Do IRC clients support markdown syntax these days?

Re:Making IRC better by Gigachad on Wednesday July 06, 2022 @09:59PM (#32009136)
Irc users like being stuck in 1995. What they don’t like is that no one else agrees with this.
Re:Making IRC better by sdfhdhjdw3 on Wednesday July 06, 2022 @09:59AM (#32001006)
> Because IRC works. It's open, battle tested and has huge support by an endless amount of tools and documentation. There is not much reason to not use if all you wanna do is chat.

How about encryption?

Re:Making IRC better by dijit on Wednesday July 06, 2022 @10:01AM (#32001043)
TLS is accepted, since you implicitly trust the server when using chatrooms in most cases.

for everything else there are plugins in your client for OMEMO (what signal is based on) and OTR (which is purely session based).

Handling this as a third-party implementation for E2EE is probably the only true way to gain trust anyway. If your provider provides the infrastructure and the client then how can you really trust it?

Re:Making IRC better by rdpintqogeogsaa on Wednesday July 06, 2022 @10:26AM (#32001491)
Yet there are also large networks like QuakeNet or UnderNet with no TLS anywhere.
Re:Making IRC better by dijit on Wednesday July 06, 2022 @10:31AM (#32001575)
And you can choose not to be there.

My network forces TLS: https://darkscience.net

Forcing people to "do what you want" is against the spirit of open standards.

Re:Making IRC better by sdfhdhjdw3 on Wednesday July 06, 2022 @10:03AM (#32001086)
I'm not a security expert, but my humble view is that if you need plugins for really basic functionality, that's a red flag.
Re:Making IRC better by dijit on Wednesday July 06, 2022 @10:06AM (#32001128)
"E2EE" is not really basic functionality.

I believe it is hard to get right, and only worthwhile on 1:1 communication, it doesn't work well at all on large chat rooms, which is what IRC is.

Regardless, if you don't want E2EE to be handled by a plugin- it is possible to make a client with it baked in.

There is no company that will stop you creating a third-party client unlike slack/discord which are extremely hostile to those endeavors by comparison.

EDIT: If you downvote without replying I'm just going to assume you're ideologically opposed to open standards, because I'm not sure what else I can take from it.

Re:Making IRC better by sdfhdhjdw3 on Wednesday July 06, 2022 @10:50AM (#32001857)
I didn't downvote, and I'm very pro open standards. I just think we can do better than IRC, that's all.
Re:Making IRC better by dijit on Wednesday July 06, 2022 @10:56AM (#32001935)
Ah, I wasn't accusing you of downvoting, just the community.

You can't downvote direct replies.

---

Speaking about standards; have you looked at IRCv3? Is there anything specific that you feel is not being addressed?

Encryption (to my mind) is better handled outside of the spec itself, just like HTTP vs TLS wrapped HTTP (HTTPS) where the "TLS" has no bearing at all on how HTTP is implemented.

IRCv3 is attempting to address the persistence issues, though many people like the lack of persistence in general.

I don't want to guess at what your concerns are, but more generally; have you looked at the spec?

Re:Making IRC better by numpad0 on Wednesday July 06, 2022 @11:50AM (#32002727)
It's not really a green flag when values offered by major commercial alternatives seems like then-experimental client features of existing thing(cf. Twitter, Slack, Discord)
Re:Making IRC better by ThatGeoGuy on Wednesday July 06, 2022 @08:16PM (#32008524)
There are OTR / OMEMO implementations out there for private queries. For rooms, I'm not convinced that we need full Olm / Megolm style encryption in most wider chats - encrypting messages across 200 individual parties with a magnitude more individual clients is perhaps just yak-shaving.

I even learned today that irssi has an OTR module by default! On Debian systems you have to install irssi-plugin-otr but once you have that and irssi you can just `/load otr` and `/otr init` in a query and that's it, you have E2EE.

This is handled client side though, so I suppose if your clients cannot agree on a protocol you'll be left with nothing. I'm not sure that's really the responsibility of the server though, since we're talking E2E after all.

Re:Making IRC better by astrobe_ on Wednesday July 06, 2022 @12:47PM (#32003525)
When a platform does not guarantees secrecy you just avoid holding private conversations on it. In my opinion, that's the spirit of IRC, a public place to discuss, because often channels are logged and the logs are accessible for anyone from the web. If not, just about anyone on a channel can do it without your consent. That's also true I think with other "secure" chat platforms because that's the nature of any-to-any chat protocols.

Then you use private messages as a way to talk to someone without disturbing, or away from the noise, of the channel - still not for really private conversations ; for this you don't need encryption, you need end-to-end encryption, but at this point you'd rather use a messaging platform rather than a chat platform.

Re:Making IRC better by ddevault on Wednesday July 06, 2022 @10:04AM (#32001112)
If IRC were designed today, it would probably be with E2EE. But it's not the end of the world that it wasn't, it just limits what use-cases it's appropriate for. Asking about kernel config options in #linux does not really demand particularly high privacy, and TLS between you and the server is generally sufficient. E2EE is also pretty difficult to get right and dramatically increases the complexity of the system -- Matrix's third-party client ecosystem is hamstrung by this requirement and very poor compared to IRC as a result. E2EE is nice to have and essential for some use-cases, but I don't think that IRC is invalid because it lacks it.
Re:Making IRC better by sdfhdhjdw3 on Thursday July 07, 2022 @02:33PM (#32017647)
So I tried Matrix yesterday for the first time in my life, despite having heard about it in passing for years on end.

It's fucking slow.

Re:Making IRC better by Idesmi on Friday July 08, 2022 @06:49AM (#32025404)
The matrix.org server is slow.
Re:Making IRC better by 300bps on Wednesday July 06, 2022 @10:19AM (#32001375)
I haven't heard of a widely-used IRC client that doesn't have built-in SSL connectivity for quite some time. For example:

https://www.mirc.com/ssl.html

Re:Making IRC better by usrn on Wednesday July 06, 2022 @01:22PM (#32004031)
Encryption makes more sense for small or one on one chats with people you know, the sorts of things people use XMPP for (which has OMEMO: signal style encryption.) In large public chat rooms being able to use a pseudonym is much more important than encryption.