Pidgin 3.0.0 Experimental 1 Announcement | Hacker News Reader

Pidgin 3.0.0 Experimental 1 Announcement

discourse.imfreedom.org

93 pointsrw_grim47 comments

Am I reading this right? Is the only protocol being supported now the most recent version of IRC? Are there plans for others? Part of Pidgin's "calling card" has been its ability to support many different chat protocols. However, that's become difficult since most protocols have become walled gardens and harder to reverse engineer. I'd imagine supporting Matrix, XMPP, Signal, other protocols with open specifications, etc. would be a good idea

I'm guessing at the time of this experimental alpha "don't use it full time but alongside pidgin 2.x to test things, all plugins are incompatible", only IRCv3 is supported.

yes all the other protocols will eventually be supported. we're a small team of open source developers and these things take time..

Not sure if my memory deceives me, but did not Pidgin support XMPP?

I suppose for now I may stick to Gajim (and Conversations on Android).

Pidgin supports lots of protocols, including xmpp, via plugins. It just sounds like they choose to make breaking changes to the plugin API for v3. I'd expect popular protocols to get ported at some point before GA.

Yes this 100%. As the post says, nearly all of the API in libpurple and pidgin has changed. So we have to rewrite all of the protocols yet.

Sure it did. Our company chat was XMPP 10 years ago and pidgin was one of the most widely used clients. I assume it's just the experimental version that doesn't support it yet.

Nowadays we use zulip at work which has a better model (topics). For IRC I use quassel because I can have the backend running on a server and when I connect using the frontend I see the channel history and messages I might have received weeks ago...

The last time I followed Pidgin development was a while ago, but I'm actively using it for various protocols (slack, discord, googlechat, skype), so I'm curious about the decision to break the API.

I understand that the Pidgin community is small, so claiming that "the other protocols will eventually be supported" seems unrealistic. This change requires the community to take on unnecessary porting work and we are already dealing with constant protocol changes, playing whack-a-mole with protocol providers.

For instance, this guy [1] is doing incredible work, but I doubt he will want to invest his time in unnecessary rewrites unless someone steps in to assist with each plugin.

[1] https://github.com/EionRobb/purple-discord.git

Presumably because the old API was broken/lacking/etc. Calling the resulting porting work "unnecessary" without even considering why the API might have been changed is not fair.

The pidgin project is 25 years old, and libpurple is 17 years old. It's entirely fair that a rework is required, and much appreciated that this work is continuing.

So to put this very simply, there is no way in pidgin 2 without breaking API to address a message after it's been displayed. Eion has been hacking around this for years via commands in protocols and then replaying messages and stuff. While this works, it's not a great user experience.

So at a bare minimum we needed to totally change the way messages work. Which inside of a chat client, is a pretty big change. That obviously led to other changes and so on and so on.

As for the other protocols... We needed something to prove out most of the abstractions and trying to prove out abstractions while implementing multiple protocols is a ton of work. But on our radar are new XMPP, Bonjour, and even a Matrix plugin. All of these will be coming from us and will be in tree.

That said, there will not be any proprietary protocols in our official source tree. If you want more information on that, check out this post [1].

As for Eion's plugins, I've been talking to him through much of this as you would expect. Obviously his time is precious and we didn't make this changes to spite him or anything, but we (me specifically) have offered to start porting his protocols for him.

[1] https://dev.to/grim/in-tree-protocols-15ne

Thank you for sharing the details!

Would be good for Matrix to have more stable client than Element or Flutter based apps.

pidgin 3 looks very exciting. it’s worth noting that Element X already provides an spectacularly more stable Matrix client than classic Element.

Signal is open source but they only let the official builds connect to the official Signal servers

It's nice to see pidgin still alive and kicking!

I hate to have different apps for similar stuff and I also prefer "native" clients over web UIs so I used pidgin a lot in my life. I impressed many people with its "telepathic" feature that would open a chat when someone starts typing, to let you type in something before they even finish their initial sentence. Fun times…

However, when I finally got a smartphone, and started a more conventional line of work which came with extra devices, I switched to grouping my chats using server-side gateways for better multi-device consistency. I started by using Spectrum [1] which leverages on libpurple (pidgin's "backend") to do that. Since spectrum is on life support (maintenance mode, no modern chat features like emoji reactions which I happen to like despite my old age), I actually started my own hobby project [2], because I'm not really interested in learning C++ (which spectrum is written in). (shameless plug, I know)

The hype these days is more around mautrix [3], but the permacomputing enthusiasth in me prefers XMPP over Matrix. I'm not that religious about it though, my gateway project includes the most feature-rich XMPP/Matrix gateway out there [4] which I try to improve and maintain when time and motivation allow it. Unfortunately, XMPP support in pidgin has historically been pretty poor and IMHO partly responsible for some of the hate XMPP gets. I'm not blaming the pidgin devs of course, I should contribute to libpurple-xmpp (if that's how it's called) instead of complaining. ^^

[1] https://spectrum.im/ [2] https://sr.ht/~nicoco/slidge [3] https://github.com/mautrix/ [4] https://git.sr.ht/~nicoco/matridge

what is more permacomputingy about xmpp than matrix, ooi?

Resource usage in general, especially server side? Although XMPP clients also seem to win too, cf [1] (debatable methodology in this link, I agree).

And this is not "implementation detail stuff", by design serving Matrix is resource-hungry, while serving XMPP is very lightweight. Others have phrased it better than me, something along the lines of:

> Matrix is a distributed, eventually consistent database; XMPP is just message passing.

Both have their own merits.

But you know all of that Arathorn ;). I'm replying for readers anyway.

[1] https://decentim.grafana.net/public-dashboards/92602d3a4aa84...

Pidgin 3.x is using a proper secret store: https://developer.pidgin.im/wiki/KeyringSupport.html

Yes! This is great news.

Yep, libsercret/kwallet on linux, wincred on windows, and keychain access on mac.

I had no idea people were still working on this. I remember it as Gaim in the 1990s and one of the developers being particularly nasty to me in 2000 after I submitted a patch. I worked hard on that and he was a complete asshole.

Since then I've used that experience to exemplify who I never want to become and how I never want to treat people.

So thanks. Bad experiences can help you become a better person.

PS: shoutout to Daniel Reed (rpidan) who did an Ncurses naim (https://en.wikipedia.org/wiki/Naim_(software) // https://github.com/jwise/naim) from back then. He was a genuine upstanding guy. Naim in long running screen sessions that I'd log into from text terminals at the library to chat with friends over AIM - good times.

Also, meebo... Basically libpurple over 2005 era AJAX. I even made a super minimal HTTP forms based client I could use over flip phones in the early 2000s. It was optimized for T9N and per/KB metered data plans. Never really publicized it. Probably still have the code somewhere.

Well fwiw, all of those people are no longer involved in the project.

Yes, there was some drama in the early 2000s where a bunch of bad vibe people got forcefully shaken out.

I did some digging and have been able to find basically everything. I'm keeping it to myself. No dirty laundry here.

Also if pidgin could magically combine slack, messenger, telegram, discord, and Whatsapp in a single reasonable app ... It'd be a return to the glory days. The current state of half a dozen web apps is such a pain. Just text chat. Nothing fancy.

What a dream...

In the meantime, you can try Beeper which does that through Matrix. It connects all the apps you mentioned in a single app.

Amazing. Will try it tomorrow

Or self hosted matrix with bridges, which is basically what's underneath beeper.

I use this ansible role to deploy it: https://github.com/spantaleev/matrix-docker-ansible-deploy/

Had not heard of this. Got it set up relatively smoothly. Great stuff, thanks!

Beeper seems to be working ok for me for the moment

there's basic text support for most of those in pidgin 2 right now https://pidgin.im/plugins/

Blast from the past; I was maintainer for a couple of releases of funpidgin; a fork started over an argument about if there should be an option to make the message input resizable https://supine.com/posts/2008/04/forking-for-funpidgin-and-p... https://m.slashdot.org/story/100702

My pride and joy was the Adium theme support (which was resurrected from another rejected patch) https://www.adiumxtras.com/index.php?a=search&cat_id=5&sort=...

AJAX... that takes me back! I remember when it started becoming trendy, you did not have to reload the page for updates! It was magical.

Fwiw, meebo was based on libpurple. They employed Mark Doliner, aka Kingant, who was also a Pidgin developer and eventual lead developer.

Meebo was a lifesaver in college where I could log in via school computers and talk to my friends before my parents got me a cellphone lol.

It was a long long time again.

Good memories.

pidgin and adium were my dream apps for a long time.

I wish this project godspeed. I used Pidgin on and off over the years. Last time I gave it serious use was around 6-8 years ago, when I created a full Lync / Skype for Business setup for Linux with it. I could even do screen shares and everything. It was a blessing because Linux was still a BYOD thing in my company, no support at all. And of course the stack was Microsoft based. Office already had competent enough web clients, but Teams wasn't on the horizon yet (for our company), and so, I had to keep around a Windows VM for S4B and other random stuff. That is, until I tinkered with Pidgin enough to set it up as a client. I really appreciated that it enabled me to do that, and I wish the best with the current efforts as well. It would be really, really funny to send iMessages with it one day.

Timing context: Debian sid currently distributes 2.14.13, which is the 2024-02-13 bugfix release of the 2020-06-10 minor release 2.14.0 which was already clearly specified to be the last of v2.

Is there a chance you will support Signal protocol in the future?!

Maybe, as others have noted here, signal doesn't allow third party clients on their networks... That said, neither did AIM, MSN, Yahoo, and ICQ...

My most vivid memory of using pidgin on a daily basis is inability to copy-paste using ctrl+c, cuz someone decided to bind this hotkey for something else. Luckily this was many years ago. Textbook example of what happens when developers design UX/UI.

While there have been clipboard bugs, we've never used those keystrokes in the UI. However, GTK2 did have a way to let users change the menu shortcuts by hovering over the menu item and hitting a new keystroke and I suspect that's what happened here.

is this good news?

I hope so? :-D

For me it's bittersweet news.

On one hand, it's great Pidgin is still alive and thriving like this. I have plenty of memories using it to talking with friends, people from uni, crushes, etc. And it had tons of cool funny plugins - I used to use one that let you change your profile pic from several from a given directory, sort of like how a "wallpaper slideshow" thing works. Do try to do that now with Whatsapp/Telegram/Signal/Slack/whatever...

On the other hand, the thing I always complain about - the state od IM right now is absolutely miserable thanks to all those walled gardens that force you to use their shitty client "apps" because of "privacy" or "features" or whatever, so you end having to use a bunch of different "apps" to talk with different people. It's insane.

But we're trying to fix the thing you're always complaining about...?

I cleaned up the language on the irc bit in the post as it was confusing people. But we are very much aiming to support everything including all the big chat networks. That's why so much of our internal API had to change.

This release is to get something out there even with limited functionality rather than not releasing until it's at feature parity with Pidgin 2.x which is going to take a long time on our current trajectory.

Reply on news.ycombinator.com