Gajim - 2024-11-26

jstein, also attempted to bump gajim in Gentoo right now and ran into the issue :)

Hello, I have a question. Gajim has a feature called "Note to myself". Is the conversation entirely local or stored on the server?

Another question, why can I enable OMEMO on that "Note to myself"? Does OMEMO support encrypting to yourself? Why can't OMEMO work on a MUC with a single person then?

Feel free to correct me if my observations/understandings are wrong, I'm just a newbie 😅

reso: the latter and unencrypted if you use omemo otherwise.

Hmm ok, then if it's stored on the server, that means that "I'm chatting with myself". But I tried to add myself as a contact but Gajim didn't let me do that.

So then how is it possible to have a chat with yourself when adding yourself as a contact is impossible?

Am I correct here with my observations?

> I'm just a newbie 😅 Welcome! :) Sorry, I'm busy, someone else needs to continue ;)

reso, Adding a contact also goes via the server, which might reject it because otherwise you'd be subscribed to your own presence, which is starting to get a bit too meta... But you can send messages to people (or, more accurately Jids) which are not in your roster (ie, a "contact").

I see

So I take it that one should not use the "Note to myself" feature because it is unencrypted and can be read by the server admins?

reso, Gajim might encrypt a "note to myself" via OMEMO; there's no reason it couldn't. And equally, a MUC containing just one person would work OK with OMEMO, but be pretty pointless, since that one person would have to be the sender. Encrypting to multiple people isn't handled by OMEMO; it's a 1:1 cryptoosystem (like most of them). To encrypt to a group, you either need to send a symmetric key encrypted to each occupant in turn - and this gets wildly inefficient very quickly - or you use a groupwise cryptosystem like MLS (and possibly nothing else).

Ah, I see

reso, If you can't trust your own server admin, you're sunk anyway. A server admin could just replace your advertised keys on the server, and decrypt your messages, re-encrypting them to the keys you think you have after reading them.

reso, I mean, unless you formally validate your keys with each person you talk to, anyway.

reso, This is unlikely to be the case, mind. Most commercial server admins would want your messages to be encrypted, since it avoids a bunch of legal requirements, like subpoenas - this being why most commercial consumer IM systems now do end-to-end encryption.

Thanks for the info, dwd.

reso: note to yourself can be encrypted but you need at least a second device on your account

> reso: note to yourself can be encrypted but you need at least a second device on your account Oh, got it ↺

Thanks for the info 😁

Hi there, I can't add conversations.im account in Gajim. It shows error "not well formed (invalid token). Does anyone have an idea of this? Thanks

Hi, I'm trying to run last dev (master) version of Gajim but it fail at building venv with a "ValueError: Unknown library Gsk" error, what is Gsk? Is it a new dependency I forgot? Thanks

Codimp: you'll need gtk4

I already have it

Which version?

"gtk4 1:4.16.7-1" from Archlinux extra repository

I tried to make some research but can't find anything relevant about this Gsk thing missing ✏

ctrl+n got much faster on the gtk4 branch lately, thanks for thet :)

> jstein, also attempted to bump gajim in Gentoo right now and ran into the issue :) Thank you for trying and reporting. I can understand that it is also a lot of work to setup a new/"real" build system, but it seems the only clean solution. ↺

even if somebody figures out how to get it built one time with many workarounds it will break all our scripts on distribution side an will break soon again and soon another dev will have to reverse engineer how all these workarounds need to be fixed

jstein, you are exagerating greatly, as you see in the arch file its like 4 lines to build and install gajim

i checked meson, and i found no particular features that supports what we need to do, i will not use a build system like meson just to shell out to our own scripts

which feature did you miss in meson?

> jstein, you are exagerating greatly, as you see in the arch file its like 4 lines to build and install gajim This. It's just that's gentoo's packaging is overcomplicated. ↺

meson to build python package?

isn't a bit overkill?

im not getting into a discussion if meson can do something or cant, im sure it can do anything, make.py is 200 lines of code of which 1/4 is boilerplate. meson is buildsystem for languages like C where you compile a lot of stuff. What we need is one source file modified, gettext called to build translations and copy 5 files to a location of choice

not sure what you think would happen if we use meson, you dont need to read the README anymore and simply type in the same 3 meson commands for any package?

I should choose a different nick :-P

> I should choose a different nick :-P freemason? ↺

> which feature did you miss in meson? meson currently cannot specify required python module versions https://github.com/mesonbuild/meson/issues/3551 ↺

> hyol: sounds like a very specific use case of yours. Not necessarily applicable to other users cal0pteryx: I get that we might not want to stretch resources too thin by overextending, but I think there's really no reason that xmpp should not be considered as a better alternative to email, unless your argument is "nobody uses xmpp".

deport: you've got to set priorities. Gajim is a chat application and I'd like to improve it on that.

deport: it's not that big of a deal. When composing a very long message, you can type it in your text editor and then copy paste it

jstein, starting with https://github.com/gentoo/gentoo/blob/master/net-im/gajim/gajim-1.9.3.ebuild I assume one only has to copy the PKGBUILD changes from https://gitlab.archlinux.org/archlinux/packaging/packages/gajim/-/commit/ff670fcc620442b1ddd0955a1e62bd5a7f21e97b

> deport: you've got to set priorities. Gajim is a chat application and I'd like to improve it on that. I get that, its true. I guess I just saw this desired feature of a taller text composition box as being easy to implement, probably not conflicting with anything else, with a potential big gain to those who use it.

Maybe I underestimate cumulative complexity increases from many "small" changes :)

> I should choose a different nick :-P 😂 ↺