Gajim - 2024-05-15

The messages I send from Android device with Conversations app, when I open them on Gajim, it shows like this: "This message was encrypted with OMEMO, but not for your device."

I use OMEMO in both devices,

Is there away that I can see all my messages in both devices?

Polarian: gajim never segfaults due to corrupted configs.

Segfaults in general are rather rare and mostly triggered by libraries

Since gajim is python, almost every issue results in a traceback

Sally: open the account in Conversations and change the trust setting under "Other devices" (for Gajim fingerprint)

Hi I'm on Windows. When I try to play a sound from the chat it plays out of my laptop's speakers instead of my Bluetooth headphones. My Bluetooth headphones are the main sound device and YouTube on my web browser plays out of them.

Maybe you can route different sounds to different devices

Check system sounds

lovetox How do I test iq_request_task stuff

you don't like coroutines but use coroutines just done wrong

is there an example of a @iq_request_task'd function being unit tested

lovetox: In the Audio/Video section of preferences all the options are disabled because video calls are not enabled on Windows

So I can't change the settings to route it to a different device

https://xmpp.jix.im/upload/945e6e922a22dd0bd15ac626fae70cc3d7c542d8/UjVmdWpl5Tw3phNVc08Ph5LuaBCkHswziwIjU6Ml/01614104-0617-4f98-8856-7d147a835b76.png

Another issue is that I can't open images on Windows because the path doesn't exist

Maybe this client isn't really very Windows friendly....

Although I'm very new to XMPP and I don't know of any better options

At least this one has a nice UI

treefingers: did you install via Microsoft Store?

treefingers: if you haven't already, try connecting your Bluetooth headphones _before_ starting gajim; also check any *windows* system audio options/settings

Yes

22m5 that's what I did

treefingers: the audio video section in gajim's preferences is not about notification sounds, so there are no relevant settings there.

I'd suggest to check the windows sound routing, but I'm no expert on that

treefingers: the file open issue is known to be a problem of the ms store version. I'll have to investigate

Oh actually, setting the headphones in the Windows volume mixer to Gajim does work

I'm used to it happening automatically, but this is fine

It's okay

Nice application

https://xmpp.jix.im/upload/945e6e922a22dd0bd15ac626fae70cc3d7c542d8/pWdjM0z0IwHZm2L3QdyK045BGiM3uXNsHGEewGuA/0e3e9029-8cfa-4db5-bf63-2ac0d34eba50.png

treefingers: I'm glad that works, thanks for the feedback

I'll check why opening files doesn't work as intended

It's probably the Ms store version

chud, https://dev.gajim.org/gajim/gajim/-/merge_requests/995/diffs#604c873585e1777d16c20ff7606d35fe19081c75

chud, https://dev.gajim.org/gajim/gajim/-/merge_requests/995/diffs#604c873585e1777d16c20ff7606d35fe19081c75_162_222 ✏

in the MR from intelfx, basic JMI and service discovery are already added in a for me acceptable state, so i dont think its a good use of your time to try and work on that.

I think the best way would be, that i can merge intelfx, MR which refactors and fixes probably many little things in the jingle code

then you could simply write your own jingle content module, which lives beside jingle_rtp.py, and is used whenever datachannels are available

Wdym datachannels...

intelfxs stuff implements dtls rtp too although I figure this will still take plenty of time to push changes into gsteamer etc. but then I guess if time were a priority here perhaps my stuff would already have been merged a long time ago.

Anyway I think both me and him want to hear a final decision from you guys to determine which of us should do any further work.

maybe i was misunderstanding this the whole time, i understand that both libs use dtls-srtp, but there is this XEP https://xmpp.org/extensions/xep-0343.html

Signaling WebRTC datachannels in Jingle

and your lib is called libdatachannel, so i thought, it does use these signaled datachannels here

and i thought farstream works without these datachannels

making it 2 libs which do the same, but in a different way

Yeah

Webrtc datachannels can be used for jingle ft (are already used by Conversations). That's part of webrtc as well as the audio/video stuff. Libdatachannel is just the name they picked. The audio/video done with libdatachannel is entirely compatible with Conversations and Dino, and thus by extension would be compatible with farstream patches. So it's the same thing in terms of protocol. Just different approaches.

I have even successfully tested webrtc datachannel file transfer against Conversations but didn't implement an interface.

Crazy idea: what if we had optional trans-workspace "superpinned chats" ? I think I'd like that feature.

chud, hm ok so you are saying they are doing exactly the same

lovetox: Yes... You didn't know both are compatible with conversations etc?...

compatible does not mean, both do the same thing, i can transfer files via direct tcp connection, via an xml stream, via webrtc datachannels

this is a diverse eco system, not all client will support the one and only thing everyone uses

clients use the thing thats available for theire env, language, os, whatever ✏

Converstaions just uses google's webrtc library, it will refuse to work without dtls rtp, Dino will too, and imho that's for the better

if also filetransfer could work in the future with libdatachannel, its probably a good choice. But the work in Gajim is not what library we pass the negotiated parameteres to. if the jingle negotiation works correctly in Gajim, exchanging the lib at a later point is probably a trivial thing. Thats what i was trying to convey the whole time, i dont care about the current transport library of the month. Thats why i want a good abstraction of jingle in Gajim, so when something new becomes hip and cool, i add a single module NewAndCoolTransport, where i implement the exchange of the data. The negotiation layer in jingle will stay the same.

how has that been working out with the broken file transfer and calls?

also, libdatachannel does fewer things than farstream, it doesn't encode/decode audio/video, I do that with gstreamer

are you saying you wrote the first code in history that does not need any maintainance at all anymore after merging?

it will break the same as all other code will break, when things change over the years

I am saying it's not broken, while the code you believe to be easier to be maintained isn't maintained.

I see you want to have another abstraction level but jingle is itself pretty trivial already at least when you're doing calls. You could of course mess around with routing stanzas in fancy ways between callbacks, state machines, additional modules just to do a simple extdisco request etc. But imho it is much more trivial if the code is linear thus easier to reason thus making less need for abstraction. Less abstraction less code less to maintain to begin with.

anyway, I could shove libdatachannel along with encoding/decoding into the unmaintained jingle code somehow I guess. Would still probably be less maintenance in the future bceause libdatachannel is actually used maintained and sponsored by people other than gajim. But up to you. ✏

Someone running Gajim under Voidlinux? Do I need to install further dependencies to get a call-button to call or to get called by Conversations-clients?

goran, calling is not currently supported

chud, i think we need to agree to disagree on the point of architecture. I agree that libdatachannel is a probably a good choice for the future. The MR from intelfx builds on the existing code and tries to improve it, which i feel more comfortable with. I want to keep the current abstraction. My plan is to merge the improvements to the jingle code that are in intelfx code, independent of the fact if farstream patches get merged or not. This means working JMI, and also external service discovery. If you are afterwards still interested to add libdatachannel as an alternative to the (then maybe working) farstream, and are willing to integrate it into the existing architecture, im happy to merge it.

Maybe you get the libdatachannel bindings in the meantime published, python typehints for the bindings would also be a good addition.

he's still talking about this?

chud, you can just create a fork if that's what you want...

lovetox, two years ago I could call a Conversations client.

Who has paid to remove that feature? 😂😂 And how much?

the world moved to new protocols which we are in the processing of getting implemented

nicoco, about the nickname MR

i think this will only work if you send the nickname also in all messages

presence is not gonna cut it

lovetox: I’m not against the idea, so we check if the punycoded-resource matches the nickname on each message? I’ll keep it in presence too since that’s what cheogram uses, but sure, I can add it to every message too.

You did merge the change required for processing the MUC nickname in nbxmpp though 😄

I know you’re in the ‘let’s put everything in every message’ team, I remember you’d like the same thing for hats. It’s non-standard (the XEP is experimental though), but I can put hats in all messages too if you want 😉

lovetox: it would be super cool if that transport icon in room avatar PR makes it to the next release too, it’s a bit sad that we have these icons for 1:1 chats but not MUCs 😕 ✏

if I made a merge request to make the debug log name be in ISO 8601 format, would that be acceptable?

fjklp, show an example please

I don't have one

oh

time format?

nicoco, its important for presence too, there are people in the room which never write a message

the same as sometimes we receive a message without the person beeing present in the room

but you dont need to adapt your MR, i currently refactor the nickname code, so it will be later a lot easier to add this feature

222m5, It's a bit tricky, but finally I found it, thank you so much

Why did gajim automatically consider the trust of omemo keys of my account on other devices as "Not Decided" instead of "Blind Trust"? This was causing my messages to be unreadable on other devices. ✏

either because you disabled blind trust

or because you trusted one of your other devices

I don't know if I ever trusted one in the past but there wasn't one trusted presently

blind trust is enabled

I don't understand why trusting a key would disable blind trust

maybe you distrusted a device

this is what I'm talking about https://en.wikipedia.org/wiki/ISO_8601

but there are options, apparently, to do things differently

so, YYYY-MM-DD or YYYYMMDD

Thh:mm:ss or Thhmmss

if it's used in unambiguous contexts, which I assume ours is:, you can ommit the 'T' ✏

so hh:mm:ss or hhmmss

it already looks like that for me

Really? for me it's DDMMYYYY-hhmmss-debug.log

ah ok you mean the filename

yes

yeah shure why not

but i rather not have ":" in my filename

so simply turning the date around, i guess is what you mean

so its better sortable

That's mostly what I wanted, but there are additional possibilities

2007-04-05T123018−02:00 has the T separator and the UTC offset

- as separator is probably easier to read

2007-04-05-123018

2024-05-15-225115-debug.log

is that alright for you?

yeah

ok i push this myself

oh, nice :)

lovetox pushed 6 commits to branch gajim/master https://conference.gajim.org:5281/pastebin/1d13fdff-6cd6-40da-9c66-58da6050b34c

thanks

wurstsalat pushed 1 commits to branch gajim/master other: Set correct pyGObject stub config for dev_env script - https://dev.gajim.org/gajim/gajim/-/commit/31820540d789353243c044f44b1d56dd6a194b6d

“17:55:36 chud> Converstaions just uses google's webrtc library, it will refuse to work without dtls rtp, Dino will too, and imho that's for the better”, that’s wrong about Dino, it uses Gstreamer not libwebrtc, and it also implements unencrypted calls for compatibility with e.g. Gajim or Pidgin or Psi or the PSTN.

nicoco, how should mentioning someone work for such a contact that supplies a nickname?

do i need to mention the punycode name or the one from the nickname tag

probably the one supplied from the nickname, as this is how they would see it on the other side of the bridge i guess

chud, intelfx, do either of you already implement screen sharing in your MRs? Do you support selecting PipeWire camera and/or audio streams? What about through a portal?