[00:00] <arraybolt3> So, the solution... is probably to delete all known WiFi connections if there's a state transition from "Connecting" to "Not Connected" on a WiFi device.
[00:00] <arraybolt3> And probably change the connection status label to "Connection failed - bad password?".
[00:01] <arraybolt3> So I'm going to try and implement that and we'll see what happens
[00:04] <arraybolt3> Actually, there's a StateChangeReason thingy I can use that should make this doable.
[00:05] <arraybolt3> Might even allow popping up a "Bad password" dialog.
[00:13] <arraybolt3> Also going to implement some intelligent combo box locking
[00:26] <arraybolt3> Almost worked - I didn't get the password popup due to a race condition, but at least I didn't totally lock myself out of being able to connect to that network at all :P
[00:26] <arraybolt3> which was the previous behavior
[00:32] <arraybolt3> I feel like I'm developing for some sort of embedded device, coding in a VM and then transferring binaries for testing onto my test laptop :P
[00:35] <arraybolt3> Woot, it works!
[00:35] <arraybolt3> OK, now to change some text a bit...
[00:45] <arraybolt3> And it works. There's a segfault that needs help, after that I'll push.
[00:48] <arraybolt3> oof, actually the segfault isn't so easy to fix, I'm just going to push this for now and tackle the segfault later. (Basically if you try to input a WiFi password that is too short, NetworkManager refuses it, which ends up with an invalid pointer or something along those lines sneaking into the code, and then boom.)
[00:51] <arraybolt3> Pushed, ready for others to test. Works on an old Chromebook over here.
[00:52] <arraybolt3> There's one very unlikely race condition I left in because I didn't see any good way to fix it (NetworkManagerQt's way of doing things didn't work), but you'd have to type your WiFi password very shockingly fast to run into it (like you'd have probably a sixth of a second to type it, maybe not even that).
[00:53] <arraybolt3> So unless someone's typing speed is approaching the speed of light, not a problem.
 well how fast could you paste your password from say a password manager
 fyi discourse should be back
[01:12] <arraybolt3> lynorian: Valid point, but if someone's managed to sideload a password manager onto the live ISO, import their password database, load the WiFi password, and copy it, all before LXQt is started and while the installer prompt screen is active... That's probably unlikely to happen.
[01:13] <arraybolt3> Besides, I would expect most people don't keep their WiFi password in a password manager (with the exception of KDE Wallet which I think acts as a secret service on KDE-based distros, but that's not technically being used like a password manager usually would be).
 Can you please describe the race condition in more detail? So I can take a look at it?
 With enough red bull I may be typing that fast :P
 Also, could you be more specific on your testing for the segfaults? So I know how to reproduce it and can add appropriate air handling and both the u I and the back end
 If there's a certain limit for how big password should be. We should be aware of that and prevent that in the u I
[01:44] <arraybolt3> If you remove a connection with NetworkManager::Connection::remove(), it gives you a QDBusPendingReply. If you then waitForFinished() that, it returns before the connection is actually fully removed. Once waitForFinished() returns, the code then calls handleWiFiConnection() (which I tweaked slightly) to ask for the password again. If you type your
[01:44] <arraybolt3> password and click Connect REALLY SERIOUSLY FAST, you might be able to accidentally add a connection that still exists because NetworkManager isn't done removing it.
[01:44] <arraybolt3> But again, that would require you to type and connect so insanely fast that you can beat NetworkManager's connection removal routine, which gives you around 1/6 of a second to do the whole thing.
[01:44] <arraybolt3> I would imagine something like openQA might trip this, but a human? Never.
 There is a maximum. I don't remember what it is.
[01:45] <arraybolt3> For the segfaults, it's easy to trigger, just type a one-character password and click Connect. Boom.
[01:45] <arraybolt3> We should prevent too small or too large passwords, but that will probably require breaking the password dialog into its own class.
[01:46] <arraybolt3> As it is we're making the UI on-the-fly in the code itself, which makes it difficult to impossible to change the clickability of the Connect button depending on the currently entered password size.
[01:47] <arraybolt3> Since we probably need QLineEdit::textChanged to some routine that can change whether the button is clickable or not. Actually, that *may* be doable without needing an extra class now that I think about it... but really an extra class would be a lot cleaner.
 Yeah honestly it need some general clean up
 This should really be in a couple separate files plus as you may see, the language stuff is not quite implemented yet but again, I think that will be pretty easy compared to all of this networking stuff
[01:47] <arraybolt3> Agreed.
 Also, here's the other thing did you consider that? Maybe the bug with the removed connections is an upstream. Bug in behavior and that it actually should weights to return it because it is calling a specific weight for finished method. That should already be waiting for that. It doesn't seem intuitive or proper for it to not fully remove the connection before it returns
[01:49] <arraybolt3> Dunno, it seems like it probably returns once the removal starts, which might be expected behavior. There might be some other signal I can connect to in order to learn when the connection was really removed though.
 Please add a nice large to do comments on top of that. So we know to revisit it and potentially do an up stream. Tweak on that behavior so it's not as racy
 Looks like pre shared keys are 8 to 63 ascii characters
 Aaron from a preliminary look, your code looks great! Thank you very much!
[01:54] <arraybolt3> Thanks! :)
 You're more than welcome to go the extra mile and implement the language stuff. But that's totally up to you if you want to move on to something else. And let me handle that and the tagging of one point oh later this week
 Basically, my super-secret plan is to announce all the new installer stuff and that comparison and installer. Prompt in one go this weekend
[02:07] <arraybolt3> ahh nice
[03:04] <arraybolt3> ls
[03:04] <arraybolt3> nope, wrong window
 ls -lah
[04:48] <arraybolt3> sigh, these translations have me virtually stumped. If I use lupdate and lrelease manually, it seems to work, but CMake seems to just be ignoring my requests to do translations.
[05:12] <arraybolt3> YAY it works!
[05:12] <arraybolt3> Grief. CMake is cool and all but AAAAAAA it can be hard sometimes.
[05:21] <arraybolt3> ugh, one last test, do the strings, you know, *actually translate* now that the translator loads?
[05:23] <arraybolt3> WOOHOO they do!
[05:23] <arraybolt3> Alright. Need to add some documentation, and then finally I'll push.
[05:47] <arraybolt3> Alright, pushed. That was way harder than it needed to be.
 discourse down again @teward001 503 service unavailable
 lxqt-pm> Yup, known bug, file it and assign it to me ;P (re @lubuntu_bot: (irc) <arraybolt3> tsimonq2: So remember the old lxqt-powermanagement issues we had before, where systems without a battery would get a notification? We now have the reverse problem - upon booting a system with a battery a notification is made that the battery is present and charging.)
[14:14] <arraybolt3> +1
 that's a different error i'll fix it (re @Leokolb: discourse down again @teward001 503 service unavailable)
 on a work meeting at the moment after that i'll poke
 fixed, problem was the NGINX reverse proxy that sits in front of the Discourse docker containers was not running.  It's up again (re @Leokolb: discourse down again @teward001 503 service unavailable)
[16:45] <tsimonq2> sudo systemctl enable --now nginx
[16:45] <tsimonq2> XD
 `sudo systemctl start nginx && sudo systemctl enable nginx` but ye basically (re @lubuntu_bot: (irc) <tsimonq2> sudo systemctl enable --now nginx)
 Confirm @teward001 all good now tks (re @teward001: fixed, problem was the NGINX reverse proxy that sits in front of the Discourse docker containers was not running.  It's up again)
[18:23] <tsimonq2> Just did a respin of the daily to add those five languages to ship-live like we talked about.
 @tsimonq2 did we accept vorlon's merge request targeting noble to drop muon?
[18:24] <tsimonq2> I think we +1ed it but didn't accept it.
[18:24] <tsimonq2> doing
 ack, tyvm
[18:25] <tsimonq2> commit e57b65a19c0711462fb29c3dcfaca89c3f3ca505
[18:25] <tsimonq2> Merge: 55f2522 2cb1140
[18:25] <tsimonq2> Author: Steve Langasek <steve.langasek@canonical.com>
[18:25] <tsimonq2> Date:   Mon Dec 4 13:04:48 2023 -0800
[18:25] <tsimonq2> Merge branch 'muon-decay' into noble
[18:25] <tsimonq2> so already donbe
 nice
 well now that my work day is over i'mma bash my head on a wall, get some lunch, take a nap, then yell at Microsoft for $DAYJOB stuff at 4PM local
[18:29] <tsimonq2> hah nice have fin :)
[18:29] <tsimonq2> *fun
[18:29] <tsimonq2> http://cdimage.ubuntu.com/lubuntu/daily-live/current/noble-desktop-amd64.list
[18:29] <tsimonq2> Looks like langpacks landed.
[18:30] <tsimonq2> @Roberalz ^^^^^^^^^^^^^^^^^^^ sudo apt install language-pack-es language-pack-gnome-es language-pack-kde-es no longer requires an internet connection \o/
 \o/ (re @lubuntu_bot: (irc) <tsimonq2> @Roberalz ^^^^^^^^^^^^^^^^^^^ sudo apt install language-pack-es language-pack-gnome-es language-pack-kde-es no longer requires an internet connection \o/)
 Great
[18:31] <tsimonq2> Increased the ISO size by 21.72928 MB. Snaps do that on a weekly basis ;P
 @Roberalz Also, in case you haven't been following, that installer prompt will also automatically switch the live system language when a new one is selected.
 I think all the localization folks, especially the Spanish folks, will be really happy to hear that :)
 We'll formally announce later this week
 Perfect!!! (re @tsimonq2: @Roberalz Also, in case you haven't been following, that installer prompt will also automatically switch the live system language when a new one is selected.)
 Good news!!!
 💯
 As always, please let us know if anything is up with translations :)))
 OK no problem (re @tsimonq2: As always, please let us know if anything is up with translations :))))
 Thanks again for all your help with the localization teams @Roberalz :)))) we really appreciate it!!!!
 Thank you all for the work you do (re @tsimonq2: Thanks again for all your help with the localization teams @Roberalz :)))) we really appreciate it!!!!)
[19:04] <tsimonq2> arraybolt3: Do you wanna see something *really* *super* *cool*?
[19:04] <tsimonq2> arraybolt3: It's so amazing. Just top class.
[19:04] <tsimonq2> arraybolt3: On the live ISO, log out and log back into the session like meh, 5-10 times. Either via the GUI or restarting SDDM.
[19:05] <tsimonq2> arraybolt3: loginctl list-sessions just keeps growing and growing.
[19:05] <tsimonq2> arraybolt3: And according to ps -aux, all of those prior sessions are still running.
[19:05] <tsimonq2> Fun stuff, huh? :P
[19:10] <tsimonq2> That extra session - and the fact that pcmanfm-qt apparently works for translations but the desktop entries in the panel aren't being translated - will make the live session switcher just *slightly* more challenging. :P
[19:11] <tsimonq2> If anyone can figure out a reproducible way to change the *entire* live session language, just let me know.
[19:15] <tsimonq2> Pretty likely going to be a fix in the same area of the code adding the translations for the Ubuntu desktop entries... since those are the only translations working right now, lmfao.
[19:17] <tsimonq2> Oh nevermind! I figured it out!!
[19:18] <tsimonq2> My usual testing language is es_MX, so this worked (along with a log out/log back in): sudo update-locale LANGUAGE=es_MX LANG=es_MX.UTF-8 LC_ALL=es_MX.UTF-8
[19:19] <tsimonq2> Before that, I just had to install language-pack-gnome-es and language-pack-kde-es.
[19:19] <tsimonq2> Now, the only untranslated entry in the menu is Apply Full Upgrade, and we have a fix for that already pending :)
[19:21] <tsimonq2> And Calamares even launches with es_MX when launched too \o/
[19:25] <tsimonq2> @RikMills: Can we please drop k3b-l10n to Recommends or Suggests on all the KDE langpacks? Volunteering to do the work, asking for an ack before proceeding.
[19:50] <arraybolt3> tsimonq2: Fun stuff :P re logout not really logging out
[19:56] <RikMills> tsimonq2: we can
[19:56] <RikMills> ack
[20:07] <arraybolt3> tsimonq2: So, the failing logouts raise a question for me...
[20:07] <arraybolt3> ...lemme test something
[20:08] <arraybolt3> but does logout work on an installed system either?
[20:08] <arraybolt3> If it doesn't, that's bad.
[20:18] <arraybolt3> yeah so logout is leaving a pile of sessions left even in the installed system.
[20:22] <arraybolt3> Testing on 22.04 LTS to see what happens.
[20:27] <arraybolt3> alright so sessions do **not** pile up on 22.04.3 LTS out of the box without updates.
[20:27] <arraybolt3> So now if I install updates...
[20:28] <arraybolt3> (don't expect this to change anything, but after this I install 22.04 LXQt Backports, and I *do* expect to see changes from doing that.
[20:35] <arraybolt3> tsimonq2: btw we should be thinking about how to get the needed changes into Casper to allow the lubuntu-live-environment session to become the default rather than the Lubuntu session - that way we avoid the "desktop pops up but then so does the installer prompt" weirdness that we have now.
[20:38] <arraybolt3> also I noticed that pcmanfm-qt on RISC-V64 in the Backports repo FTBFS with a depwait. Should I perhaps poke it and see if it will build now?
[20:39] <arraybolt3> oh, I see what happened, we need some NCRs in order to make that work
[20:39] <arraybolt3> in face we'd need to NCR the whole entire Backports stack
[20:39] <arraybolt3> *in fact
[20:41] <arraybolt3> tsimonq2: do you think it's worth me NCRing the whole Lubuntu Backports stack just for the sake of getting RISC-V builds in there? I feel like it might not be worth it.
[20:41] <tsimonq2> arraybolt3: If I thought it would be worth it, it would have been done already. ;P
[20:41] <arraybolt3> Alright, I have confirmed that the login pile-ups are LXQt's fault.
[20:41] <tsimonq2> arraybolt3: casper> ack, file the bug! "Bring out the list!"
[20:42] <tsimonq2> "Bring out the whole ocean!"
[20:42] <tsimonq2> arraybolt3: Bisect it using your mind again? XD
[20:42] <arraybolt3> I cannot reproduce them on a fully updated 22.04 LTS VM, but I can reproduce them on a fully updated 22.04 LTS + Backports VM.
[20:42] <arraybolt3> tsimonq2: lol
[20:42] <tsimonq2> https://www.youtube.com/watch?v=oBw4EllCluo
[20:44] <arraybolt3> tsimonq2: just watch it be the lxqt-menu-data change that b0rked things :P
[20:48] <arraybolt3> actually, probably not, probably something changed in whatever ships lxqt-leave
[20:55] <arraybolt3> guiverc, tsimonq2: Soooo... it looks like switching to DEB822 is what broke command-not-found. It can't parse /etc/apt/sources.list.d/ubuntu.sources.
[20:56] <tsimonq2> Does command-not-found support DEB822 at all?
[20:56] <arraybolt3> Probably not.
[20:56] <arraybolt3> I imagine it wouldn't be difficult to add support for it though.
[20:56] <tsimonq2> It is a goal for this feature; if you'd like to JFD the implementation, there's libraries for at least C++ and Python, and you can add it to your core dev app :P
[20:56] <tsimonq2> *for this release
[20:58] <arraybolt3> tsimonq2: out of curiosity why did you disable the nonfree sources? They've been enabled by default for years, no?
[20:58] <arraybolt3> What if someone wants to install rar? :P
[20:58] <tsimonq2> arraybolt3: That's what the installer checkbox is for. :P
[20:58] <arraybolt3> ohhhhhh
[20:58] <arraybolt3> nifty!
[20:58] <tsimonq2> Also, juliank thinks they should all be able to fit into one file.
[20:58] <tsimonq2> File the bug! :P
[20:58] <arraybolt3> I've been experimenting with that as little as possible since it's still in an unfinished state.
[20:59] <tsimonq2> Nah man, they want to push that as default *this cycle*.
[20:59] <arraybolt3> kk, well then command-not-found will need some help
[20:59] <guiverc> arraybolt3, I'm not surprised; i've experienced it with xubuntu i think and thus believed it impacted all
[21:00]  * guiverc just swore at myself my use of `rcp` again...  being removed from ubuntu/debian now by default...
[21:00] <tsimonq2> rcp?
[21:01] <tsimonq2> arraybolt3: How's the MOTU stuff coming btw? I know, I keep sidetracking you. :P
[21:03] <arraybolt3> tsimonq2: somewhat stalled right at the moment, one of the scripts I need to set up a local Britney instance has vanished out of its linked location and I'm not exactly sure who has it :P I pinged bdmurray about it a couple days ago but he must have missed it. I should ask him again.
[21:03] <arraybolt3> I think I'm going to miss the December 11 meeting though.
[21:36] <arraybolt3> no one here happens to have a spare Kinetic ISO they could shoot me, do they?
[21:36] <arraybolt3> I can try to fish it out of archive.org if not
[21:37] <arraybolt3> so far I've determined that the login issue doesn't exist in 0.17.0 and does exist in 1.2.0 (Lunar) so that *might* be close enough to bisect comfortably, I dunno though.
[21:40] <tsimonq2> arraybolt3: $ ls ISOs/lubuntu-22.10-desktop-amd64.iso
[21:40] <tsimonq2> ISOs/lubuntu-22.10-desktop-amd64.iso
[21:42] <arraybolt3> tsimonq2: great, throw it into https://transfer.zip/ and give me the download link if that's OK (that'll peer-to-peer transfer it to my computer)
 Sure, just give me like 10-15 mins. I have a gigabit connection so your side will be the throttle :P
[21:43] <arraybolt3> np
[21:44]  * guiverc has a load of ISO & manifest files if ever useful
[21:55] <tsimonq2> arraybolt3: Link sent over Matrix.
[21:55] <arraybolt3> Thanks, appears to be working :D
[21:56] <tsimonq2> Wow, you weren't kidding about your connection speed ;P
[21:56] <tsimonq2> "5 hours later"
[21:56] <tsimonq2> I guess I can't do more debugging requiring me to toggle my network access ;P
[21:57] <arraybolt3> Yeah it appears to be a bit pokey atm
[21:57] <arraybolt3> Usually it's faster than this though, perhaps the peer-to-peer transfer nature makes it slower.
[21:59] <arraybolt3> Turns out the script for using Britney against PPAs is more lost than I thought (bdmurray said "nobody knows" when I asked him where it was :P) so I guess I'm relegated to coding an alternative myself
[21:59] <arraybolt3> which will be fun since I've never used Britney locally before :P
[22:45] <arraybolt3> Well now we know how to break a VM :P
[23:02] <arraybolt3> https://bugs.launchpad.net/ubuntu/+source/lxqt-build-tools/+bug/2045706
[23:02] -ubottu:#lubuntu-devel- Launchpad bug 2045706 in lxqt-build-tools (Ubuntu Noble) "Logging out of Lubuntu does not fully terminate earlier logged-in sessions" [Undecided, New]
[23:43] <arraybolt3> I think this is probably actually SDDM's fault.
[23:44] <arraybolt3> I can't be sure *yet* since downgrading SDDM didn't fix the problem once I caused it, but it's the only thing left that looks like a real suspect (installing most of the new LXQt stack didn't reproduce it and everything left over looked non-suspicious).
[23:54] <LimeOn> Hello, i dont know if this would help in any way, but i have the last lxqt version and tried to reproduce those steps in the bugreport you made. I tried to log-in/logout 3 times and always see just 1 session. I do not know if theres something revelant you could need, but as i said some days ago, i am under arch linux. Just some info to throw some light in the issue
[23:54] <LimeOn> 'loginctl' changes its number, but in the end it says '1 sessions listed.'
[23:54] <LimeOn> so, i guess, theres just 1 session running
[23:54] <LimeOn> sddm here too
[23:55] <tsimonq2> arraybolt3: Can we reproduce it with just a pure Openbox session?
[23:55] <tsimonq2> If we can, it's SDDM. If we can't, it's LXQt.
[23:59] <tsimonq2> >____________<