Thursday, 2021-07-08

opendevreviewMerged openstack/diskimage-builder master: Add element block-device-efi-lvm
*** ykarel|away is now known as ykarel04:34
*** redrobot4 is now known as redrobot04:52
*** zbr is now known as Guest16405:03
*** ysandeep|away is now known as ysandeep05:18
opendevreviewMerged opendev/infra-manual master: Switch docs from referencing Freenode to OFTC
*** rpittau|afk is now known as rpittau06:59
fricklerfungi: ^^ added a comment about a possible follow-up here, but considered it more important to get this finally merged now07:01
*** amoralej|off is now known as amoralej07:07
*** ykarel_ is now known as ykarel07:14
*** jpena|off is now known as jpena07:19
*** sshnaidm_ is now known as sshnaidm08:25
*** ykarel is now known as ykarel|lunch08:41
*** ysandeep is now known as ysandeep|lunch08:49
*** ysandeep|lunch is now known as ysandeep09:51
*** ykarel|lunch is now known as ykarel10:46
*** jpena is now known as jpena|lunch11:31
*** slaweq_ is now known as slaweq11:42
*** amoralej is now known as amoralej|lunch12:02
mnasertwo hours ago we hit change #800000 :)12:17
*** jpena|lunch is now known as jpena12:27
fungifrickler: thanks for the idea! i'll put something together12:37
fungimnaser: i guess we need to declare today an opendev holiday of sorts12:37
* fungi puts on his party hat12:38
*** rpittau is now known as rpittau|afk12:45
opendevreviewJeremy Stanley proposed opendev/infra-manual master: Re-add suggestion to register IRC nicks
*** amoralej|lunch is now known as amoralej13:03
opendevreviewJeremy Stanley proposed opendev/infra-manual master: Re-add suggestion to register IRC nicks
*** ykarel_ is now known as ykarel|away14:18
clarkbnick registration change lgtm14:55
*** frickler is now known as frickler_pto15:03
ricolinclarkb, fungi the build for ebian-bullseye-arm64 seems completed on is that help on currently stuck jobs in check-arm64? ?15:17
fungiricolin: i have a feeling those images are failing to boot, we'll have to check the nodepool-launcher logs for them15:19
clarkbthough failing boots should result in NODE_FAILURE15:19
clarkbcould be iterating really slowly through the retries though15:19
clarkblinaro and osuosl both show instances in use generally though so likely something specific to that image like boot failures15:20
clarkbI'm following the TC meeting because we'll be discussing the ELK stuff in a bit but I'll see if the logs show anything obviously wrong15:21
clarkbthere are 18-19 hour old bullseye arm64 images in linaro and osuosl15:23
fungiyeah, i just watched node debian-bullseye-arm64-linaro-us-regionone-0025448707 boot and go ready 15:24
fungii wonder if we have a bunch of stale node request locks again15:25
clarkbthere is one in use right now15:25
clarkbbut that is the only current arm64 bullseye image I see (that should rule out consistent boot failues though)15:25
fungii'm going to pick one of the older ones and see if i can find evidence the node request is being held indefinitely by the launcher after it should have been released, but am expecting that's what i'm going to find, in which case we can remedy it with a launcher restart to free all the stale node request locks15:30
clarkbsounds good15:30
corvussome thoughts on matrix from our friends in the ansible community:
mordredit's a good read - I highly recommend it15:35
*** ysandeep is now known as ysandeep|away15:44
fungiso the vast majority of the builds with queued debian-bullseye-arm64 requests are from 5.5 days ago, corresponding to the time of the last scheduler restart... for a clearer picture i'm going to pick one which isn't quite that old, 799348,2 (126.5hr instead of 134.5hr)15:58
fungithe kolla-build-debian-source-aarch64 node request for that patchset-uploaded event was 300-0014659571 (logged by the scheduler 2021-07-03 09:24:14,161)16:07
*** marios is now known as marios|out16:07
fungithat node request is still outstanding as of a few minutes ago, from the scheduler's perspective:16:08
fungi2021-07-08 15:58:32,707 DEBUG zuul.nodepool: [e: 4271459da6414a8c8e6fd0d269de61f0] Resubmitting lost node request <NodeRequest 300-0014659571 <NodeSet kolla-debian-bullseye-aarch64 [<Node None ('primary',):debian-bullseye-arm64>]>>16:08
fungilooks like nl03 declined 300-0014659571 after failure to boot in osuosl-regionone when it raised Exception: Unable to find flavor with min ram: 800016:12
fungibut never unlocked the node request, so node_failure was not returned16:13
fungisince that's confirmed, i'm willing to wager the remainder are in similar situations, so restarting the launcher to free any node requests it may still be sitting on locks for16:13
fungiunless anyone wants the process preserved in this condition for further investigation (would a thread dump be at all useful in this case?)16:14
corvusfungi: i'm not planning to debug that right now, but a thread dump is easy so may as well?16:15
corvusfungi: if you want to restart all the launchers, that would be convenient16:16
corvus(we'd be prepared for the upcoming zuul change which requires that)16:16
fungisure, i'll trigger a thread dump, and then happy to restart all 4 launchers16:16
*** jpena is now known as jpena|off16:22
fungicorvus: the thread dumps don't look very helpful anyway:
fungioh, nevermind16:30
fungii shouldn't have only grepped for loglines with timestamps16:30
clarkbdo we need to do pairs of thread dumps on odepool due to yappi?16:32
fungiyeah, i did16:32
clarkbI know that has a big impact on the zuul side but I'm not sure if we have yappi installed on the nodepool launchers16:32
clarkbah yup your paste shows the starting and stopping of yappi16:32
fungiokay, proper thread dumps are now in
fungithat should include everything16:33
fungii'll get to work restarting all the launchers16:33
clarkbI need to find some breakfast but then I'm going to dive into zuul reviews. Then I need to pick up the gerrit account classification work for the accounts that are left. Then hoping on Friday to look at the EMS setup16:36
clarkbI guess Friday is tomorrow now16:36
clarkbwe haven't seen any complaints over the previously disabled gerrit accounts have we?16:36
fungi#status log All nodepool launchers restarted on "latest" docker image (built from change corresponding to the nodepool 4.2.0 release)16:40
opendevstatusfungi: finished logging16:40
fungiclarkb: i haven't, no16:40
*** amoralej is now known as amoralej|off16:41
fungiso anyway, there's at least no stuck nodelauncher thread for that node request, assuming the numbers at the end of the nodelauncher thread name are expected to correspond with the numbers at the end of the node request itself16:47
fungiand if they're not supposed to correspond, i don't see any other way of trying to match them up16:47
fungilooks like those stuck queued builds ended up returning node_failure results after the nl03 restart16:52
clarkblooks like a kolla change is running some debian arm64 jobs now at least17:06
clarkbnot sure if those are bullseye though17:06
fungiwell, builds were running before too17:07
clarkbbut not for debian17:07
fungiit was just that there were also some stuck node requests, looked like17:07
clarkbya so debian may have been coincidence17:07
fungiwell, whatever causes these requests to get held onto seems to happen a lot more when there are launch failures of some kind, so it's possible there were specifically problems launching debian nodes for a while there which resulted in the majority of the stuck requests being for debian17:09
clarkber ah17:09
fungithough the specific one i dig into was rejected for osuosl because the min ram we specified didn't seem to correspond to any available flavors. that was a bit weird17:10
fungier, dug into17:10
clarkbfungi: I think some of the arm jobs run on xxlarge flavors in linaro17:11
clarkbthat those would be rejected by osuosl17:11
fungi"Unable to find flavor with min ram: 8000"17:12
fungithat doesn't look particularly large17:12
fungiwe're still logging that, roughly once an hour17:12
fungii wonder if some of the flavor list calls are erroring17:12
clarkbfungi: ya the xxlarge is 16GB ram iirc17:15
clarkb8000MB should be fine17:15
fungiright, i think i see the problem in the config17:16
fungilooks like we specify a flavor name for every label, and the min-ram only takes effect if the specified flavor is unabailable17:16
fungiin osuosl we specify opendev.large on all our labels, except debian-bullseye-arm64 which is set to use a flavor name of os.large instead, which i bet doesn't exist17:17
fungilooks like a copy-paste from a different provider which actually has that flavor name17:18
opendevreviewJeremy Stanley proposed openstack/project-config master: Correct flavor-name for bullseye nodes in osuosl
fungiclarkb: ^17:21
clarkbonce gerrit accepts my vote anyway17:22
fungiit's being testy today17:24
opendevreviewMerged openstack/project-config master: Correct flavor-name for bullseye nodes in osuosl
clarkbfungi: when linaro was down those bullseye nodes would only be able to launch in osuosl, then we had the problem with the flavor. That would explain backing up the queues. I wonder if that behavior makes the locks problematic17:51
fungiyeah, that's the sort of scenario18:10
fungiwe see it a lot with the airship-special node types when there's a problem in citycloud and rather than reporting node_failure some of them just get their node requests held indefinitely18:11
opendevreviewMerged openstack/project-config master: update channel config with new registered nic
clarkbmordred: for setting up EMS things should I use or will it allow us to have a number of admin accoutns and we can use personal email? Probably a good idea to use infra-root@ anyway but thought I'd ask20:05
opendevreviewMerged opendev/infra-manual master: Re-add suggestion to register IRC nicks
mordredclarkb: it allows us to have a number of admin accounts20:20
mordredwell - hang on - let me re-state20:21
mordredthe EMS control panel has a single login - use infra-root@ for that20:21
mordredthen - that control panel lets us define the other admin accounts for the actual matrix service20:21
clarkbmordred: cool that was the distinction I wanted to know about20:23
mordredclarkb: be sure: "Advanced Options: Custom DNS: On"20:25
mordredeverything else is super easy (that's also super easy - but I messed it up once so I'm mentioning)20:25
corvuseven messing up is super easy!20:33
*** mgoddard- is now known as mgoddard20:50
opendevreviewmelanie witt proposed openstack/project-config master: Update ceph grafana for the current jobs
*** dviroel is now known as dviroel|out21:44
mordredcorvus, clarkb: do we happen to know - in the gitea image, if we put a random file in custom/public - it should get served by gitea, yeah?22:23
clarkbmordred: yes I blieve so. And if not we have apache there now too and can easily serve it from apache22:24
mordredshould we do that instead?22:24
mordredI was going to write you a patch for the well-known files real quick22:24
clarkbI'm not sure I have a preference. apache is largely there to filter dos stuff currently22:26
clarkbBut could easily serve a file22:26
clarkbon the other hand our docker images aren't really reconsumable so no big deal to make them more specific22:26
fungiwhich well-known files?22:26
mordredlet's try the docker imge for now - it's the easiest22:26
mordredwe can always make it harder later22:26
clarkbfungi: for matrix to find the matrix server once it exists22:26
fungioh, it expects a webserver in the parent domain?22:27
opendevreviewMonty Taylor proposed opendev/system-config master: Add matrix well-known files for opendev
corvusfungi: eg:
corvusor what mordred just pushed :)22:27
mordredI'm assuming that's the name clarkb is going to pick for our host fwiw :)22:28
mordredclarkb: so - you know - pick that name - or else I'll have to UPDATE my patch *yikes*22:28
fungiis that to be able to make the accounts be instead of
mordredfungi: more for the channels22:28
corvusbut yes, that too for the bots22:29
fungiahh, okay, but same22:29
mordredah - yes - also for bots22:29
mordredso they'll be @awesomebot:opendev.org22:29
fungiand this is as an alternative to using something like srv rrs in dns i guess22:29
corvusand probably instead of '' instead of instead of ''22:29
corvusfungi: yes, there is a dns way, but it gets messy with certs22:30
mordredthere's a dns option too22:30
mordredsince we do have a webserver we can deploy too - it's fairly simple22:30
fungithanks, makes sense22:30
mordredclarkb: it's basically going to look like going through half of the setup page at element, then coming and landing the well-known files and a dns cname change, then finishing22:31
mordredactually - I don't know if we want to do the cname ... what do you think corvus ?22:31
mordredI don't think we actually need to have a since nobody is going to be logging in to it22:32
clarkbadmins might?22:32
clarkbbut ya I think we can just use the actual name22:32
mordredyeah - but admins can just log in to opendev.ems.host22:32
fungiadmins can also just... yeah22:32
mordredyeah - no need to brand things for us22:32
clarkbwhat sort of caching is involved in the advertising of that info? just wondering if it is safe to land that nowish and then update later if things change22:33
clarkbalso might do the wrong thing until the server exists (like someone could squat it?)22:33
fungithey could also squat it now22:34
fungieven without that change22:34
mordredyeah. and then we could just change it if they did22:34
mordredand they'd be silly22:34
clarkbright its fine to be squatted now if we don't add well known files pointing at them22:34
clarkbthen when we make the server we can ensure the well known file matches22:34
clarkbbasically I think I'm saying lets not merge that until the server exists22:34
mordredwell - we COULD update that to be an invalid target just so we could verify adding teh files like that works22:35
mordredlike - point it to matrix.example.org22:35
mordredthen update it when clark does the form thing22:36
clarkbI would do as we control that domain22:36
clarkbbut ya then we can confirm that gitea serves the files as expected22:36 is a purposefully invalid tld22:36
clarkbbut could theoretically be shadow dns'd in $organizations22:37
clarkbI want to say google did something like that then when all these new tlds opened up stuff broke for them22:37
opendevreviewMonty Taylor proposed opendev/system-config master: Add matrix well-known files for opendev
mordredclarkb: good idea re: matrix.opendev.org22:38
rm_workooo matrix-talk :D when do I get to connect up? :P23:14
clarkbunsure at this point. I am planning on poking at a server setup tomorrow23:14
clarkbrm_work: that said many people here are already connected via matrix23:15
clarkbso you could go that route if you want an early jump23:15
rm_workah, yeah I have run a homeserver for years, with our own stuff, just curious when things will be more official :D23:15
rm_workjust IRC-bridging then?23:15
mordredrm_work: zuul is working on moving to matrix native23:17
clarkbrm_work: well irc briding is already a thing thanks to matrix.org23:17
mordredrm_work: corvus and I are both here via matrix bridging23:18
clarkbbut ya zuul specifically is wanting to be matrix native. I have no idea what it may look like for other projects at this point23:18
rm_workyeah, right23:18
rm_workis there a doc/etherpad somewhere about this so I can sate my curiosity without asking you a bunch of questions you've answered 10+ times? :D23:18
mordredrm_work: in fact, if you go to the zuul community page: - you'll note that the link to the #zuul channel is actually a matrix link :)23:18
mordredrm_work: yah - link coming23:18
rm_workjust doing my weekly drive-by scan to see what ya'll are up to and got intrigued since i love Matrix :)23:19
clarkbmordred: I'm finally catching up on the ansible doc and I'll be honest it feels super preachy and I think we need to avoid that in our transition23:19
mordredclarkb: I agree - it's not the right tone for us23:20
clarkb"it was the right choice a decade ago; it is not today" <- statements like that claerly indicate where the author's biases come from even if that may or may not be true23:20
clarkbyou can have a transition without telling people that like irc that they are old and should feel bad23:20
mordredyah - I think he goes on in that paragraph to say the real meat of that sentence23:20
mordredin a more useful way23:20
clarkbya but the setup just kills it imo23:21
mordredrm_work: also: and 23:21
corvusclarkb: oh yeah totally agree there23:21
mordredclarkb: yah. for me part of the big win is that you can have your cake and eat it too23:21
mordredwithout either eating or cake being "bad"23:22
corvusi love irc and think matrix is better in ... basically every way23:22
corvuswhat mordred said :)23:22
mordredcorvus: ++23:22
corvusmostly because i see it as a near complete superset of irc23:22
rm_workis there an easy way to join these channels via matrix if i already have all my matrix stuff set up, just like "join room" ?23:23
clarkbya I mean I'm the odd person that will say the clients aren't nearly as good in their current state (though they aren't bad)23:23
clarkbI also think that emoji and photos and gifs more often degrade conversations than contribute to them (and an explicit embrace of those tools is likely to be painful for many users)23:23
clarkbbut as far as federated chat goes matrix has a really neat setup23:24
corvusrmwork: for this room: /join #oftc_#opendev:matrix.org23:24
corvusrmwork: for this room: /join #_oftc_#opendev:matrix.org23:24
corvussorry, i'm having an underscore problem with this client :)23:24
corvusi will be fixing that this weekend23:24
rm_you[m]cool, thanks23:24
mordredrm_work: that same pattern works for any oftc channel23:24
rm_you[m]I see that rich-presence works, and that I get a neat [m] badge apparently lol23:25
clarkbnickserv and chanserv incantations are definitely painful though and matrix channel management (from what little  I've seen so far) is much much nicer for users23:25
mordredrm_you: yah - all the matrix people get rich things ... like reactions and whatnot23:25
corvusrmyou, rmwork: you can change your irc nick by chatting with 23:26
clarkb(I called out the emoji and gif thing bceause its literally the first reason listed for switching on that ansible doc)23:26
mordredclarkb: I have to admit - I'm still not an emoji fan - but I *do* like being able to ++ a specific line of text and have it be an annotation23:27
rm_you[m]<mordred "clarkb: I have to admit - I'm st"> hmm, does this work23:27
clarkbmordred: I haven't run into that on matrix yet, does it indicate who annotated things?23:27
rm_you[m]oh, ok that's ... uhh23:27
corvusi want to make the weechat plugin take "++" and turt it into a reaction23:27
mordredclarkb: yup23:27
clarkbmordred: beacuse discord totally doesn't and it drives me nuts23:27
mordredclarkb: if you look at this chat in matrix you'll see two line annotations - one from me and one from rm_you 23:28
* corvus uploaded an image: (17KiB) < >23:28
corvuslik that23:28
clarkbI don't think I get history if joining afterwards? I didn't with zuul iirc23:29
rm_you[m]yeah no history23:29
rm_you[m]but that makes sense23:29
mordredif not- we should change that - I think it should be configurable23:29
rm_you[m]I wonder where my image uploads go...23:29
corvuswe'll need admin access to change it23:29
*** ChanServ sets mode: +o mordred23:29
rm_you[m]yeah typically you can set in matrix whether new users get history, but I assume it may depend on which homeserver you're on -- for instance, when I connected, my homeserver didn't HAVE history -- not sure if it'd fetch it remotely?23:30
corvusrm_you: it can fetch it if the room allows23:30
mordredpoop. not something we can set on the bridge channel with irc ops23:30
mordredwe'd have to ask the oftc bridge admin to change it for us23:30
clarkbalso if you click the join a channel button in element this channel doesn't show up23:31
corvusbut the matrix-irc bridges don't allow it by default because that's not how irc works by default23:31
clarkbI swear it did for zuul though23:31
corvusclarkb: that's because we set the channel to be hidden23:31
clarkbso now I'm wondering why this channel is different23:31
rm_you[m]yeah this23:31
* rm_you[m] uploaded an image: (191KiB) < >23:31
corvusno, it doesn't show up for both23:31
clarkband we did that because the spamming23:31
rm_workhmm interesting, my image actually is served from instead of from my own homeserver23:32
*** ChanServ sets mode: -o mordred23:32
corvusclarkb: yep, you may have just looked at zuul during the brief time it wasn't hidden23:32
mordredrm_work: I think that's because the room is hosted at matrix.org23:32
mordredthe bridge room23:32
clarkbcorvus: ok so is /join a valid element command in a random channel buffer?23:32
rm_you[m]yeah just kinda surprising, typically uploads go to my own I thought regardless23:32
corvusmordred: technically, the room is hosted on every homeserver with a user on it :)23:32
rm_you[m]<clarkb "corvus: ok so is /join a valid e"> I think so23:32
mordredcorvus: good point23:32
corvusi got an link for rm_you 's image23:33
clarkbnew client grump: no power user command console23:33
rm_you[m]eugh have to remember that "reply" functionality is shitty on the IRC side23:33
clarkbI don't want to accidentally send #zuul a message if I get the command wrong23:33
mordredclarkb: want me to invite you to the room?23:33
corvusclarkb: can you elaborate?  i don't fully understand?23:33
mordredpoo. that doesn't work either23:33
rm_workah so ... IN MY CLIENT (element) I get a link from my homeserver, but on the IRC side it re-hosts it for the bridge, makes sense I guess23:34
rm_workso I guess it just gives a link from whatever homeserver is used23:34
rm_worknever knew that :) TIL23:34
clarkbcorvus: typically irc clients have a separate command buffer and if you screw something up no one ever hears because its a local buffer just for getting commands23:34
mordredrm_work: oh - yea - that makes total sense23:34
clarkbcorvus: but in element I don't seem to have anything like that so doing an explicit /join requires I type that into an existing buffer which will get the message if you forget the / or otherwise do it wrong23:34
* corvus uploaded an image: (19KiB) < >23:34
corvusclarkb: i typed "/notacommand" and got this: 23:34
corvusoh yeah if you forget the /23:35
Clark[m]ya so the usual concern is if you drop the / or have a prefix space (or some other string)23:35
Clark[m] /notacommand23:35
Clark[m]I've always found it good practice to have that separate buffer23:35
corvusClark: you could set up a room for that -- that's what weechat and irssi really do anyway23:35
rm_workcan Element have custom channel groupings? not seeing that (though I should go Google it I guess heh)23:35
corvusrm_work: try enabling "spaces" it's a beta feature23:36
rm_workah kk23:36
Clark[m]corvus: you can but thats still a bit different because a remote server will always see the commands in that case23:36
rm_workyeah I see it, now my client looks like Slack lol23:36
Clark[m]its not the end of the world though23:36
corvusrm_work: you can make your own private spaces, but public spaces will also be a thing (which is great for communities)23:36
rm_workyeah, uhh do we just... have a public opendev/openstack space?23:36
corvusClark: only your homeserver will see it, and if you're the only person in an end-to-end encrypted room, you're the only one who will see the content23:37
rm_workyeah, i can't be the only one with an encrypted chat with only myself XD23:37
corvusrm_work: you aren't :)23:37
rm_work(though I get warnings about chatting to myself a lot, I assume it thinks I am lonely or something lol)23:38
clarkbI guess that is the hack around that23:38
mordredI have a "general" room on my homeserver23:38
mordredit serves no purpose, since I';m the only user23:38
mordredI've startedusing it to leave myself notes23:38
clarkbstill feels odd to not hae a command console if commands exist (though maybe they have hidden it away to keep unwary users from stumbling on it)23:38
mordredclarkb: I mean ... there is a REST API - you could do all of these things via that and then it's REALY a command console ;)23:39
rm_you[m]<rm_work "yeah, uhh do we just... have a p"> so, this?23:39
mordredrm_you: no, not at this time23:39
rm_you[m]seems like it would be valuable23:39
mordredthat space woudl want to be hosted in the opendev homeserver23:40
clarkbit is still on my todo list to use the weechat plugin then I'm sure all these client concerns will become less of a concern23:40
mordredwhich does not yet exist23:40
rm_workmakes sense23:40
corvusclarkb: though underscores will be a concern :)23:40
mordredbut also - so far only zuul has decided to do anything with matrix23:40
clarkbcorvus: oh is it the weechat plugin causing that?23:40
corvuszuul's space would be like one channel for now, so limited use atm23:40
mordredclarkb: yeah. it's being "helpful"23:40
mordredcorvus: can you make spaces with channels on other servers?23:40
clarkbI see the reactions with proper attribution.23:41
corvusmordred: that works with private spaces; i assume it would work with public too23:41
fungiclarkb: sounds like the weechat plugin for matrix thinks you want to _underline_ text23:41
corvusyep, it treats it as emphasis23:42
corvussame as element actually23:42
fungiand has lousy heuristics for telling that apart from "normal" use of underscores23:42
Clark[m]oh that will take getting used to. I hate when slack clients do similar. My favorite thing is when slack decides things are https urls23:42
Clark[m]because the internet only exists in https23:42
corvuselement has good heuristics, weechat bad23:42
Clark[m]Makes it impossible to talk about dns records in slack23:43
corvus * foo_bar_baz  this was element23:43
corvus * foo_bar_baz  this was weechat23:43
corvus * foobarbaz  this was weechat23:43
rm_workyeah i just set up a bunch of spaces, seems it's really just an organization tool if you create "private" ones, works well, cross-homeserver/whatever23:43
clarkbinterestingly I haven't received corvus' messages in matrix23:44
corvusclarkb: those were edits23:44
rm_you[m]oh I wonder how PMs work23:44
corvusrm_you: they work fine23:44
corvusrm_you: they're just new matrix rooms23:44
rm_you[m]yeah but I assume they don't go over OFTC23:44
mordredthey do if one of the people is in oftc23:45
clarkbcorvus: that is a weird bride behavior23:45
mordredrm_you: but if I chat with you, it'll be matrix23:45
rm_workyeah i just sent a test message to corvus 23:45
rm_workseems it's matrix-direct23:45
rm_workwhich is fine23:45
rm_workwas just curious23:45
mordredyeah - matrix basically knows which of us are in via matrix and matrixes us via all the matrix23:46
corvusclarkb: yeah, on account of irc not supporting editing messages, they had to pick something.  you learn to read it.  ;)23:46
corvusbut it's a good example of a "degraded" experience in irc that's better in native matrix23:47
Clark[m]corvus:  I would've expected a bit more context, but ya now I know the "  *" prefixed lines refer to edits against previous lines23:47
rm_workit makes sense, that's what I manually type anyway when I edit things23:47
rm_work*when I edit stuff23:47
corvusClark: yeah, as long as folks don't wait too long to edit, it's easy enough to pick up.  and yeah, as rm_work says, it's a semi-convention in irc anyway23:48
rm_you[m]so the difference here is ...23:48
rm_you[m] * so the difference here is ... what?23:48
rm_workah k23:48
clarkbyup the other common convention is the ed substitution23:48
rm_workyeah that just automatically does it, easy enough23:48
rm_works/easy enough/very easy!/23:48
Clark[m]I wonder if that works in element23:48
mordredweechat should TOTALLY display edits in ed format23:49
Clark[m]s/works in element/works in the element client/23:49
mordredI wish I could configure element to show me edits in ed format23:49
rm_you[m]can I just get a side-by-side diff of edits plzkthx :P23:49
mordredmaybe I'll go learn enough of the codebase to add that option23:49
corvusmordred: you like writing js right?23:49
rm_you[m]ok this has been fun, thanks ya'll 😛 good end to my day23:50
mordredcorvus: this might be a useless enough feature that I could nerdsnipe myself23:50
mordredcorvus: it should be a sub-option of "experimental IRC mode"23:50
rm_you[m]time to ditch irccloud and set up all my channels here23:50
mordredrm_you: I've been using this for IRC for a bit over a year now23:50
mordredand have been quite pleased23:50
corvusrm_you: ++ before you go lemme dig up a link23:50
rm_you[m]yeah i have played with IRC bridges before and they sucked23:50
rm_you[m]this implementation is A++23:50
rm_you[m]actually I had an irc-bridge plugin running on my matrix homeserver several years ago and it was ... bad... memory leaks and crashes23:51
rm_you[m]had to uninstall23:51
rm_you[m]glad it's all fixed up23:51
mordredthe global federated bridges are super handy23:51
Clark[m]I think matrix has come a long way over all in the past few years too23:51
rm_you[m]actually i wonder when I started running that server23:51
rm_you[m]I say "a couple years" but like... i think that's just me being old and losing track of time XD23:52
Clark[m]I should set one up for family chat just to troll my brother23:52
Clark[m](he had to use matrix a few yaers ago and apparently the experience was not good at all)23:52
corvusrm_you: a lot of this you already know -- but near the end of this msg there's info about using matrix for both native and irc bridged channels --
Clark[m]but none of his complaints about performance seem to be an issue today23:52
corvusrm_you: there may be some useful info for you there, especially about the bridge commands23:53
rm_workyeah, will need to take-over my existing nick23:54
Clark[m]corvus: for weechat how does it handle things like images, reactions, and emoji? I guess the emoji are probably just displayed with whatever is in your font23:54
rm_workrm_you is for personal stuff, rm_work is for work :P23:54
* corvus uploaded an image: (5KiB) < >23:56
corvusClark: ^23:56
corvusClark: that's images; reactions i don't see23:56
Clark[m]Ok very similar to how it ends up across the bridge then23:56
rm_you[m]yeah reactions just don't show up ☚ī¸23:57
Clark[m]a clickable link and no reactions23:57
rm_workwell, my IRC client expands the link so, it's basically the same as element23:57
Clark[m]wee slack actually does the reactions in slack23:57
rm_workoh neat23:57
Clark[m]it gives you to them in text form with a counter that updates. No attribution though23:57
corvusi guess there's no reason weechat-matrix couldn't be extended to do that23:58
corvusit's python23:58
Clark[m]ya I imagine a similar thing to what wee slack does is possible. Not perfect but it at least gives you some indication23:58
*** rm_you[m] is now known as rm_you23:58
*** rm_work is now known as rm_work|23:59
*** rm_you is now known as rm_work23:59
rm_workuhh ok... takeover complete? :D23:59

Generated by 2.17.2 by Marius Gedminas - find it at!