[11:10] ScottK: just out of curiosity....how difficult would it be to mark a package as "requires reboot"? [12:13] macvr: There's no mechanism for knowing in advance. Currently you touch /var/run/reboot-required the postinst of the relevant package. [12:13] can we change that? [12:13] lamalex: We can change anything. It's free software. [12:13] ScottK: i undersatnd that there is no present mechanism... but how difficult is it going to be to implement it? [12:14] macvr: It shouldn't be really hard, but I'm not sure it's at all necessary. [12:15] With the exception of Firefox, generally you can install the new version and just catch it on the next boot and keep working. [12:15] The previous version keeps working just fine. [12:16] ScottK: the problem is updates requiring reboot can be done either at shutdown/login. while the rest are done in-session [12:16] ah... [12:16] so... there is no *need* for a reboot! [12:17] but to the contrary i'v found some packages misbehave :( [12:17] I'm of the opinion that the optimal solution in this area is to have a subtle, but noticeable indication to the user that there are updates available, they install them when they think it's a good idea, and you also give them a similar indication that a reboot is required and they can do that when they feel like it too. [12:18] macvr: There may be others. Firefox definitely needs an immediate restart (not the whole system though). [12:18] ScottK: hate firefox! [12:18] it messes up a lot of things , addons dont work! [12:18] * ScottK isn't a huge fan either, but it's an important use case for the system. [12:20] i think a lot of people would , jump ship to Chrome as soon as chrome is made for native linux and has provision to port firefox extension to chrome [12:32] ScottK: i like you idea... all that i required is re-wording... not use reboot , since it is a warning/harsh word for the users coming from windows [12:32] ScottK: could you mail your idea to the list? [12:33] macvr: Feel free to copy/paste/mail the IRC log. I'm unlikely to have time to write something better for a while (doing $WORK). [12:33] if i quote other members , i was said not to quote [12:57] ScottK: sent :) [13:57] lamalex: i guess you still arent awake ;p [13:58] huh? [13:59] I've been up for hours [13:59] It's 3pm [13:59] macvr: ^ [14:00] weird ctcp says 9 am! [14:01] ah, my vps is in the USA [14:01] but I am in france [14:03] au revoir monsieur ;p [14:03] tu pars? [14:03] yeh, he's too ashamed to stay [14:04] :) [14:05] macvr: It may have escaped your notice, but the way I described thinking updates should work is the way it worked before people started 'improving' it. [14:05] SiDi_: no... i think lamalex needs sleep [14:06] ScottK: i +1 you [14:06] ScottK: no...i know what you said... see my recent mail! [14:06] the problem is imo educational, and not in our implementation of updates [14:06] i'm ok for the update manager to be intrusive when security updates are pending [14:06] but not for regular updates [14:06] macvr: if *you* read what he said, ScottK is describing the old system of updates [14:07] SiDi_: I disagree. Even security updates are rarely urgent for typical users. [14:07] ScottK: no, but they are for the image of Ubuntu :) [14:07] think about users who'd get hacked cause of a flaw known for several weeks/months [14:08] all the "IT" websites would give us a nice advertising [14:08] and this is an issue [14:08] SiDi_: My point is that it's very rare the security issue that could cause a typical user to get hacked. [14:08] ScottK: i know that the update solution needs more "improving" but i'm just mentioning the rewording for a papercut [14:08] (not to mention massive attacks if our end users begin acting like windows's ones) [14:08] ScottK: indeed, but we need to make sure it never happens at all :d [14:08] SiDi_: +1 to ScottK: [14:08] macvr: this is not papercut-ish [14:08] the question isnt trivial [14:09] SiDi_: papercuts are all about rewording! dont confuse the two threads~ [14:09] there are currently 4 issues i can identify : some updates require reboot in order to be fully performed | some updates require app restart | updates shouldnt be intrusive | updates MUST be performed [14:09] for me updates on shutdown fails to address any of these issues [14:10] SiDi_: oh my god! i'm not trying to solve the updates! just proposed a papercut solution! [14:10] SiDi_: Go back and look for the last time there was a security fix for an issue that allowed remote priviledge escalation. It's been quite some time. We generally get only a handful per year of those and a small hand at that. [14:10] updates on login is a big amount of effort that fails to address #3 but that partially addresses #2 and #4 [14:10] ScottK: indeed, but security is all about being paranoiac :) [14:11] macvr: papercut for what ? at the moment when an update makes it needed to restart an app, it says "This app must be RESTARTED" it doesnt say rebooted [14:11] SiDi_: we get paranoid... but not the average user! most of them just dont care [14:11] SiDi_: update on login also solves 1 [14:11] SiDi_: It's about being rationale and balancing risk and benifit. If you've connected to the internet you've already made some compromises. [14:11] and totally addresses 2 [14:11] oh, i misread [14:11] lamalex: #1 is really not a desktop issue... only server issue since servers must do the updates AND stay up all time [14:11] #1 can be delayed for a day without problem [14:12] SiDi_: holly crap! I'm only talking about the updates requiring reboot! [14:12] ScottK: indeed :p [14:12] yah [14:12] it doesn't fully fail to address #3 [14:12] papercut for those! [14:12] macvr: we have to care instead of them. That's why mpt took such a drastic decision as update-manager popping up [14:12] i agree with the goal but not with the way it is achieved ~ [14:12] SiDi_: It doesn't acheive the goal either. [14:12] SiDi_: me too [14:13] lamalex: i think several mecanisms are needed, and most of them are educational [14:13] It assumes the wrong problem. [14:13] SiDi_: highly intrusive! [14:13] The problem isn't people not being able to figure out there were updates available, but that they didn't care. [14:13] ScottK: indeed [14:13] The caring part won't be fixed in software. [14:13] from my own testing, the "guy-who-knows-about-computers" has to shout in order to get the "end-user" to perform updates [14:14] Yep. [14:14] haha ^so true [14:14] I periodically ssh into all the desktops in the house and update them. [14:14] Can we agree on the 4 goals i wrote above, and try to see how each proposed solution addresses them ? [14:15] ScottK: im too lazy for that, i just ask if they thought about it when i have them on phone :) [14:16] SiDi_: the goals are right [14:16] SiDi_: I agree, except that other than in very rare cases there is almost zero urgency for updates to get installed. [14:16] (updates on shutdown with other mecanisms can address #2 actually : if we add preinst mecanisms to say that an app will have to be restarted after update, and if we check if the app runs during the update, we can ask the user if (s)he wants to delay the update upon closing this app / the session) [14:17] (and for upon closing the session, we gently propose it again when s/he does because s/he can have changed her/his mind [14:17] SiDi_: shutdown achieves #1 #3 #4 [14:17] Firefox updates as a class also tend to be somewhat urgent. [14:17] #1 is not a desktop issue macvr ^.^ [14:17] ah! [14:18] alright, i'll write a wiki page with the goals, and then i dont know what ill do but i have a few mins to find out ~ [14:19] by the way people, when you change title, please add (Was : $OLD_TITLE) [14:19] SiDi_: no... its more fun when people get confused ;p [14:21] djsiegel_: ping! [14:22] macvr: what's up, man! [14:23] djsiegel_: hi...are you interested in fixing this for the karmic papercut cycle? https://lists.launchpad.net/ayatana/msg00502.html [15:18] Takes time to write this damn wikipage [16:25] Amazing [16:25] i had a kernel panic when i had just *finished* the wiki page [16:25] im so pissed [16:28] * SiDi noticed the wiki saves templates automaticaly so ... https://wiki.ubuntu.com/Ayatana/UpdateIssues#preview [16:28] its almost finished [16:38] SiDi: the issue 4> remove the "non-negligible" , it sounds rude [16:38] SiDi: just "A proportion" / "A fair proportion" [16:38] macvr: It's wiki [16:39] Does it really sound rude ? [16:39] ScottK: i knowi can edit...... but he is stil working on it [16:39] ScottK: im playing with the template for adding proposals of solutions so its locked [16:40] SiDi: You might go ahead and put in "Revert to what it was in Intrepid" as a proposal is you're taking IRC edit requests. [16:40] ScottK: feel free to do so :) [16:40] ill let you know as soon as i stop editing [16:41] ScottK: im for a mix of previous + current behaviour + some addons [16:41] i'll explain it all in little time [16:41] OK. [16:41] Not that what was there before was perfect, but I think it's a better basis for further development than the Jaunty experiment. [16:50] i learnt about the current situation late : in xubuntu we kept the old behaviour [16:50] https://wiki.ubuntu.com/Ayatana/UpdateIssues can now be edited [16:50] im gonna work on some idea on my side [16:52] ScottK: macvr ^ [16:52] SiDi: already on it ;p [16:53] i'll announce it on the ML [18:28] SiDi: edited it... check it out [19:06] macvr: dont put a table of contents on the page pls ^.^ i put some damn long titles everywhere, the toc will be too big :p [19:07] SiDi: its better to navigate that way... title dont matter , it even easier if in future someone adds an idea, providing a link to the idea makes references easier [19:08] then i have to strip it down [19:08] bah, i'll see this tomorrow in details macvr, i might aswell cut it into several pages [19:08] what matters now is content :p [19:08] (btw you had left the ubuntu-art icon in the ToC Q.Q) [19:09] oops! [19:09] SiDi: do you have a screenshot of the restart dialogue? [19:10] no macvr [19:10] not relevant anyway : im on xubuntu [19:10] ;p [19:10] ;) === vorian is now known as heHATEme === heHATEme is now known as vorian