Status Meetings:2007-08-22:Log

From Camino Wiki
Jump to: navigation, search
[01:14am] You changed the topic to "http://wiki.caminobrowser.org/Status_Meetings:2007-08-22:Agenda".
[12:01pm] pinkerton joined the chat room.
[12:01pm] pinkerton was promoted to operator by chanserv.
[12:01pm] smorgan was promoted to operator by you.
[12:02pm] mento: welcome pinkerton!
[12:02pm] ardissone|food: indeed!
[12:02pm] pinkerton: heh
[12:02pm] smorgan: Who's the new guy?
[12:02pm] ardissone|food: ?
[12:02pm] mento: introductions, introductions anyone? 
[12:03pm] ardissone|food: oh, we're having fun
[12:03pm] ardissone|food: sorry to spoil it 
[12:03pm] ardissone|food: however, let's all open http://wiki.caminobrowser.org/Status_Meetings:2007-08-22:Agenda in our Camino du jour
[12:04pm] mento: mine's 1.5.1!
[12:04pm] ardissone|food: a very fine choice; i hear it's a great vintage
[12:05pm] mento: aged grapes from a satisfying branch
[12:06pm] ardissone|food: i moved 1.6 front-and-center this week; 1.5 is the last great vintage, but we need to start thinking about a new one
[12:06pm] ardissone|food: i still don't know what other bugs y'all talked about at the meet-up for 1.6
[12:07pm] ardissone|food: so if someone who does will update the scoping document and bugzilla, that would be great
[12:07pm] ardissone|food:        http://wiki.caminobrowser.org/Development:Planning:Camino_1.6    
[12:07pm] smorgan: We should talk about auto-update a bit at some point in the meeting
[12:07pm] ardissone|food: k
[12:07pm] ardissone|food: bugs we've fixed are strikethrough on that page
[12:07pm] ardissone|food: basically, most of tab scrolling is done
[12:08pm] ardissone|food: there's good progress of some sort on Tabsposé
[12:08pm] ardissone|food: but no "news" about the other major features
[12:09pm] ardissone|food: batwood and murph are both working on their smaller projects; batwood has patches for both of his
[12:09pm] ardissone|food: so that's where we stand
[12:09pm] ardissone|food: it's probably time to start really thinking about dates and a1 (even if we don't
[12:09pm] ardissone|food: "set" them yet)
[12:10pm] murph: I'll have the search editor in momentarily, I had to step back and make a few minor changes to accommodate Stuart's (much welcomed) WebSearchField
[12:10pm] ardissone|food:                           
[12:12pm] ardissone|food: 1.5.1, then?
[12:12pm] ardissone|food: we stole sam from the soul-sucking folks and released 1.5.1 friday
[12:12pm] ardissone|food: last i heard, early numbers were low, but i haven't gotten an update this week
[12:13pm] smorgan: feedback shot up, so someone is downloading it
[12:13pm] ardissone|food: it's still crashing plenty, too 
[12:14pm] ardissone|food: sam was having problems with the reports, but he ran one manyally last night
[12:14pm] ardissone|food:        http://talkback-public.mozilla.org/reports/camino/CM151/index.html    
[12:14pm] ardissone|food: we should look into the hex crasher at #2
[12:15pm] ardissone|food: several of the libobjc crashes are 1passwd still, just from a quick glance at them last night 
[12:16pm] ardissone|food: we mentioned this last week, i think
[12:16pm] ardissone|food: there's a new problem we've been hearing about where we stop loading pages or windows
[12:16pm] peeja joined the chat room.
[12:16pm] ardissone|food: smorgan's running that one through Bug 387085 
[12:16pm] peeja was promoted to operator by you.
[12:17pm] ardissone|food: anything else about the great vintage of 1.5.1?
[12:17pm] ardissone|food: ok, 1.0.6
[12:18pm] ardissone|food: our RC is ready and out there
[12:18pm] ardissone|food: smorgan: have you heard back from the 10.2 guy?
[12:18pm] smorgan: Nope, at this point I say we assume I never wil
[12:18pm] smorgan: l
[12:18pm] ardissone|food: ok
[12:19pm] ardissone|food: i've heard nothing from the other guy, either
[12:19pm] ardissone|food: mento: have you heard when 1.8.0.13 goes final final?
[12:19pm] mento: no
[12:19pm] mento: ss?
[12:20pm] ardissone|food: he's not here
[12:20pm] ardissone|food: afaik
[12:21pm] ardissone|food: we should poke him this PM PDT and then email marcello and markus?
[12:22pm] ardissone|food: skipping the meet-up category (er, broken record)
[12:22pm] ardissone|food: dun-dun-dun
[12:22pm] ardissone|food: SoC 2007!
[12:23pm] peeja: AS: Bookmarks are up for a second r
[12:23pm] ardissone|food: starring jeff and peeja 
[12:23pm] ardissone|food:                           
[12:23pm] peeja: it's targetted to cl|flying, but if anyone else has time, steal it from him
[12:24pm] peeja: he's been pretty busy
[12:24pm] peeja: and when that's sr I'm...halfway done!
[12:24pm] peeja: toolbar items already have a patch, though, which I need to respin
[12:25pm] ardissone|food: has comments for your comments
[12:25pm] peeja: and after/alongside that will be...drumroll
[12:25pm] ardissone|food: dun-dun-dun
[12:25pm] peeja: semi-automated testing!
[12:25pm] ardissone|food:  !!!
[12:25pm] pinkerton:                           
[12:26pm] pinkerton: imagine being able to tell when someone breaks bookmarks!!!
[12:26pm] peeja: I'll create a test suite that should flex everything that AS cat get at
[12:26pm] ardissone|food: awesome
[12:26pm] peeja: ardissone|food: I expect a lot of ideas from you 
[12:26pm] smorgan: Want to do unit tests for camino after that? 
[12:26pm] ardissone|food: and fixing script bugs?
[12:27pm] ardissone|food: i need to just start filing things i run into, and we can WF them if appropriate
[12:27pm] ardissone|food: that way i can at least search to see if we've discussed it before 
[12:27pm] peeja: "script bugs" being bugs in the scripting support?
[12:28pm] ardissone|food: mhm
[12:28pm] peeja: yeah, that's pretty much the rest of what's on my plate
[12:28pm] ardissone|food: when does SoC officially end?
[12:29pm] peeja: shrugs.
[12:29pm] peeja: jeff?
[12:29pm] jeff: 'pencils down' was on monday 
[12:29pm] smorgan: Yeah, it's done
[12:29pm] jeff: runs out the door
[12:29pm] smorgan: But don't tell jeff and peeja
[12:29pm] peeja: hey, I don't really care 
[12:29pm] ardissone|food: oh 
[12:30pm] ardissone|food: well, good to see them still working 
[12:30pm] jeff: so on the tabspose front, I am going to submit a new patch for the titles today and just take out the string truncation for a another patch
[12:30pm] jeff: and have a patch up for disabling most of camino's features when tabsposé is open
[12:31pm] jeff: smorgan is working on some gecko thumbnailing code so that we dont get as many crashes with plugins
[12:31pm] ardissone|food:                           
[12:31pm] smorgan: Yeah, that should go up today, maybe tomorrow
[12:32pm] smorgan: Now plugins will be invisible, which is weird but better than crashing
[12:32pm] ardissone|food: and if that works well, and jeff's other stuff lands, we start backporting to the branch? or is more polish needed first?
[12:32pm] ardissone|food: heh
[12:33pm] smorgan: Someone will need to do some work (hopefully not much) to get my code branch-safe
[12:33pm] smorgan: Because getting the image content from Gecko is different
[12:34pm] jeff: I think the one thing that needs to be cleaned up is just some of the distortion the view's get when resized
[12:35pm] ardissone|food: k
[12:36pm] ardissone|food: well, it certainly feels like a productive summer from where I am (even only seeing caps of the tabsposé half) 
[12:37pm] ardissone|food: the r queue is staying pretty full
[12:38pm] ardissone|food: our problem seems to be finding enough people to be here so we can rotate through them well
[12:38pm] ardissone|food:                           
[12:38pm] ardissone|food: hwaara thinks he should still be able to do reviews (and maybe some patches) weekends/etc
[12:39pm] ardissone|food: so we can keep trying to target him
[12:39pm] ardissone|food: (though ATM he seems to have hands full with transparent XUL)
[12:39pm] ardissone|food: ready for bugs?
[12:40pm] peeja: are we ever!
[12:40pm] thebot joined the chat room.
[12:40pm] ardissone|food:        Bug 185436    
[12:40pm] thebot: ardissone|food: Bug https://bugzilla.mozilla.org/show_bug.cgi?id=185436 enh, P1, Camino1.6, nick.kreeger@park.edu, NEW, Software update for release builds
[12:40pm] ardissone|food: kreeger asserts this will be really easy
[12:40pm] ardissone|food: smorgan has other thoughts
[12:41pm] smorgan: The API is reasonable
[12:41pm] smorgan: However:
[12:41pm] smorgan: - It will need a wrapper, like everything else
[12:42pm] smorgan: - It puts us at the mercy of the entire Moz update creation process, which is poorly documented as far as I can tell
[12:42pm] smorgan: - It has unknown interaction problems with our event model, like everything else in Gecko (maybe it'll be fine, maybe not)
[12:43pm] peeja: not to be belligerent, but why did we decide not to use sparkle again?  and could we add that reason to the bug?
[12:43pm] smorgan: My big question is: is incremental update something we want enough to eat what will undoubtedly be a higher dev and maintainence cost of Sparkle
[12:43pm] smorgan: peeja: ^^^
[12:43pm] smorgan: s/of/as compared to/
[12:44pm] ardissone|food: what's the cost of getting Sparkle into the tree?
[12:44pm] smorgan: I haven't tried, but from their docs it looks very easy
[12:44pm] smorgan: Or do you mean licence issues?
[12:44pm] ardissone|food: license and such
[12:44pm] ardissone|food: s/tree/repo/
[12:45pm] thebot: New build added to MozillaTest: macosx_update_verify (status: Success).
[12:45pm] smorgan: That I forgot, so not sure
[12:45pm] jeff: I believe Sparkle is BSD
[12:46pm] ardissone|food: re-downloading the entire app is so 1980s 
[12:47pm] smorgan: mento, pinkerton: thoughts?
[12:47pm] ardissone|food: the other question is what's the likelihood of getting Sparkle to do partial updates?
[12:49pm] peeja: sparkle reads an rss appcast and extracts a new app from a dmg
[12:49pm] smorgan: There's a ticket requisting it with no comments at all
[12:49pm] ardissone|food:                           
[12:50pm] peeja: what's so great about partial updates?
[12:50pm] smorgan: ardissone|food apparently uses dial-up
[12:50pm] smorgan:                           
[12:50pm] ardissone|food: they're very quick
[12:50pm] ardissone|food: save bandwidth
[12:50pm] ardissone|food: you can go nightly-nightly with no effort
[12:51pm] ardissone|food: ask froodian when you next find him
[12:51pm] pinkerton: i'm fine with re-downloading the whole app
[12:51pm] smorgan: Erm... it's really unlikely we would ever do updates for nightlies
[12:51pm] pinkerton: yeah, this would be for releasees
[12:51pm] peeja: is there a way to do a useful but simple "diff" of the whole bundle?
[12:51pm] peeja: we might be able to send out a patcher
[12:51pm] smorgan: peeja: that's what the Gecko update system does
[12:52pm] peeja: but could we do it without pulling gecko strings?
[12:52pm] ardissone|food: do our own wrapper on bsdiff?
[12:52pm] peeja: we might be able to leverage sparkle or something like it
[12:53pm] smorgan: There's no way I'm writing a new binary patching update system from scratch
[12:53pm] smorgan: that would be madnes
[12:53pm] smorgan: I'm not saying the Gecko version isn't doable
[12:53pm] smorgan: Just that it means added complexity now and forever
[12:53pm] smorgan: (as will any diffing system)
[12:54pm] peeja: doesn't know enough to know what bits can be separated out.
[12:54pm] smorgan: It's not about separating out
[12:54pm] smorgan: Let's assume that it's really easy to make the .mar updates
[12:54pm] smorgan: It's still another thing we need to build for every release
[12:54pm] smorgan: It's another upgrade path to test
[12:55pm] kreeger joined the chat room.
[12:55pm] thebot: Firefox: 'WINNT 5.1 qm-winxp01 dep unit test' has changed state from Test Failed to Success.
[12:55pm] smorgan: There's no question that a patching system is more fragile than a replacement system
[12:56pm] smorgan: I just personally don't think it's worth it to save people 10 megs every 2 or 3 months
[12:57pm] peeja: is there no way to patch the binary from outside of the app?
[12:57pm] smorgan: peeja: I don't understand what you are asking
[12:57pm] peeja: using a separate executable to do the patching
[12:57pm] ardissone|food: that's what mozUpdate does
[12:58pm] ardissone|food: "updater"
[12:58pm] peeja: oh, then why does it hit our event model?
[12:58pm] kreeger: updates are a matter of creating the diff and then modifying a XML file...
[12:58pm] smorgan: peeja: because there is polling and notification happening
[12:58pm] thebot: Just appeared in Planet Mozilla - http://planet.mozilla.org/ : 
[12:58pm] You kicked thebot from the chat room. (you are sooo annoying)
[12:58pm] kreeger: Since we are already bootstrapping under Gecko, might as well take advantage of it
[12:59pm] smorgan: I don't see that argument
[12:59pm] smorgan: It's going to be harder than Sparkle
[1:00pm] peeja: could we build a small updater that uses the gecko system and downloads via sparkle?
[1:00pm] kreeger: peeja: even harder
[1:00pm] smorgan: Making our own system will be harder than any other option
[1:00pm] smorgan: We should do one or the other
[1:01pm] kreeger: does the sparkle license align up?
[1:01pm] peeja: kreeger: ok, that's really what I've been failing to ask properly 
[1:01pm] peeja: "The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software."
[1:02pm] peeja: otherwise no limits
[1:03pm] smorgan: kreeger: did you find docs on making the patch?
[1:03pm] kreeger: ja... let me look it up again
[1:03pm] smorgan: Knowing what kind of size we can actually expect would perhaps be useful information in deciding
[1:04pm] kreeger: here is info on creating the update(s) http://wiki.mozilla.org/Software_Update:Testing
[1:04pm] kreeger: there is one I found for actually producing the binary...
[1:05pm] smorgan: Yeah, that's the useless one that says "see darin's notes" as the only instructions for making a patch
[1:06pm] peeja: does the system checksum the result of the patch to make sure it worked?
[1:07pm] kreeger:        http://wiki.mozilla.org/Software_Update:HowToManuallyGenerateMARFiles    
[1:07pm] kreeger: http://wiki.mozilla.org/Software_Update (is the home URL)
[1:08pm] smorgan: Um... "will need mbsdiff.exe"?
[1:09pm] smorgan: hopes Windows is not the only option...
[1:09pm] ardissone|food: "Important Note: this is is not how the official mars are created, this is just how (as a developer), you can generate mars for testing software update."
[1:10pm] ardissone|food: the tinderboxen generate them
[1:10pm] ardissone|food: so there's a "Mac" app built
[1:10pm] ardissone|food: so
[1:11pm] ardissone|food: have we reached a decision, or are we still "discussing" ?
[1:11pm] smorgan: apparently I'm the only one who has concerns about partial updates
[1:12pm] peeja: i have concerns
[1:12pm] kreeger: smorgan: whats your concerns (sorry I was in a meeting)?
[1:13pm] ardissone|food: [12:41pm] smorgan: The API is reasonable
[1:13pm] ardissone|food: [12:41pm] smorgan: However:
[1:13pm] ardissone|food: [12:41pm] smorgan: - It will need a wrapper, like everything else
[1:13pm] ardissone|food: [12:42pm] smorgan: - It puts us at the mercy of the entire Moz update creation process, which is poorly documented as far as I can tell
[1:13pm] ardissone|food: [12:42pm] smorgan: - It has unknown interaction problems with our event model, like everything else in Gecko (maybe it'll be fine, maybe not)
[1:13pm] smorgan: I have deeper concerns, really
[1:13pm] smorgan: namely
[1:13pm] smorgan: That they are more fragile by nature, complicate the testing matrix, and have limited utility given our release frequency
[1:14pm] kreeger: thats a good point, we don't release that often
[1:14pm] peeja: personally, I don't really like the idea of a process mucking about in my applications, even if it's doing it to itself
[1:14pm] peeja: unless it's actually trashing one version and replacing it with another complete package
[1:14pm] smorgan: If we released every week, sure, it'd be huge. But we don't, so the win seems very small
[1:15pm] smorgan: And having another question to ask everyone with a problem ("did you auto-update, or download the disk image") doesn't appeal to me, for example
[1:16pm] kreeger: Whats the minimum SDK for sprinkle 10.3?
[1:17pm] peeja: kreeger: sparkle 
[1:17pm] kreeger: (only an issue for branch),.... tomato tomatoes
[1:18pm] ardissone|food: so, if we're doing sparkle (and it sounds like we are), we should come up with a list of things to nail down answers on, and then someone (pink?) needs to talk to gerv/whomever about getting permission to land it in the tree, and then we can proceed with writing code
[1:19pm] pinkerton: ok. what about ss?
[1:20pm] ardissone|food: ?
[1:20pm] ardissone|food: what about him?
[1:22pm] smorgan: One other question, which is general to either solution: do we have https capabilities for this somewhere?
[1:22pm] ardissone|food: with mozUpdate, probably
[1:22pm] pinkerton: he sits next to gerv, maybe he should talk to him
[1:22pm] ardissone|food: ?
[1:22pm] ardissone|food: gerv lives in London
[1:22pm] ardissone|food: is all confused
[1:23pm] pinkerton: when did that happen?
[1:23pm] ardissone|food: gerv has always been in london
[1:23pm] ardissone|food:        http://wiki.caminobrowser.org/Development:Planning:Software_Update    
[1:24pm] ardissone|food: fill in as needed
[1:24pm] mento: meeting STILL going on?
[1:24pm] ardissone|food: mento: yes 
[1:24pm] smorgan: 'cause I'm ornery
[1:24pm] mento: let me catch up - i saw sparkle come up again
[1:25pm] smorgan: mento: if you have arguments one way or the other, now's a good time 
[1:25pm] ardissone|food: that stub doc now linked from the development home page, too (tracking docs section)
[1:25pm] kreeger left the chat room. (Quit: kreeger)
[1:25pm] mento: catching up
[1:26pm] ardissone|food: smorgan: https for downloading the update hints, or the actual updates, or something else?
[1:27pm] smorgan: Everything needs to be either https or signed (signing is an option for the Sparkle downloads, but there's a logistical issue that would come with that)
[1:28pm] mento: ok, so partial updates - how useful are they?  probably not very.
[1:28pm] smorgan: We don't want someone in a café to be pwned by a fake Camino trojan
[1:28pm] ardissone|food: right
[1:28pm] mento: do we have an idea of the space savings that firefox partials get?
[1:28pm] mento: i'm pretty sure i don't care
[1:28pm] pinkerton: probably nothing compared to the talkback symbols
[1:28pm] pinkerton: or nightly builds
[1:28pm] mento: the argument is that users don't have to download talkback symbols
[1:28pm] mento: but i still don't really care
[1:28pm] pinkerton: oh, those space savings
[1:28pm] pinkerton: me either
[1:29pm] ardissone|food: 2.0.0.5 to 2.0.0.6 partial is 400 KB
[1:29pm] ardissone|food: 425 KB, to be exact
[1:29pm] ardissone|food: er
[1:30pm] mento: that is pretty small
[1:30pm] ardissone|food: not sure if that's really that, or rc1-rc2
[1:30pm] smorgan: On my sucky DSL, that's the difference between essentially instant and a minute or two
[1:30pm] ardissone|food: yeah, that was rc1-rc2
[1:30pm] smorgan: I still don't care once every couple months
[1:31pm] ardissone|food: i don't see other partials lurking 
[1:31pm] smorgan: It's certainly a bigger deal for dial-up users though, which I acknowledge
[1:32pm] ardissone|food: no, that was .5 to .6
[1:32pm] ardissone|food: .4 to .5 was 1 mB
[1:32pm] mento: 2.0.0.4-5 partial 1.5M, 2.0.0.3-4 partial 2.2M, 2.0.0.2-3 partial 385k, 2.0.0.1-2 partial 1.8M
[1:32pm] peeja: can sparkle resume lost connections?
[1:32pm] ss: Woah, the meeting's still going on?
[1:33pm] ss: I was going to be here on time, but got in a minor car accident.
[1:33pm] ardissone|food: ss: software update
[1:33pm] ardissone|food:                           
[1:33pm] mento: i think it's just easiest to ignore partials though
[1:33pm] ardissone|food:                           
[1:33pm] mento: with partials, you worry about the app getting into a funky state during the update
[1:33pm] ss: Where are we?
[1:33pm] mento: and you need to qa the partial
[1:33pm] mento: suckiness
[1:33pm] ss: You do, that's true.
[1:33pm] ardissone|food: about to adopt sparkle, it seems
[1:33pm] ss: But it's not too bad. You just need to make sure it updates properly
[1:34pm] mento: we've got en-US and all to worry about
[1:34pm] ardissone|food: pending some requirements-checking
[1:34pm] ss: And possibly make sure others update properly as well.
[1:34pm] ardissone|food: so
[1:34pm] ss: I'm okay with Sparkle. It's not as nice as partial updates.
[1:34pm] mento: are sparkle's requirement checking features strong enough for what we need?
[1:34pm] ss: But that's life.
[1:34pm] ss: Should be, yeah.
[1:35pm] mento: say, 1.6.x updates to 2.0.x at some point unless you're running 10.3, in which case you only get 1.6.x updates
[1:35pm] ardissone|food: and if you move to 10.4, do you get re-offered 2.0.x
[1:36pm] mento: gotta provide a way to turn off update checking from inside the app (unlike adium - with apologies to cbarrett for the jab)
[1:36pm] ardissone|food: so
[1:36pm] ardissone|food: i think what we need to do
[1:36pm] mento: one thing i don't like about sparkle is that it doesn't check to see if you have permission to update the app first
[1:36pm] mento: or at least it didn't
[1:36pm] ardissone|food: seeing as it's 1:30
[1:36pm] ardissone|food: is add these things to the wiki
[1:36pm] ardissone|food: and have people investigate them, or add comments
[1:36pm] ardissone|food: depending on the type of item
[1:36pm] ardissone|food: and revisit this with facts next week 
[1:37pm] ardissone|food: everyone hear that? 
[1:37pm] mento: i don't care, i just got back from lunch 
[1:38pm] ardissone|food: and we've run really long with this, so let me just poke kreeger about Bug 384796, and poke everyone on 10.4 about the "needs project patches" backlog, and building Bug 385485 to see if X2 likes the project changes
[1:38pm] ardissone|food: crap, kreeger left 
[1:39pm] ardissone|food: ok, so everyone work on http://wiki.caminobrowser.org/Development:Planning:Software_Update and on your review queues
[1:39pm] ardissone|food: and have a good week
[1:39pm] smorgan: We need a "pay_attention?" flag for bugs 
[1:39pm] peeja left the chat room.
[1:40pm] ardissone|food: finds a board, whaps people with it
[1:40pm] ardissone|food: how's that? 
[1:42pm] pinkerton left the chat room.
Personal tools
Namespaces
Variants
Actions
Navigation
Toolbox