Status Meetings:2006-10-25:Log

From Camino Wiki
Jump to navigation Jump to search
12:04 <@ss> Shall we begin?
12:04 <@kreeger> sure
12:04 <@ss> Camino 1.1 was released after our meeting last week
12:04 <@ss> Success!
12:04 <@ss> Err...
12:04 <@ss> Camino 1.1a1
12:04 <@ss> We've had a few good bugs come out of it, but not that many yet.
12:04 <@ss> Good feedback though, overall.
12:05 <@ss> We do have a topcrasher though that's random hex strings.
12:05 <@Wevah> "yay"
12:05 <@smorgan> That links to a recently fixed bug
12:05 <@smorgan> Is that a lie?
12:05 <@ss> That bug wasn't "fixed"
12:05 <@ss> It was resolved WFM because it was in JEP.
12:05 <@ss> But I'm not sure that's true.
12:06 <@smorgan> "Appears fixed by the new JEP"
12:06 <@ss> fwiw, bug 295299 also contains the hex string
12:06 <@ss> That was also resolved WFM.
12:06 <@ss> (By me even)
12:07 <@ss> It's worth some investigation anyway.
12:07 <@ss> We should probably file a bug for it based on the comment users are leaving.
12:08 <@ss> Beyond that, looking forward to Camino 1.1...
12:08 <@ss> kreeger: Any update on the kqueue stuff?
12:08 <@kreeger> about done, i should have the time to finish it up on Saturday
12:09 <@ss> Sounds good
12:09 <@ss> froodian finally got more info on the bookmark destruction bug.
12:09 <@ss> But it doesn't seem to be useful info (?)
12:10 <@froodian> well, possibly
12:10 <@froodian> it's certainly confusing info
12:10 <@smorgan> I still think it looks like an outside force is responsible for both the 
                 0-byteness and the failed write of the side file
12:10 <@froodian> it certainly looks like it
12:11 <@ss> There was talk a while ago about it potentially being Spotlight, iirc.
12:11 <@froodian> but it's not a 3rd party Camino thing
12:11 <@ss> Is there any basis to that talk?
12:11 <@smorgan> Spotlight doesn't write files, or lock them at the Cocoa level
12:11 <@pinkerton> what was the firefox interaction though?
12:12 <@smorgan> Another possibility is that 'outside force' is something Gecko-y being done to 
                 the profile folder.
12:12 <@smorgan> They were using a carbon call
12:12 <@smorgan> IIRC, it was restricted to that level
12:12 <@hwaara> ?
12:12 <@hwaara> what and who? :)
12:13 <@hwaara> and why?
12:13 <@smorgan> Why would anything come hose the profile folder?  Presumably a bug
12:13 <@smorgan> Just tossing it out there
12:13 <@ss> We definitely need more research into it and such. There seem to be more and more 
            reports of it.
12:14 <@pinkerton> bizarro
12:14  * froodian nods
12:14 <@smorgan> That's all well and good to say, but it's not really a meaningful statement
12:14 <@ss> Not as though it's getting worse, maybe just more outspoken people.
12:14 <@pinkerton> brb, stepping away to put something the microwave
12:16 <@kreeger> could be something with the gecko writing to the profile dir (like smorgan 
                 said) maybe anything the likes of history or cache
12:17 <@kreeger> that might be a bit tricky to trace down
12:17 <@pinkerton> writing to the same dir shouldn't be a problem.
12:19 <@smorgan> I'll try writing something switching to writeToFile:options:error:
12:19 -!- hwaara_ has joined #camino-mtg
12:19 < hwaara_> we don't use the same bookmarks format as firefox, right?
12:19 < hwaara_> or do we?
12:19 <@Wevah> we don't
12:20 < hwaara_> so then they can't affect it
12:20 <@pinkerton> yeah totally different
12:20 < hwaara_> unless they know the file name, which I'm pretty sure we don't expose via 
                 AppDirServiceProvider, iirc
12:20 -!- hwaara has quit [Ping timeout]
12:20 -!- hwaara_ is now known as hwaara
12:20 <@smorgan> I'm just saying there's something weird happening at the dir level at that 
                 point, since writing a new file fails
12:20 <@Wevah> even if they did, they shouldn't be doing stuff with it since our file is a plist
12:21 <@pinkerton> could it be a permissions problem?
12:21 <@pinkerton> can we add logging to write out the permissions of the dir?
12:21 <@smorgan> Like I said, I'll try to rig up something that we'll get error data out of on 
                 why it failed
12:21 < hwaara> do we have bm backup?
12:21 <@froodian> no
12:21 <@smorgan> Flatten the whole thing to NSData and using the API that gives us an NSError
12:22 < hwaara> if we add it, the problem is not as bad (assuming it's something like an OS bug 
                that we'll have trouble solving anyway)
12:22 < hwaara> right?
12:22 < hwaara> I forget, are the bookmarks blown away, or what
12:22 <@smorgan> We have no idea how they are being truncated
12:23 <@smorgan> So we have no idea whether the same thing would happen to a backup
12:23 <@pinkerton> right
12:23 < hwaara> I mean, the actual bug for the end-user
12:23 < hwaara> is the result that all bookmarks are lost?
12:23 -!- Pinolo has quit [Ping timeout]
12:23 <@smorgan> Yes
12:23 < hwaara> I know firefox always has a few bookmark copies
12:23 <@smorgan> But if we have a way to ensure files don't get emptied, we don't need a backup
12:24 <@smorgan> And if we don't then what's the point?
12:24 <@smorgan> Because something is truncating our bookmark files
12:24 < hwaara> another idea that might or might not work, is to lock the bookmarks.plist
12:24 < hwaara> and unlock it before we write
12:24 < hwaara> then re-lock
12:26 <@pinkerton> can we get a full system profile dump from the guy who saw the latest suckage?
12:26 <@pinkerton> maybe he is running something he's unaware of in the bg
12:26 <@froodian> (she)
12:26 <@froodian> how do I ask for that?
12:26 <@froodian> i'm not sure what constitutes a full system profile dump
12:26 < hwaara> Go to the "About this computer" app
12:26 < hwaara> there's a way to make a profile for the computer
12:27 <@pinkerton> click "more info", it'll open system profiler
12:27 <@pinkerton> save that profile and send it
12:27 <@froodian> k
12:27 <@froodian> i'll email her
12:28 <@pinkerton> thanks
12:28 <@froodian> next?
12:29 <@maxr> yes
12:29 <@maxr> ss had to step out
12:29 <@pinkerton> slacker
12:30 <@froodian> so, tinderboxen
12:30 <@smorgan> What does that item you skipped mean?
12:30 <@froodian> oh, i didn't see it
12:31 <@froodian> we're not focused enough on 1.1-targetted bugs
12:31 <@smorgan> I fix mostly 1.1 bugs
12:31 <@froodian> especially regression and follow-ups
12:31 <@froodian> yeah :)
12:31 <@froodian> i'm partially to blame
12:31 <@smorgan> I'm looking at a number at the bottom that says most of the bugs fixed were 1.1
12:31 <@smorgan> And a number of the bugs that were just sr'd are 1.1
12:32 <@froodian> yeah
12:32 <@smorgan> So I'm wondering how that's a bad ratio
12:32 <@froodian> so i guess it should be more like "1.1 bugs are going well, feature follow-ups 
                  need more focus"
12:33 < hwaara> I don't see delliot around for tab things, unfortunately
12:33 <@froodian> we'll get there
12:33 <@froodian> first
12:33 <@froodian> tboxen
12:33 < hwaara> I thought we were talking about 1.1?
12:33 < hwaara> oh well
12:33 <@froodian> (there's a thing on tab things specifically for specific bugs)
12:33 <@froodian> but tab scrolling isn't 1.1
12:33 <@froodian> (probably)
12:34 < hwaara> ok, that's probably a good thing
12:34 <@froodian> we got boxset back, pawn's been having problems
12:34 <@froodian> the biggest of which is no grapsh
12:34 <@pinkerton> what happened to it?
12:35 <@pinkerton> did they reinstall something?
12:35 <@ss> Boxset was unplugged or something.
12:35 <@ss> rhelmer said it was "off" when he went to check on it.
12:35 <@ss> I don't know much more than that.
12:35 < hwaara> any status on new tboxen? :)
12:36 <@pinkerton> i'm so done with these fucktards
12:36 <@ss> New? No. They're going to be upgrading them though.
12:36 <@pinkerton> how can you just UNPLUG a tinderbox?
12:36 <@ss> pinkerton: I'm not 100% that it was unplugged.
12:36 <@ss> It might have just been powered down for no reason.
12:36 <@pinkerton> mmm
12:36 <@pinkerton> even better!
12:36 <@ss> But either way it's bad.
12:36 <@Wevah> .\m/
12:36 <@Wevah> very
12:37 <@ss> It probably got powered down for a new OS but never got the new OS
12:37 <@ss> Or something along those lines.
12:37 <@pinkerton> "yay"
12:37 <@ss> In any case, they are going to be upgrading them which is why the xserve is there.
12:38 <@ss> Beyond that, we need better build graphs.
12:38 <@ss> The main issue is filed and mento has CCed the appropriate people.
12:38 <@ss> It's just a matter of when they'll get to it.
12:38 <@ss> Fx2 too a lot of build time
12:38 <@ss> And they still have Fx 1.5.0.8 coming in a week
12:38 <@ss> So that might take more time away.
12:39 <@ss> Hopefully after that we can steal away more of their time.
12:39 <@pinkerton> ok
12:39 <@ss> As far as our queues go, we're looking "alright", but there still are some hold ups 
            (mostly r not sr)
12:40 <@ss> pinkerton likes to go sr crazy every few days. ;)
12:40 <@ss> I want to clarify something...
12:40 <@ss> pinkerton: Can josh do SRs for Camino?
12:40 <@ss> I was never clear on that.
12:41 <@pinkerton> ss: yes, if he feels comfortable
12:41 < hwaara> doesn't it look weird if someone writes sr= when they're not a mozilla 
                super-reviewers?
12:41 <@ss> Alright, sounds good.
12:41 <@pinkerton> naw, i was doing it for years before they made me an sr against my will
12:41 < hwaara> heh
12:41 <@ss> So josh has said he's up for more r/sr
12:41 <@ss> Give him some lovin'
12:42 <@pinkerton> ok, i also wanted to add something here
12:42 <@ss> (But not too much lovin'; we don't want to scare him away)
12:42 < hwaara> josh seems pretty busy with fx3 :)
12:42 <@pinkerton> ha
12:42 < hwaara> he's the only full-time guy with cocoa all over his hands
12:42 <@Wevah> yeh
12:43 <@pinkerton> since i see the next few months being even more increasingly busy, i think 
                   it's time to start training someone to do camino-level SRs
12:43 <@pinkerton> smorgan and I have talked and he's stepped up to take on that role
12:43 <@pinkerton> i picked stuart because he's within easy kicking range when he messes up 
                   (just kidding, mostly)
12:44 <@ss> When you say "training"... ?
12:44 <@pinkerton> he's been doing sr-level code reviews for a while now, i feel comfortable at 
                   the level he's been reviewing code
12:44 <@Wevah> wfm
12:44 <@froodian> cool :)
12:44 <@smorgan> ss: It's like, pink does real sr's, and I sweep the floor ;)
12:44 <@pinkerton> for small things, things that i would probably rubber-stamp anyway (that's 
                   what rs=pink means), he's available
12:44 <@ss> smorgan: Yeah. ;)
12:44 <@pinkerton> for bigger things, we'll tag-team as he gets his feet wet
12:45 <@pinkerton> for large things, mento, smfr, or i are still the right contact for now
12:45 <@ss> Sounds good
12:45 <@Wevah> haha
12:45 <@pinkerton> i'll be relying on his judgement to some extent on when he feels something is 
                   bigger than his expertice, so bear with us while we figure that out
12:46 <@pinkerton> and certainly anything with huge amounts of gecko changes will need extra 
                   checking
12:46 <@pinkerton> i hope this will free up the long lists of small reviews that are sitting 
                   around until i get a chance to do them
12:46 <@ss> Naturally. Anything with Gecko changes needs real SR
12:46 <@pinkerton> though somone will still have to sr his patches :)
12:47 <@smorgan> ss: not if it's lots of Gecko use in Camino-only code
12:47 <@ss> Ah...
12:47  * ss read that differently
12:48 <@pinkerton> let me know if this isn't helping and we can try something different, though 
                   i think this is stiill a big step in the right direction for the future
12:48  * froodian can dig it
12:48 < hwaara> sure
12:48 <@ss> We could, of course, do this the Fx way.
12:48 <@ss> Or the Fx/toolkit way
12:49 <@ss> One r from a "peer" where you define what peers are.
12:49 < hwaara> that's what I meant when I said it's weird with camino-specific r's, instead of 
                just defining "peers"
12:49 < hwaara> er, camino-specific sr's
12:49 <@ss> But that's probably a conversation for another time, if this doesn't work out well 
            enough
12:49 <@pinkerton> yeah, but the tools still refer to sr
12:49 < hwaara> where review from one peer is usually enough for simple things, and you want 
                more for bigger changes
12:49 <@ss> pinkerton: We can easily change that to first-review and second-review
12:50 <@ss> I could have timeless do that in a matter of minutes (if he's around)
12:50 < hwaara> I think sr implies a lot of things, given the super-reviewer guideliens on 
                mozilla.org etc
12:50 <@kreeger> lunch
12:50 <@pinkerton> my concern is "what's simple enough" and there are still small changes with 
                   which i find disagreement
12:50 -!- kreeger is now known as kreeger-lunch
12:50 <@ss> pinkerton: They also have ui-review
12:50 <@ss> For changes to UI and such.
12:51 <@ss> And UI review can only come from one or two people.
12:51 <@ss> (I believe beltzner and mconnor atm)
12:51  * froodian likes the sr model
12:51 <@pinkerton> let's stick with the sr model for the time being, i'm open to asjusting the 
                   terminology if things get to weird
12:51 <@ss> Yeah, just something to think about for the future.
12:51 !sand.mozilla.org ss invited thebot into the channel.
12:51 <@pinkerton> noted
12:51 -!- thebot has joined #camino-mtg
12:51 <@ss> bug 355490
12:52 < thebot> ss: Bug https://bugzilla.mozilla.org/show_bug.cgi?id=355490 nor, --, Camino1.2, 
                nobody@mozilla.org, NEW, Need a way to access all tabs quickly from the tab bar
12:52 <@ss> For that bug, what exactly do we want to do?
12:52 <@pinkerton> i gave stuart a suggestion last week
12:52 <@pinkerton> did he share?
12:53 <@froodian> not that I recall
12:53 < hwaara> how about click-hold?
12:53  * smorgan didn't get around to it
12:53 <@pinkerton> heh
12:53 <@froodian> click-hold should scroll
12:53 <@froodian> like a scroll button
12:53 <@froodian> (in fact, it does)
12:53 <@smorgan> A widget on the right side with the big menu delliot's talked about
12:53 < hwaara> oh right ;)
12:53 <@pinkerton> why not just have a popup menu where the mouse is on the item of the current 
                   tab. tabs before are above it in the menu, tabs after are below it
12:53 <@froodian> the major issue here is whether it should be added functionality to existing 
                  UI or new UI
12:54 <@froodian> pinkerton: i like that, i think
12:54 <@smorgan> It's the single-menu that's been kicked around, but solving the issues I had
12:54 <@pinkerton> with a widget on the rightside to show said menu, rather than overload the 
                   buttons
12:54 < hwaara> pinkerton: not sure I understand it
12:54 <@smorgan> and the big thing is that in needn't be perfect
12:54 <@smorgan> It needs to be done so everything can go no branch
12:54 <@pinkerton> hwaara: it works like the popup select menus. the currently selected item is 
                   shown under the mouse
12:54 <@ss> Is that how Fx2 does it?
12:55 <@smorgan> Before our lives become insane
12:55 <@froodian> i'm just not sure i like the "widget on the rightside" thing.  I mean, i guess 
                  it's starting to form as a standard for it...
12:55 <@smorgan> Although not until 1.1 is out, obviously
12:55 <@froodian> but it seems strange and lopsided
12:55 <@pinkerton> froodian: me either, but there's not much other choice
12:55 <@froodian> could we have it be the context menu (same) from either scroll-button?
12:55 <@froodian> or is that too confusing?
12:55 < hwaara> where would the menu button be?
12:55 <@smorgan> People won't find it, is the concern with context menus
12:55 <@froodian> yeah
12:55 <@froodian> :(
12:57 <@froodian> well, i'd be ok with it like that.  it seems less than ideal to me, but it's 
                  way better than what we have now
12:57 <@ss> Hm... could context menus be secondary UI then?
12:57 <@ss> I can see them being beneficial.
12:57  * hwaara would like a simple mockup or something, as he don't grok the descriptions in 
          words
12:58 <@pinkerton> can someone who does udnerstand work with hicks to get some mockups?
12:58 <@froodian> there's something "like" it
12:58 <@froodian> http://photos1.blogger.com/blogger/1961/2708/1600/demo.jpg
12:58 <@froodian> but it'd have the selected item be the "root" of the menu, and stuff to the 
                  left above it, and stuff to the right below it
12:58 <@pinkerton> problem is, that's a popdown, always rooted at the same place
12:58 <@pinkerton> we don't want that
12:58 <@froodian> right
12:59 <@pinkerton> but yeah, for the overflow icon idea
12:59 <@pinkerton> it's not the best solution, but it unsticks us
12:59 <@ss> While we're on this bug... can we talk about the "Tab palette"?
12:59 <@pinkerton> what we have now sucks for many and his prohibiting many any real progress 
                   for getting it on the branch
12:59 <@smorgan> Are we branching for 1.1?  If not, I guess that blocks us too
13:00 < hwaara> I wish the release after 1.1 had some more radical tab preview thingy
13:00 < hwaara> like tab palette, or expos?
13:00 <@pinkerton> i would love tabspose
13:00 < hwaara> along with delliot's stuff
13:00 <@smorgan> Unless we plan to get this in for 1.1, but that seems a bit scary
13:00 <@froodian> let's get tab scrolling first, i say ;)
13:00 <@froodian> smorgan: i agree.  1.1 should block it IMO
13:00 < hwaara> we could make a release that is mostly tab-related enhancement
13:00 < hwaara> s
13:01 <@smorgan> Let's not plan 1.x for values of x > 1 right now
13:01 <@froodian> heh
13:01 < hwaara> bah
13:01 <@froodian> another question is, who is going to implement it?
13:01 <@froodian> will delliott?
13:01 <@froodian> i wish he were here to ask that
13:01 < hwaara> I don't think it's safe to assume that
13:01 <@ss> He hasn't been around much as of late.
13:01 < hwaara> if you want to work on something, just do it
13:01 <@ss> I'd say take it from him unless he says otherwise.
13:02 <@Wevah> yeah
13:02 <@froodian> k
13:02 <@ss> Anything else?
13:02 <@froodian> i think pink is right though, we should get hicksian mockups first
13:02 <@ss> CC him and ask. He'll be more than willing.
13:02 <@froodian> ok
13:03 <@froodian> Shit, i'm late
13:03 -!- froodian is now known as froodian|classes
13:03 <@pinkerton> i think getting him something to actually move forward with will help, rather 
                   than just random ideas prone to controversy
13:03 < hwaara> while you're at it, ask if hicks has any other crazy tab-expose ideas
13:03 < hwaara> tab-preview, or whatever
13:03 <@smorgan> Ask that in a completely different bug though
13:03 <@smorgan> The one about tab expose
13:03 < hwaara> yes
13:04 <@smorgan> bug 343296 real quick, if we are done with that...
13:04 < thebot> smorgan: Bug https://bugzilla.mozilla.org/show_bug.cgi?id=343296 nor, --, 
                Camino1.1, stuart.morgan@alumni.case.edu, ASSI, Spell-checker requires text 
                cursor to be in word to get context menu
13:04 <@smorgan> I think hwaara is looking into doing this right (yes?)
13:05 < hwaara> I've tried
13:05 <@smorgan> But I'm inclined to get the hack in short-term if that doesn't happen soon
13:05 <@smorgan> Because right now, everyone things spelling is broken
13:05 <@smorgan> I'd rather have a hack and a bug for doing it right then ship another 
                 alpha/beta where spelling "doesn't work"
13:06 < hwaara> isn't there a risk the fake doubleclick will go to the wrong location?
13:06 <@smorgan> ?
13:06 < hwaara> since you're not operating on the actual click event, if I understand it right
13:07 <@smorgan> It forges clicks at the right-click location
13:08 < hwaara> smorgan: can you make sure there's a bug filed for fixing it right, if we get 
                that in?
13:08 < hwaara> I just don't want it to get in there, and then we move on
13:08 <@smorgan> Yeah, we'd definitely want a bug
13:08 <@smorgan> pinkerton, thoughts?
13:09 <@pinkerton> can we hack it better with smfr's suggestion, or did that not pan out?
13:09  * smorgan looks at hwaara
13:10 < hwaara> maybe if there's an exposed API for selecting what double-clicking would select?
13:10 < hwaara> like the current word, or whatever
13:10 <@smorgan> I flailed around in gecko for a while before doing the hack, and got no-where.  
                 I can do so again starting from smfr's suggestion.
13:11 < hwaara> that sounds like a "good" hack, if any hack can be that :)
13:11 -!- maxr [maxr@moz-D0C45A82.dreamhost.com] has left #camino-mtg []
13:11 <@smorgan> I'm really unfamiliar with all of that code, so I'm not likely to be a good 
                 person to find the right way.
13:11 <@pinkerton> yeah, seems like firing a command is the better "hack"
13:11 < hwaara> I'm also gonna see again if there's some way to do that in gecko, once and gor 
                all
13:13 < hwaara> I mean, fix the real bug
13:13 < hwaara> there are so many related bugs to how textfields should work on mac, anyway
13:15 <@smorgan> Are we done?
13:16 <@ss> Yes.
13:16 <@ss> See you all next week