/srv/irclogs.ubuntu.com/2011/04/06/#launchpad-yellow.txt

=== Ursinha-bbl is now known as Ursinha-afk
daniloswow, we've got a nice flower on the board09:02
danilosnice *yellow* flower, that is09:04
gmbWe are a delicate flower.11:35
danilosgmb, heh11:38
danilosgmb, btw, do you feel like reviewing a JS branch that lacks tests? I'd ask an OCR but I think self-reviewing non-trivial changes is not our common practice :)11:39
gmb:)11:39
gmbdanilos: Sure. Link me.11:39
danilosgmb, https://code.launchpad.net/~danilo/launchpad/add-subscription-link/+merge/5639211:40
daniloscheers :)11:40
gmbnp11:40
danilosgmb, a movie to prove that at least at one point, it all used to work: https://devpad.canonical.com/~danilo/screencasts/add-subscriptions.ogv11:40
danilos(or that I am pretty good at montaging a screencast)11:41
gmb:)11:43
gmbFair enough.11:43
gmbdanilos: 354+        '<div style="margin: 1em 0em 0em 1em"'+12:02
gmb355+            '      class="subscription-filter"></div>')12:02
gmbI'm not keen on manually setting styles.12:02
gmbCan you create a class for that and add it to the CSS?12:03
gmbOr is there a reason for doing it this way?12:03
danilosgmb, that's just moved code12:07
gmbAh, okay.12:07
gmbMaybe file a tech-debt bug for it then.12:07
danilosgmb, I am not very keen on that either, I even had to add a hack to fix the problem on removal where margin-top was left to 1em12:07
gmbEch.12:07
danilosgmb, basically, I didn't add any plain-HTML node creation, I just factored it out into smaller and yet smaller methods :)12:08
gmbRight :)12:08
gmbdanilos: So, file a tech-debt bug for it; it can be taken care of by a maintenance team (probably us).12:08
danilosgmb, we might decide to fix that sooner rather than later, but sure, I'll file a bug12:09
gmbCool.12:09
gmbOtherwise, r=me12:09
* gmb -> lunch12:09
danilosgmb, thanks12:19
gary_posterdanilos, gmb, I commented on https://bugs.launchpad.net/launchpad/+bug/752406 fwiw, in case you want to add some voices of dissent :-)13:12
_mup_Bug #752406: Do not set style directly in structural-subscriptions.js <story-better-bug-notification> <tech-debt> <Launchpad itself:Triaged> < https://launchpad.net/bugs/752406 >13:12
danilosgary_poster, well, for one thing, it's very hard (or very ugly) to test stuff with direct style setting, whereas with classes it's so much easier to test for the (non-)presence of a class, imho13:16
danilosgary_poster, that's one big reason for not setting the style directly13:16
bacgary_poster, danilos, gmb: regarding the in-place style vs CSS issue, I'll just note there is not agreement or a consistent policy across LP.  sinzui argues adding items to the CSS should only be done for general things that are to be shared; one-offs should be done locally.  OTOH I've gotten a lot of review comments urging the other practice, the last from thumper, I believe13:16
* bac likes danilos testability argument13:17
gary_posterYeah, sounds like I'm in sinzui's camp.  I could be more convinced of the other approach if we had structural-subscription.js-specific css file, but dumping it in a giganto shared one makes things hard IMO13:18
daniloswe perhaps should not put everything inside a single CSS file though, but we do that because fetching stuff over SSL is slow (even small CSS files might cause a hit)13:18
gary_posterwe can compress the CSS AIUI13:18
danilosgary_poster, that I'd counter-argument with perhaps we can have multiple CSS files that get combined ala JS stuff we do13:18
gary_posterI don't buy the testability argument TBH, at least for the kinds of things I know we set style for.  YUI.getStyle13:19
gary_posteris easy13:19
gary_posterand not necessary to test if you are just saying (this stuff should (always) float right)13:19
danilosgary_poster, I don't like having a test that asserts that YUI.getStyle("margin-top") is zero or something like that, because even if you set it in ems, it might return pxs and such and13:19
danilosgary_poster, well, some of those things are "always, until we decide to hide it"; on removal, subscription filters kept 1em margin-top that was set as <div style="..."> because they were only hidden13:20
danilosgary_poster, when I fix a bug like that, how do I add a test that makes sure it never comes back?13:21
gary_posterclasses don't hep answer that particular question, fwiw13:21
danilosI mean, we don't test CSS itself, but at least it'd be outside the code, and we'd know the code does what it's supposed to13:21
gary_posterhelp13:21
danilosgary_poster, imo they do, but I guess that's an argument we can have some other time :)13:22
gary_poster<shrug> we disagree but I'm happy to defer on this :-)  What do you think we should do about the bug?13:22
danilosI'd like us to solve it, but I'd also like us not to introduce mess into CSS; how about we add our own CSS file and establish a practice for that? (bin/combine_css has a simple names=[] variable where we can manually list it for now)13:23
daniloswhere "solve" try to use more of the CSS classes instead of setting style directly13:24
danilos"solve" ==13:24
gmbComing late to the party here I think  that danilos suggestion is a good one.13:24
danilosgmb, you must be the sound of reason since at this point of the party, all the booze is gone :)13:25
gmb:)13:25
daniloss/sound/voice/ (see what partying does to a man? :))13:25
gary_posterheh13:25
gary_posterthat sounds good to me danilos.  Do you mind that not being "high"?  I'm happy to have it as a "fix it if you want to" card in the backlog though13:26
gary_posterbac benji (who is not here) danilos gmb: please make sure kanban is up to date now-ish13:27
gmbyup13:27
danilosgary_poster, yeah, non-high is just fine13:27
gary_postercool13:27
gary_posterI'll up it to medium, and make a backlog card13:27
danilosgary_poster, thanks13:30
gary_posterbac, you on Skype?13:30
bacoops13:31
bacon now13:31
danilossorry about that, new, better headset is on order :)13:34
gary_poster:-)13:35
=== Ursinha-afk is now known as Ursinha
* danilos -> lunch14:05
* gary_poster to car garage. back soon.14:15
* benji is back from the dentist.14:18
* gary_poster back too.14:49
gmbgary_poster: Any objection to me taking bug 424849?14:57
_mup_Bug #424849: Launchpad should batch attachment notification emails <lp-bugs> <story-better-bug-notification> <story-better-notification-sending> <Launchpad itself:Triaged by gmb> <apport (Ubuntu):Invalid> < https://launchpad.net/bugs/424849 >14:57
gmb(I think I know how to do it, so it seems the post obvious option)14:57
gary_postergmb, ok, fire away14:57
gmb*most14:57
gmbCool.14:57
gary_posterbac, benji, danilos, gmb, I want us to try and do a better job tracking what is released.  I'll try to lead this.  For now, I've made the following changes to the kanban board.15:38
gary_poster1) I split up Deployment: Ready into three sub-lanes.15:38
gary_posterIf you land something into db-devel put it into the db-deploy lane.  If it is a normal devel landing, put it in no-downtime.15:38
gary_posterIf you've landed something in db-devel inbetween a db-devel deploy code freeze and a db-devel deployment, like we did with the "mute team subscriptions" stuff recently, you can put it in "Next db-deploy".15:38
gary_poster2) I moved all of our "ready" cards to "done-done" because of today's deploy.  Please move cards back to "no-downtime" (or other) if you know they are not actually deployed.15:38
gary_posterthat's it, thanks.15:38
gmbgary_poster: Righto. Makes sense; I admit it was a bit unclear how to handle the db-devel versus devel stuff.15:39
gary_postercool15:39
bacgary_poster: i'm unhappy with the way i've been working on cards with no bugs and having to use no-qa.  i'd like to start filing more bugs for task tracking.  you ok with that?15:40
gary_posterbac, absolutely.  I sometimes make cards when I land, just for the qa.  Do whatever works for you.  The tension between kanban cards and bugs is not lost on me. :-/15:41
baca bug seems like a too heavy artifact for a lot of things but without one we lose qa-ness15:41
gary_posterBut I mostly need the kanban cards15:41
gary_posterright15:41
gary_posterbac, sorry, I meant "I sometimes make *bugs when I land, just for the qa."15:41
danilosgary_poster, ack15:43
gary_posterthanks15:43
gary_posterbenji, I want to know how smart the code is that handles IJSONRequestCache(request).objects--in particular, if it reuses the same data structure in the JSON if the same object is used multiple times, via relying on some sort of JS post-processing.  I doubt it, but it woud be silly to work around that if it were handled already.16:14
gary_posterSo you know where I should look?16:14
gary_posterDo16:14
* benji looks.16:15
benjigary_poster: since you have to add objects by name, and the cache is a simple dictionary, duplicates aren't eliminated16:22
benji(see src/lazr/restful/jsoncache.py)16:22
gary_postergot it, thanks benji16:22
* gmb -> brain stall; increasing tea reserves16:24
* gary_poster back to garage16:27
* danilos off17:18
gmbYou know that feeling when you've just spent half an hour on a really, really good test, only to suddenly realise it doesn't exercise the code you want it to? Yeah, that.17:33
danilosactually, my branch failed with some test failures, resubmitted with fixes, and pushing my merged-with-db-devel to lp:~danilo/launchpad/db-add-subscription-link (includes some devel revisions so I'll ec2 land that one as well)17:51
danilosgary_poster, actually, I can't land that easily without an MP, Gary, can you please confirm that all mute-related functionality works as expected and then ec2 land it on db-devel if so?18:01
danilos(I was supposed to be out over an hour ago, and I just wanted to get this finished, so db-devel doesn't hit a conflict that's hard to resolve)18:02
danilosgary_poster, I also didn't check any tests, I am sure some have broke18:02
* danilos -> really off18:02
gary_posterdanilos, ack, will do, thanks18:08
gary_posterThe JS tests on danilos' branch pass...18:22
* bac -> lunch18:27
gary_posterqa on his branch works...18:27
benjiWARNING: GIANT, MAN-EATING BIKESHEAD AHEAD18:40
benjiThe subscription listing wording for the mute link seems quite odd: "Do not receive emails from this subscription".18:40
benjisuggestion: "Stop emails about this bug"18:41
benjiend-of-bikeshead18:41
gary_posterThe problem is that your simplification is not true18:41
gary_posterWe have a concept of muting bugs18:41
gary_posterwell, emails about a bug18:41
gary_posterwhich is handled elsewhere18:42
gary_posterthis is for the case in which you are in a team18:42
gary_posterwhich has a subscription to [something[18:42
benjioh, that's not muting18:42
gary_posterright18:42
gary_posterthat's why I didn't call it muting in the interface18:42
benjiI'm sure coming up with good words for that was (is) hard, but I'm still complaining about them ;)18:42
gary_postercomplaining is only mildly interesting without a compelling solution.  :-) Alternatively, make a card for "ask Huw about wording" for that feature18:43
benjilast shed-swatch: "Stop emails from this subscription"18:44
benjiI'll make a card and shepherd it.18:45
gary_posterI'd be fine with that wording benji.  That said, I'm not clear on it being a win, but that's not any kind of endorsement of or belief in the current wording.  User testing is the proof of the pudding.18:46
gary_posterFeel free to just make the change.18:47
benjigary_poster: ok, the "Stop..." wording assuages [that's a hard word to spell] my inner-shed18:48
gary_posterI feel kinda blech18:50
gary_posterKaryn is telling me to take a half day sick18:50
gary_posterI think I might do that18:50
gary_posterguys, I got permission for a sick day18:52
gary_posterdanilos branch is in ec2 land18:52
gary_postersick half-day I should say18:52
gary_posterHopefully I'll feel better tomorry18:52
gary_posterTTYL18:52
baci like how the kanban favicon now looks like it was custom-made for the yellow squad21:57
=== Ursinha is now known as Ursinha-afk

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!