launchpad-dev team mailing list archive
-
launchpad-dev team
-
Mailing list archive
-
Message #00980
ReviewerMeeting20090916
#FORMAT rst
=======================
ReviewerMeeting20090916
=======================
summary
=======
* all blueprint page updates are ui=rs
* all mechanical updates are ui=rs
* these still need code reviews
* 4-space indents in css
* thumper describes his cool new `assertStatementCount()` method
logs
====
ameu
----
::
10:04:19 > barry: #startmeeting
10:04:19 < henninge: Hey barry!
10:04:20 < MootBot: Meeting started at 09:04. The chair is barry.
10:04:20 < MootBot: Commands Available: [TOPIC], [IDEA], [ACTION], [AGREED], [LINK], [VOTE]
10:04:36 > barry: hello everyone and welcome to this week's ameu reviewer's meeting. who's here today?
10:04:39 < BjornT: me
10:04:40 < abentley: me
10:04:43 < henninge: me
10:04:43 < bac: me
10:04:45 < noodles775: moi
10:04:49 < sinzui: me
10:04:52 < gary_poster: me
10:05:05 < EdwinGrubbs: me
10:05:15 < henninge: noodles775: toi?
10:05:39 < flacoste: me
10:05:39 < noodles775: henninge: nah, it's just "Australish"
10:05:44 < adeuring: me
10:05:45 < intellectronica: me
10:05:48 < henninge: ;-)
10:06:38 < EdwinGrubbs: noodles775: instead of "ping" do you say "Oi!"
10:06:57 < noodles775: You got it Ezza ;)
10:07:07 > barry: deryck, salgado cprov bigjools danilo-afk allenap mars ping
10:07:15 > barry: [TOPIC] agenda
10:07:16 < MootBot: New Topic: agenda
10:07:17 < cprov: me
10:07:20 < allenap: me
10:07:20 < bigjools: meh
10:07:23 < deryck: me, sorry
10:07:23 < salgado: me
10:07:23 > barry: * Roll call
10:07:23 > barry: * Action items
10:07:23 > barry: * UI review call update
10:07:23 > barry: * new heading/title rules [barry]
10:07:26 > barry: * 4-space indents for CSS styles [barry]
10:07:29 > barry: * Peanut gallery (anything not on the agenda)
10:07:32 > barry:
10:07:36 > barry:
10:07:42 > barry: [TOPIC] * Action items
10:07:43 < MootBot: New Topic: * Action items
10:07:46 > barry: [TOPIC] * Action items
10:07:47 < MootBot: New Topic: * Action items
10:07:54 > * barry thanks mootbot
10:08:02 > barry: * gary_poster and barry will transfer review guidelines from the old wiki and old old wiki to the new wiki
10:08:17 < gary_poster: postponed to post 3.0
10:08:24 -!- danilos!n=danilo@canonical/launchpad/danilos has joined #launchpad-meeting
10:08:25 > barry: sorry, gary_poster asked me to get mrevell a list of pages to pull over but 3.0 is swamping me
10:08:28 < danilos: me
10:08:32 > barry: what gary_poster said :)
10:08:41 > barry: * cprov to update guidelines to clarify how code sensitive to env changes should be written
10:08:52 < cprov: oi! sorry I forgot about it.
10:09:17 > barry: cprov: we'll just carry that one over. i'm hoping we'll be less stressed after this week or maybe next
10:09:21 > barry: so no worries
10:09:34 < cprov: barry: sure, thanks.
10:09:37 > barry: [TOPIC] * UI review call update
10:09:37 > barry:
10:09:38 < MootBot: New Topic: * UI review call update
10:09:57 > barry: let me see if beuno is around...
10:10:21 < intellectronica: i think he's on holiday
10:10:29 > barry: intellectronica: ah.
10:10:32 > barry: right
10:10:36 < intellectronica: one important thing we discussed is how overburdened ui reviewers are
10:11:03 > barry: intellectronica: yes. and that this is an extra-ordinary time because of the push for 3.0
10:11:15 < intellectronica: to make it a bit easier until 3.0 is out, it's ok for people to land very simple changes (mechanical conversions and such) with ui=rs
10:11:18 > barry: beuno did say he was going to start working with a few people to get them ui graduated
10:11:24 < intellectronica: i think martin has writted to the list about that
10:11:33 > barry: yep. intellectronica thanks
10:12:22 > barry: that's all i can think of. intellectronica, anything else?
10:12:47 < intellectronica: i think that's it
10:12:50 < sinzui: Does everyone know that Blueprint pages are ui=rs?
10:13:12 -!- jmux!n=jmux@xxxxxxxxxxxxxxxxxx has joined #launchpad-meeting
10:13:26 < noodles775: Woops, I just landed the sprint-add mechanical change with r=rs and ui=rs... sorry.
10:13:50 > barry: noodles775: right, still need a normal code review
10:13:56 < noodles775: Yep.
10:14:06 < bigjools: yeehaw
10:14:10 < intellectronica: sinzui: does that include volunteers? i bet there are many people out there itching to give us a hand with blueprint conversions. if that happens, can we land their changes without a proper ui review?
10:14:20 < sinzui: noodles775: That is better than me. I was juggling 4 branches yesterday. I landed one by accident. I am glad bac approved it
10:14:29 < noodles775: heh
10:15:01 > barry: i promised bac to land one thru ec2 and then promptly pqm-submitted it ;)
10:15:19 < sinzui: intellectronica: I do not have an answer
10:15:35 > barry: intellectronica: for volunteers, we still have to land the branch so i think a quick look couldn't hurt
10:15:37 < flacoste: intellectronica, sinzui: if they are mechanical changes, yes
10:15:52 < flacoste: i still think we should do a UI review for non-mechanical blueprints UI change
10:15:53 < danilos: intellectronica, I've got 12 templates converted to generic-edit for blueprints, already reviewed, should land soon
10:15:57 < flacoste: doesn't need to be much involved
10:16:06 < danilos: intellectronica, i.e. make that 12 templates removed and replaced with generic-edit.pt
10:16:08 < intellectronica: danilos: you're a star
10:16:09 < sinzui: intellectronica: blueprint 2.0 pages will not work when we release, so doing a bad conversion job fixes more than will be broken in production
10:16:18 < flacoste: there is only 12 unclaimed blueprints templates on https://dev.launchpad.net/VersionThreeDotO/BlueprintsConversion
10:17:32 > barry: moving on..
10:17:46 < sinzui: The two blueprint pages I converted yesterday had whitespace issues. text overlapped text. I had to make some markup changes to make the page layout
10:18:00 > barry: [TOPIC] * new heading/title rules [barry]
10:18:00 > barry:
10:18:01 < MootBot: New Topic: * new heading/title rules [barry]
10:18:20 > barry: i want to make sure any question you have about the new rules get answered
10:18:46 < danilos: barry, excellent
10:18:50 > barry: i know that the breadcrumbs/titles/headers need some refinement, but that will have to wait until after 3.0. the rules we have now will not change until after 3.0
10:19:05 > barry: so... ask your questions now! :)
10:19:08 < bigjools: plz barry just fix everythin kthxbye
10:19:24 < noodles775: barry: did you see deryck's email?
10:19:24 < flacoste: that's actually a job for salgado :-)
10:19:48 < deryck: barry, beuno, and I talked off list about breadcrumbs in titles a lot already
10:19:53 > barry: noodles775: yes, but only in the sense that it made my inbox grow larger. i haven't read it yet. but deryck and i spoke this morning
10:20:01 < noodles775: Great.
10:20:08 > barry: salgado is the breadcrumb man
10:20:19 < deryck: I just replied to my own mail with the results of our call
10:20:35 > barry: deryck: cool, thanks
10:20:43 < deryck: np
10:20:56 < danilos: ok, so, I am wondering about the h2/h1 issue... it seems the h2 with context.title is gone on most pages? is that expected?
10:21:04 < danilos: or, how do we make it appear again?
10:21:06 > barry: danilos: yes
10:21:14 < noodles775: Add view.label?
10:21:15 < danilos: (sorry if I didn't pay close attention to the new rules page)
10:21:27 > barry: noodles775: yes, it can be something like:
10:21:29 > barry: @property
10:21:39 > barry: def label(self): return self.context.label
10:21:40 < danilos: noodles775, right, I am looking into figuring these out for all the translations page now that we've basically got everything converted
10:22:09 > barry: noodles775: i would like to make that simpler (e.g. fall back to context.title) but not for 3.0
10:22:30 < noodles775: Yep - that'd be great!
10:22:32 < bigjools: we get rs=barry for fixing stuff like this I think you said?
10:22:39 < intellectronica: barry: maybe have a mixin that does that?
10:22:53 > barry: intellectronica: i think it can be done on LaunchpadView and/or base-layout.pt
10:23:15 > barry: bigjools: sorry, what was the "stuff like this" in that sentence?
10:23:53 < bigjools: barry: fixing existing 3.0 changed pages to conform to the new heading rules
10:24:01 > barry: bigjools: right, +1
10:24:17 < bigjools: we were waiting for your branch to land so we could do that
10:24:54 > barry: yep. now you can jfdi:)
10:25:15 < bigjools: thanks for the extra work :)
10:25:40 > barry: :)
10:25:51 < flacoste: actually, it's less work
10:26:04 < flacoste: because otherwise, there were tweaks on a per-page basis
10:26:09 < bigjools: for future changes, yes
10:26:09 < flacoste: now, it's the same thing everywhere
10:26:21 > barry: updating pagetests might not be fun
10:26:32 < * flacoste:26:38 > barry: which is actually the main reason we aren't going to change the <title> rules until post 3.0
10:27:02 < * bigjools:27:26 < sinzui: A lot of converted pages appear to be using pagetitle :(
10:28:14 > barry: please everyone, take a healthy swing with a large baseball bat at pagetitles.py
10:28:15 < sinzui: We'll know who cheated on day 1 of week 0 when pagetitles is deleted.
10:28:27 < deryck: heh
10:28:59 < * bigjools:29:15 < * sinzui!n=sinzui@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx uses a small cannonette
10:29:18 > barry: bigjools: just give it two healthy swings then
10:29:23 < bigjools: :)
10:29:35 < sinzui: um
10:29:45 > barry: anything else on titles/headings/breadcrumbs?
10:30:16 > barry: cool
10:30:17 < sinzui: our designer is not available and I have not seen a design for the front page. Are we changing the page macro and declaring victory?
10:31:19 < bigjools: while we're on the subject of UI, some of the two-column <dl>s are a bit borked if an item uses an icon
10:31:48 < sinzui: leading icon?
10:32:02 < * bigjools!n=quassel@canonical/launchpad/bigjools looks for an example
10:32:16 < bigjools: yes leading
10:33:47 < bigjools: sinzui: https://edge.launchpad.net/ubuntu/+source/kdepim/
10:34:00 < bigjools: anyway, OT for this meeting I guess
10:34:21 > barry: k, let's move on
10:34:24 > barry: [TOPIC] * 4-space indents for CSS styles [barry]
10:34:25 < MootBot: New Topic: * 4-space indents for CSS styles [barry]
10:34:48 > barry: fairly simple one i think. most of our css in styles-3-0.css are 4 space indents
10:35:01 < bigjools: JFDI
10:35:24 > barry: any objections to keeping 4 space indents?
10:35:43 > barry: 5
10:35:47 > barry: 4
10:35:48 < noodles775: Nope - guessing it was just 2-spaces from pre-minimization days?
10:35:50 > barry: 3
10:35:55 < sinzui: keeping it? I intentionally set it to 4, who put something else
10:36:10 > barry: sinzui: dunno
10:36:21 < noodles775: sinzui: there's old stuff in style.css with 2 spaces I think.
10:36:26 > barry: rs=me on fixing them for the next person to touch the file
10:36:46 > barry: 4 spaces it is
10:36:48 > barry: [TOPIC] peanut gallery
10:36:49 < MootBot: New Topic: peanut gallery
10:36:50 < sinzui: There should not be. stuff requires porting from specific CSS to generic
10:36:55 < * bigjools:37:02 > barry: bigjools: shuck it!
10:37:05 < bigjools: ok
10:37:22 < bigjools: real quick, but I sometimes see some obtuse page test code that does this:
10:37:36 < sinzui: There should not be a zillion p.<class_that_is really_useful_but_can_only_be_used_by_paragrpahs>
10:37:36 < bigjools: >>> browser.getLink("xxx").click()
10:37:36 < bigjools: >>> print browser.url
10:37:36 < bigjools: http://launchpad.dev/blah
10:37:36 < abentley: bigjools: Then take some pictures of it, and you'll have your own peanut gallery.
10:37:44 < bigjools: which causes an unnecessary page load
10:37:50 < bigjools: you can just do:
10:37:52 < bigjools: >>> print browser.getLink("xxx").url
10:37:54 < bigjools: instead
10:38:04 < bigjools: EOT
10:38:31 < BjornT: bigjools: shouldn't the test make sure that the link leads to the right page?
10:38:33 > barry: bigjools: oh, you mean if you aren't doing anything else on the target page?
10:38:47 < bigjools: BjornT: they are the exact same test
10:39:01 < sinzui: bigjools: barry: I think there is a misunderstanding here
10:39:11 < bigjools: barry: yes
10:39:12 < BjornT: bigjools: no. the first one makes sure the link isn't a 404. your version doesn't
10:39:41 < bigjools: BjornT: if the test is doing nothing else, my version is better. Loading a page and doing nothing is a bad test.
10:39:47 < sinzui: I do what bigjools is suggesting when I test that a view provides expected content, but a story is about link traversal
10:40:03 < BjornT: bigjools: well, it still depends on the intention of the test
10:40:05 < bigjools: it should ideally have a separate test for that page
10:40:09 < bigjools: BjornT: yes, agreed
10:40:18 < bigjools: but it's just something to look out for
10:40:29 < bigjools: I am on the rampage against slow tests
10:40:50 < bigjools: and this is low-hanging fruit for a big gain
10:41:06 < sinzui: bigjools: most of the slow tests are pagetests that are checking the contract of a view. Covert them to a unittest or a doctest of the view.
10:41:29 > barry: sinzui: +1 view tests rock
10:41:31 < bigjools: yes, I think that's a separate issue to this though
10:42:02 < bigjools: if another page test file tests the page, you don't need to load it somewhere else just to check a link exists
10:42:19 < bigjools: anyway, that's it
10:42:31 > barry: bigjools: cool thanks
10:42:39 > barry: 3m left. anyone else?
10:42:43 < sinzui: bigjools: right. That is why I argue we have far too many pagetests
10:43:31 < BjornT: bigjools: well, if some other test makes sure you can navigate to that page, yes. that's what really is happening there. from page A we can reach page B
10:44:33 > barry: okay, i think we're done.
10:44:36 > barry: #endmeeting
asiapac
-------
::
18:30:32 > barry: #startmeeting
18:30:33 < MootBot: Meeting started at 17:30. The chair is barry.
18:30:33 < MootBot: Commands Available: [TOPIC], [IDEA], [ACTION], [AGREED], [LINK], [VOTE]
18:30:42 > barry: hi guys
18:30:48 < rockstar: ni!
18:30:52 < wgrant: me
18:31:04 < thumper: hi
18:31:24 > barry: jml, mwhudson ping
18:31:35 < jml: hi
18:31:55 > barry: so. let's start with a recap of ameu
18:32:14 < mwhudson: hi
18:32:16 > barry: all updates to blueprint pages and all mechanical changes are ui=rs
18:32:22 > barry: they still need a code review though
18:32:30 > barry: wgrant: that includes volunteers :)
18:32:37 < mwhudson: blueprints? what's that again? :)
18:32:42 > barry: :D
18:32:50 > barry: use 4 space indents in css files
18:32:58 < wgrant: barry: Tempting, tempting...
18:33:01 > barry: that's it!
18:33:27 < mwhudson: cool, short and sweet
18:33:40 > barry: thumper: has something for today...
18:33:40 < rockstar: thumper landed a new toy today. He should talk about it.
18:33:49 < thumper: :)
18:34:08 < thumper: r9475 of devel has a new method on TestCase
18:34:15 < thumper: def assertStatementCount(self, expected_count, function, *args, **kwargs):
18:34:26 < thumper: it uses the StormStatementRecorder
18:34:31 < thumper: (also in r9475)
18:34:40 < thumper: you need to be careful about the state of the store cache
18:34:41 < jml: thumper, wow, that's a great idea :P
18:34:52 < thumper: jml: :)
18:35:20 < thumper: but it is good for asserting that you aren't hitting the DB when you don't expect to
18:35:38 < mwhudson: thumper: can you use the StormStatementRecorder to find out more about the actual queries than just the count?
18:35:46 < rockstar: thumper, can we find a way to get the information from StormStatementRecorder into every HTTPResponse for our dev server?
18:35:56 < mwhudson: although i'm having a hard time thinking about why you'd want that, now i've said it...
18:36:00 < thumper: mwhudson: if the count doesn't equal, it prints out the actual statements in the fail message
18:36:10 > barry: thumper: when you say "you need to be careful about the state of the store cache" what does that mean?
18:36:20 < mwhudson: thumper: cool
18:36:28 < thumper: barry: if you are creating objects in the test, then they are in the cache
18:36:39 < thumper: barry: so factory.makeBranch will already have the branch owner in the cache
18:36:48 < thumper: where as if you have just loaded the branch itself, it won't be
18:36:50 < thumper: so
18:36:55 < thumper: you need to do something like:
18:37:02 < thumper: store = Store.of(obj)
18:37:05 < thumper: store.flush()
18:37:07 < thumper: store.reset()
18:37:18 < thumper: reload the obj somehow using a utility
18:37:22 < thumper: then assert counts
18:37:44 < thumper: the store.reset() removes the hidden storm store attribute of the object
18:37:51 < thumper: so you can't just use: store.reload(obj)
18:37:53 < thumper: it doesn't work
18:37:59 > barry: thumper: this sounds cool even today when i'm more fully awake :) please email launchpad-dev to let people know about it and/or add it to our dev wiki
18:38:08 < thumper: this _maybe_ solvable with enough storm-fu
18:38:37 < rockstar: thumper, I think the landscape guys might like to know about it too.
18:38:52 < thumper: perhaps
18:39:19 < thumper: I'm using it to confirm that my "priming of the storm cache" actually does what I expect
18:40:45 > barry: thumper: this is cool, thanks for adding it, and please do let everyone know about
18:40:53 < thumper: ok
18:40:56 > barry: anything else guys?
18:41:04 < rockstar: Not from me.
18:41:23 < thumper: that's it from me
18:41:35 > barry: wgrant, jml, mwhudson ?
18:41:44 < jml: not from me
18:42:06 < mwhudson: nope
18:42:25 > barry: great, thanks everyone!
18:42:27 < wgrant: no
18:42:30 > barry: #endmeeting
Attachment:
signature.asc
Description: PGP signature
