fluid-work IRC Logs-2011-09-16

[10:53:20 CDT(-0500)] <Justin_o> fluid-everyone: here's an update about what blockers we have.. i had sent one out to the list the other day too but it looks like it didn't make it through
[10:53:29 CDT(-0500)] <Justin_o> FLUID-4445
[10:53:43 CDT(-0500)] <Justin_o> FLUID-4444 (pull request)
[10:53:51 CDT(-0500)] <Justin_o> FLUID-4443
[10:53:56 CDT(-0500)] <Justin_o> FLUID-4382
[10:54:05 CDT(-0500)] <Justin_o> FLUID-4453 (pull request)
[11:02:39 CDT(-0500)] <Justin_o> Bosmon: so looking at 1.3.1 it didn't exhibit the problem you mentioned in FLUID-4453. I think this is because the insertAnchor code was pretty different.
[11:02:44 CDT(-0500)] <Bosmon> Ok
[11:02:50 CDT(-0500)] <Bosmon> That's a good note for our testing at least
[11:02:52 CDT(-0500)] <Justin_o> i'll find you the code
[11:03:02 CDT(-0500)] <Bosmon> That would be interesting
[11:03:03 CDT(-0500)] <Justin_o> yes
[11:03:22 CDT(-0500)] <Bosmon> I mean, it's clear that IE6 throws a fit only on some manipulation of the DOM, and not all
[11:03:59 CDT(-0500)] <Justin_o> Bosmon: here's the current code in the repo now https://github.com/fluid-project/infusion/blob/master/src/webapp/components/tableOfContents/js/TableOfContents.js#L27-32
[11:04:08 CDT(-0500)] <Bosmon> My sense from looking at what failed (inserting anchors) and what worked (inserting pre-rendered markup using innerHTML) was that IE6 would explode on any attempt to create an explicit fresh DOM node from outside the iframe
[11:04:17 CDT(-0500)] <Bosmon> That is, by using a jQuery bound to a different document
[11:04:38 CDT(-0500)] <Bosmon> There are reports that this will fail in general to some extent in all browsers, which is why we went for the "hidden world" strategy
[11:04:40 CDT(-0500)] <Justin_o> Bosmon: here's what was there before https://github.com/fluid-project/infusion/blob/infusion-1.3.1/src/webapp/components/tableOfContents/js/TableOfContents.js#L32-35
[11:05:42 CDT(-0500)] <Bosmon> Ah wow
[11:05:47 CDT(-0500)] <Bosmon> It explicitly used "ownerDocument" before
[11:05:53 CDT(-0500)] <Bosmon> That isn't dissimilar to the fix I made
[11:06:17 CDT(-0500)] <Bosmon> Although much shorter
[11:06:45 CDT(-0500)] <Bosmon> Much neater, really - I can revert my fix in favour of that approach
[11:06:53 CDT(-0500)] <Bosmon> Since I made quite a lot of mess trying to locate the "correct" jQuery
[11:07:34 CDT(-0500)] <Justin_o> Bosmon:
[11:07:39 CDT(-0500)] <Bosmon> There should really have been a comment on that tricksy 1.3.1-level code explaining why it used that strategy
[11:07:52 CDT(-0500)] <Bosmon> Then you would have known not to trash it during the refactoring
[11:08:07 CDT(-0500)] <Justin_o> Bosmon: that's true.. unless the code was just like that by coincidence
[11:08:14 CDT(-0500)] <Bosmon> I think that is very unlikely
[11:08:31 CDT(-0500)] <Bosmon> That code was probably the result of some painful experimentation
[11:09:28 CDT(-0500)] <Bosmon> And then the person responsible, after they got it working, left it there with the bland assertion... "it works"
[11:09:58 CDT(-0500)] <Justin_o>
[11:10:11 CDT(-0500)] <Bosmon> The lack of useful test cases also contributed
[11:10:14 CDT(-0500)] <Justin_o> we've come a long way with our code quality
[11:10:17 CDT(-0500)] <Bosmon> I will try to produce one today
[11:10:24 CDT(-0500)] <Justin_o> Bosmon: thanks.. that's excellent
[11:10:31 CDT(-0500)] <Bosmon> Although making this kind of "cross-frame test case" will be pretty annoying
[11:11:06 CDT(-0500)] <Justin_o> Bosmon: yes.. i think so... good luck.. actually maybe you can write something up about how to do cross world unit testing properly
[11:17:42 CDT(-0500)] <Bosmon> Well, I don't think it will require anything particularly special in this case
[11:17:52 CDT(-0500)] <Bosmon> Just merely having some test cases which actually test the functionality : P
[11:21:43 CDT(-0500)] <Bosmon> Justin_o: What do you think to moving the file "UIOptionsPreview.html" into the main component directory?
[11:22:01 CDT(-0500)] <Bosmon> I discovered that currently our test case for Full Preview doesn't actually succeed in loading any preview markup at all
[11:22:19 CDT(-0500)] <Bosmon> And I think it would be very bad to have one of our automated test cases depending on a file which is in "demo"
[11:24:08 CDT(-0500)] <Justin_o> Bosmon: hmm.. i think there is a test preview file in the tests directory
[11:25:06 CDT(-0500)] <Justin_o> if we were to put a preview in the compoent we'd probably have to get jameswy to design one that he thought was appropriate... i'm not sure how happy he is with the ones we have for our demos so far
[11:28:27 CDT(-0500)] <Bosmon> Well, the test preview file isn't really very useful
[11:28:36 CDT(-0500)] <Bosmon> Since it doesn't contain anything which TOC could bite onto
[11:28:53 CDT(-0500)] <Bosmon> And I think it would be much preferable if the component actually came "out of the box" working
[11:28:54 CDT(-0500)] <Bosmon> Rather than not
[11:30:24 CDT(-0500)] <Bosmon> I'll move the demo file into components as a placeholder at least - since it is better than anything else we have
[11:30:45 CDT(-0500)] <Bosmon> We can make it a blocker for the release if you want for jameswy to redesign it
[11:30:52 CDT(-0500)] <Justin_o> Bosmon: jameswy is off to a meeting right now with colin, but when he gets back i can talk to him about it
[12:54:06 CDT(-0500)] <Justin_o> jameswy: now that you're back.. Bosmon and I were talking about the preview for UIO not being part of the component
[12:54:14 CDT(-0500)] <Justin_o> currently it only exists in the demo
[12:54:18 CDT(-0500)] <Justin_o> the preview page that is
[12:56:28 CDT(-0500)] <jameswy> Justin_o, Bosmon: I agree that it makes sense to have a default demo integrated with the component.
[12:57:02 CDT(-0500)] <Bosmon> He Yields! He answers to our call! We do not ask for more!
[12:57:09 CDT(-0500)] <jameswy> (default preview, I mean)
[12:57:13 CDT(-0500)] <Bosmon> A sturdy fellow after all, this latest Ruddi-gore....
[12:57:15 CDT(-0500)] <jameswy> But it should also be the case that we encourage the implementer to provide a site-stylized preview
[12:57:22 CDT(-0500)] <Bosmon> A sturdy fellow after all, this latest Rudd-gore!
[12:57:27 CDT(-0500)] <jameswy> Rudd-gore...?
[12:58:22 CDT(-0500)] <jameswy> All perish in unheard woe, who dare our wills defy?
[12:58:28 CDT(-0500)] <Bosmon> yes, that!!
[12:58:39 CDT(-0500)] <Bosmon> We beg your pardon, ere we go....
[12:58:43 CDT(-0500)] <Bosmon> For having a-go-nised you so!
[12:59:00 CDT(-0500)] <Bosmon> Sorry, I am somewhat crazed after a full day of IE6 testing.....
[13:00:26 CDT(-0500)] <Justin_o> Bosmon: that's how we all get, but i usually i lose the ability to form any sort of coherent thought
[13:00:27 CDT(-0500)] <jameswy> It's okay. We all quote from Gilbert and Flint after IE6 testing.
[13:05:20 CDT(-0500)] <Bosmon> http://www.youtube.com/watch?v=IhBbBGbaVr8
[13:50:43 CDT(-0500)] <Bosmon> I'm starting to receive a lot of commit messages from the last 2 days....
[14:19:44 CDT(-0500)] <colinclark> Hey Justin_o, what's the state of FLUID-4444?
[14:28:29 CDT(-0500)] <Justin_o> colinclark: it's ready for code review.. i haven't done that yet.. i have tried it either.. the change seems straightforward enough
[15:37:38 CDT(-0500)] <Bosmon> So... I think I have finally written a test case. ...........
[15:37:56 CDT(-0500)] <Bosmon> That only took about 5 hours
[15:49:30 CDT(-0500)] <colinclark> Bosmon: A test case for what?
[15:51:19 CDT(-0500)] <Bosmon> For my IE6 fix
[15:52:49 CDT(-0500)] <Bosmon> It hardly seems worthwhile... even if UIO is "working" under IE6, on my machine it takes more than 1 minute to load up
[15:53:14 CDT(-0500)] <Bosmon> So the component is effectively worthless in that environment
[16:13:59 CDT(-0500)] <Bosmon> Hmm... actually it runs a lot faster on the build site