[16:12:38] RoanKattouw: you around? I'm noticing weirdness with IE returning values for text selections that are devoid of line breaks [16:14:22] Yeah I'm here [16:14:35] What weirdness exactly? Is your code up-to-date? [16:15:01] yeah it's up to date [16:15:29] hi guyrs [16:15:31] guys [16:16:47] i select text in the iframe, and getSelection seems to retrieve it just fine, except that there's no line breaks [16:18:20] like type a word in, hit return, type another word, select both words and hit the Bold button in the toolbar and you'll see two different things in firefox vs IE [16:18:29] Trying [16:18:46] firefox gives me: adam [16:18:46] miller [16:18:54] Yeah just saw it, interesting [16:18:57] ie gives me: adammiller [16:20:16] Morning nkomura [16:20:26] morning Roan [16:20:40] thank you for following up on the bug reports from calcey [16:20:55] how do IE7 & 8 look so far? [16:21:11] if you log the output of context.$iframe[0].contentWindow.document.selection.createRange().text, it's devoid of any \n or \r or
[16:22:16] Yeah [16:22:26] Check out createRange().htmlText [16:22:32] And you'll see whty [16:22:48] Firefox uses
s to separate lines in the iframe, but IE puts each line in a

[16:23:11] hahaha [16:23:15] awesome [16:23:52] i'm not seeing that though [16:24:00] i get
tags in IE8 [16:24:06] Interesting [16:24:09] Oh wait [16:24:21] Maybe it's only newly added text that's in

s [16:24:52] nkomura: They're progressing, almost stable. We got pretty much all of the big bugs [16:25:00] that's great [16:25:24] how about bugs filed for FF? [16:25:33] I'm not aware of any :) [16:25:47] RoanKattouw: right, new text is in

tags, preexisting lines are broken with
[16:25:54] Most bugs are either universal or browser-specific where browser is not Firefox :) [16:26:10] Yeah we explicitly insert those
s. [16:26:23] there are NTOC issues for FF [16:26:27] This is bad because it potentially breaks quite a few things [16:26:55] nkomura: Are there still? If you mean the old NTOC-scrolls-all-over-the-place issue, that should have been fixed by moving to the iframe [16:27:17] i will need to check the current status [16:27:34] For instance, my DOM traversal code counts
s as newlines (text with length 1) while getContents() doesn't handle them [16:27:37] Ooooo...IE7 actualy seems to work better than 8 with this business. createRange().text returns linebreaks in 7, but not 8 [16:27:43] WTF [16:27:46] but NTOC not detecting as soon as the header is inserted was a big one i thought [16:27:50] That's the most ridiculous thing ever [16:28:02] Yeah but that works in Firefox now, right? [16:28:19] Oh IE doesn't detect anything happening yet, that's a big bug [16:29:01] adam_miller: How does IE7 do linebreaks with
compared to

? Does it insert

s like IE8 does? Does it convert them to newlines correctly? [16:29:36] nkomura: So one big bug left on IE is that IE basically refuses to tell us when something's going on inside the iframe (text inserted, cursor moved, etc.) so the NTOC doesn't update ever [16:30:12] sounds like a major issue [16:30:19] Yes [16:30:22] do you guys have an idea how to go about it? [16:30:34] I'm waiting for Trevor, he's poked at event propagation issues in IE before [16:32:35] RoanKattouw: 7 seems to work the same way as 8 with the p tags for new line breaks, br for old [16:33:00] OK [16:33:14] Does it show the newlines in range.text for both BRs and Ps? [16:34:41] yeah, 7 handles returns seemingly identical values to firefox for range.text [16:35:00] err....remove 'handles' from that sentence [16:36:02] Cool [16:36:22] So we're in the crazy situation where 7 is actually working perfectly and 8 is broken [16:37:11] (Except that

insertion still breaks the assumption made elsewhere that lines are broken with
; this'll actually work in 8 and break in 7 cause in 8 that assumption will agree with what getSelection says) [16:39:32] yeah. should I commit my changes to jquery.wikiEditor.js for the multiline indent handling? [16:39:45] seems like it will be working in IE8 when we get this issue resolved [16:45:49] RoanKattouw: have you started tagging the revisions for babaco enhancement release? [16:46:10] Not yet, will do [16:46:37] My plan for the UsabilityInitative extension was to just release trunk or a certain snapshot of trunk, so no tagging needed there [16:46:51] There are a few revs in MediaWiki core that need to be deployed alongside, I made a list of those [16:47:09] ("certain snapshot" means the state of the code as of a certain revision) [16:47:17] what are you referring to by a certain snapshot of trunk? [16:50:58] So basically I'll be saying "let's deploy r12345 of UsabilityInitiative" meaning all changes before and including r12345 are in and everything after r12345 is out [16:51:38] As opposed to picking revisions here and there, leaving out revisions in between, deploying different revisions of different files, etc. [16:52:51] how do you plan to go about excluding citron features which are not part of babaco enhancement release? [16:53:27] By deploying the code but disabling it [16:54:32] That's been done before; the upload-by-URL code for instance is deployed on the live site but disabled. I enabled it once and Tim yelled at me because it broke the search function (!) [16:55:07] Grabbing dinner now; when I'm back and Trevor's in we'll come up with a game plan I guess [16:55:08] so the current release candidate reflect they way you plan to disable it [16:55:12] ? [16:55:13] Yes [16:55:27] The deployment wiki has all of the template editor code and what not, but it's all disabled [16:55:30] k, let's chat some more later [16:55:43] enjoy your dinner :-) [18:33:39] so, the

vs
comes down to IE's use of enter ==

[cursor]

, shift+enter ==
[cursor] [18:33:50] other browsers use
all the time [18:34:02] this is annoying because you can't get in and make enter act like shift+enter [18:34:03] has to do with IE sucking, google IE sucks for more details :) he he he [18:34:03] so anyways [18:34:21] what I did was make paragraphs have no margins or padding, so they look just like normal new lines [18:34:27] Yeah [18:34:41] So either way it looks like we're stuck with having an inconsistent DOM with a mixture of
and

[18:34:43] then I added special filters on getContents that convert the paragraphs to plain old
tags [18:34:51] We can't normalize either way because of Enter and Shift+Enter [18:34:57] yes [18:35:13] OK so the DOM traversal needs to be fixed up to recognize

tags [18:35:23] so the best we can do is to improve our traversal stuff to support the

style of things [18:35:30] this would have come up eventually anyways [18:35:42] because you can paste into firefox and end up with

tags [18:35:51] Ah yes [18:36:03] Right so the traverser needs to recognize

tags, I'll poke at that [18:36:33] ok [18:36:40] the event thing, I can try and work on again [18:36:41] I'm not sure what I'm gonna do about IE 8's selection handling yet [18:37:00] Did it work once? [18:37:23] I thought so [18:37:33] but perhaps I'm mistaken [18:37:40] I will investigate and get something solved [18:37:49] *it solved [18:38:12] I think it's got to do with IE's strange way of handling inter-frame permissions [18:39:02] it could also have to do with the fact that IE dumps the body when we turn designMode on [18:39:20] which breaks all the context that jquery might have put in pace with the DOM [18:39:25] I'm not really sure [18:39:31] looking in a sec [18:39:32] Hmm [18:39:45] Theoretically the latter could be checked with a breakpoint in the iframe