Jump to content

IRC office hours/Office hours 2013-11-03

From Meta, a Wikimedia project coordination wiki

Chat on VisualEditor
Sunday 3 November 2013
00:00 - 01:00 UTC


[00:00:03] <Sky_Harbor> i was supposed to be in the earlier office hour, but i fell asleep, so i will sit in for the first part of this one
[00:00:08] <Sky_Harbor> leaving for the cemetery in a bit
[00:00:34] <Jamesofur> welcome everyone, we'll get started in a sec, just as a heads up to everyone (especially if you didn't read the topic) we are logging and the log starting now will get posted on meta
[00:01:43] <Jamesofur> We've had enough of these we might as well jump in if you have questions I'll try to stack them up. Risker: I know you dropped off before finishing your last question earlier, do you want to start with that?
[00:01:58] <Risker> thanks, Jamesofur.
[00:02:14] <Sky_Harbor> noted, James
[00:02:40] <Risker> My question is about the planned addition of an edit-tool to VE. I'm wondering what it will include, and how closely it will parallel the edit tool in use on Monobook wikitext editing.
[00:03:16] <James_F> Risker: By "an edit-tool", do you mean a character inserter or something else?
[00:03:46] <James_F> Risker: Also, "on Monobook"? I don't think there's anything skin-specific about either the wikitext editor or VisualEditor.
[00:03:58] <Risker> character inserter, I believe is what it is called on the bugzilla
[00:04:13] <Risker> https://bugzilla.wikimedia.org/show_bug.cgi?id=50296
[00:04:23] <James_F> Right.
[00:04:43] <James_F> So, I touched on this this morning, but you didn't get to see that yet, sorry.
[00:05:06] <Risker> James_F, I'm not sure whether or not there is a difference between skins for the character inserters.
[00:05:11] <James_F> The character inserter will not be exactly the same as in the wikitext editor.
[00:05:26] <James_F> Firstly, because there are three (!) wikitext editor character inserters.
[00:05:31] <rastus_vernon> Aren't there at least two of them?
[00:05:32] <Risker> ouch!
[00:05:46] <rastus_vernon> I know there's one built into the editing bar, and another that appears under the edit box.
[00:05:53] <James_F> There's only meant to be one, but, well, a community member was angry about the idea of getting rid of the excess ones.
[00:06:16] <James_F> Yeah, there's one in the toolbar in WikiEditor (the "new" toolbar provided as part of the Usability Initiative in 2007).
[00:06:36] <James_F> There's the old toolbar, MediaWiki:Edittools, that's often extended by local wikis, often to quite ludicrous extents.
[00:07:03] <Sky_Harbor> three?
[00:07:05] <James_F> And then there's the old "new" wikitext editor toolbar character inserter, which I imagine most of you have never seen, but is there for users who've disabled the "new" wikitext editor.
[00:07:10] <James_F> Yeah. Useful, huh? :-)
[00:07:11] <Sky_Harbor> wait...what are the differences between the three of them?
[00:07:12] <James_F> Anyway.
[00:08:07] <James_F> Some of the "characters" that these inserters provide are actually just chunks of wikitext.
[00:08:13] <Jamesofur> Sky_Harbor: I believe mostly that there are different places to edit them and so they can end up less then equal
[00:08:24] <rastus_vernon> I suppose there's no work right now being done on or no plan on work being done on creating a visual editor for LaTeX formulas?
[00:08:29] <Sky_Harbor> ahh, i see
[00:08:39] <rastus_vernon> I'm aware of the VisualEditor Formulae editor that will be part of Beta Features.
[00:08:50] <rastus_vernon> I tried it, but it only allows directly inputting LaTeX code.
[00:08:57] <James_F> E.g. "<ref></ref>" or "#REDIRECT[[]]".
[00:09:31] <James_F> Those don't make sense for VisualEditor, obviously, so we have to do a character inserter for language characters rather than syntax mumbo-jumbo.
[00:09:38] <Sky_Harbor> since I'm leaving soon to go to the cemetery, I'll throw in a question: has the schedule for the deployment of VE for the other Wikimedia projects (including tl.wiki, which is why I'm here) been moved?
[00:09:57] <Sky_Harbor> I've noticed that there have been no updates on that front
[00:10:00] <Jamesofur> Thanks Sky_Harbor and rastus_vernon, I'll make sure yours get answered after risker
[00:10:11] <James_F> rastus_vernon: There was some early work done by our GSoC student as part of that work, but no, there's no visual equation editor for VE yet, sadly.
[00:10:12] <Sky_Harbor> I'll be sure to read the log afterward :)
[00:10:36] <James_F> rastus_vernon: If you or someone you know would like to make a start, we'd love to support you. :-)
[00:10:43] <Sky_Harbor> I'm also asking these questions as I still have a pending bugzilla request for a new interface language for tl.wiki which James_F responded to
[00:10:48] <Sky_Harbor> https://bugzilla.wikimedia.org/show_bug.cgi?id=52957
[00:10:56] <Jamesofur> Risker: did that answer your question?
[00:11:49] <Risker> I think so, Jamesofur. Sky_Harbor's kind of in a time-sensitive situation, so perhaps his question can be answered next?
[00:11:50] <James_F> Sky_Harbor: So, yes, tlwiki is an issue.
[00:11:52] <jayvdb> math editor enhancement request: https://bugzilla.wikimedia.org/show_bug.cgi?id=43058
[00:12:02] <James_F> Sky_Harbor: VisualEditor doesn't support language variants yet.
[00:12:31] <James_F> Sky_Harbor: That means that, if we were to switch language variants on for tlwiki, we'd not be able to deploy VE until that was done.
[00:12:32] <rastus_vernon> James_F: Is that why the ULS input methods don't seem to do anything in VisualEditor?
[00:12:49] <Sky_Harbor> is that for content or for the interface?
[00:12:56] <James_F> Sky_Harbor: Which is the quandary. :-)
[00:13:08] <Sky_Harbor> I mentioned the interface in the bugzilla
[00:13:18] <James_F> Sky_Harbor: Content.
[00:13:40] <Sky_Harbor> the issue of content variants is highly contentious and we don't expect to push through with it, which is why we're only agreeing to have a new informal Tagalog interface language
[00:13:52] <Sky_Harbor> if it's just for the interface, it shouldn't be a problem, right?
[00:13:53] <James_F> Sky_Harbor: Contentious within the community, or more generally?
[00:13:55] <Jamesofur> similar to de-formal/de I assume?
[00:14:03] <Sky_Harbor> contentious with the community
[00:14:06] <Sky_Harbor> and yes, similar to de/de-formal
[00:14:09] <James_F> Yes, VE supports whatever interface languages you can get translatewiki.net to support.
[00:14:11] <jayvdb> They only want an interface translation
[00:14:12] * James_F nods.�
[00:14:28] <James_F> OK, in that case tlwiki can probably go ahead relatively quickly.
[00:14:33] <Sky_Harbor> OT: I really need to learn how to use IRC :P
[00:14:38] <Sky_Harbor> oh, wonderful :)
[00:14:42] <Jamesofur> don't we all
[00:15:13] <James_F> Sky_Harbor: I think you'd want to wait for the character inserter to be ready to insert nÍ g though?
[00:16:02] <Sky_Harbor> well, we could, but ñ is not used that often in Tagalog, so it isn't really necessary to wait for it, but if you think it's more practical then I don't see what's wrong
[00:16:19] <James_F> Sky_Harbor: Happy to throw the switch tonight if the community wants it. :-)
[00:16:25] <Sky_Harbor> ñ is mostly used in surnames :)
[00:16:29] <James_F> Right.
[00:16:38] <Jamesofur> [note for transparency: James and I are in the same room so some of the question ordering etc going on verbally]
[00:16:53] <Sky_Harbor> well, I'll have to ask, so I think I can get back to it
[00:16:54] <James_F> OK, well, we're about to start to list the next batch of wikis we think are ready to get VisualEditor.
[00:16:57] * James_F nods.�
[00:17:05] <James_F> It won't be a surprise, don't worry. :-)
[00:17:30] <James_F> Sky_Harbor: Anything more before you have to run?
[00:17:40] <Sky_Harbor> well, before I start preparing to go, how about our bugzilla for the new interface language? :P
[00:17:48] <Sky_Harbor> so far though, I think my issues with VE are fine
[00:17:55] <Sky_Harbor> I've been testing it in tl.wiki and it seems to be running well
[00:18:01] <James_F> Sky_Harbor: Not my department – will ping Siebrand.
[00:18:05] * James_F nods.�
[00:18:11] <Sky_Harbor> all right
[00:18:14] <James_F> Sky_Harbor: If you have any issues, just shout. :-)
[00:18:16] <James_F> OK.
[00:18:19] <Sky_Harbor> thanks, James_F
[00:18:21] <Sky_Harbor> will do ;)
[00:18:28] <James_F> rastus_vernon: You asked about ULS input methods?
[00:18:32] <rastus_vernon> James_F: Yes.
[00:18:37] <James_F> rastus_vernon: ULS input methods are… a challenge.
[00:18:47] <rastus_vernon> You mean integrating them in VE?
[00:19:11] <James_F> rastus_vernon: VisualEditor expects (and works very well with) some forms of IMEs, and is completely broken with others.
[00:19:26] <rastus_vernon> I can understand why that's a challenge, considering it means that it requires controlling keyboard input in contenteditable blocks.
[00:19:30] <James_F> We're working on un-breaking the other IMEs, the most obvious of which is ULS.
[00:19:57] <James_F> Exactly. Some IMEs insert and then replace characters as you type, others insert whole words, more augment, etc.
[00:20:04] <James_F> It's a nightmare. Yay for no standards. :-(
[00:20:24] <James_F> We've been working with the Language Engineering team, and obviously ULS's JS software IME is a priority.
[00:20:27] <rastus_vernon> I wasn't actually aware that the DOM even made it possible to play with keyboard input.
[00:20:34] <James_F> But for now it's quite badly broken.
[00:20:38] <rastus_vernon> (without using keyboard events in all sort of magical ways)
[00:21:18] <James_F> rastus_vernon: We've actually managed to build a test framework that lets us simulate IMEs by triggering events (compositionStart/End/keyDown/etc.) through the browser.
[00:21:42] <rastus_vernon> Oh.
[00:21:52] <James_F> rastus_vernon: So pretty soon we're expecting to have the ability to build a unit test for each IME, proving that (a) it works now, and (b) it's still working in the future as we make changes.
[00:21:57] <rastus_vernon> Do you have a framework for specifying IMEs, or are they all hand-coded?
[00:21:58] <James_F> rastus_vernon: But it's slow work, sadly. :-(
[00:22:23] <James_F> rastus_vernon: We've actually built a tool that captures IME events as a JSON file, so we can "play back" an IME's input and test against it.
[00:22:37] <James_F> rastus_vernon: So… manually-curated but not "hand-coded" per se.
[00:22:54] <rastus_vernon> So most of your work so far with input methods is with testing? ;)
[00:23:19] <James_F> rastus_vernon: Well, not just testing, but it's been a large chunk of the work.
[00:23:21] <Sky_Harbor> okay, will be going now
[00:23:30] <Sky_Harbor> thanks again to the two Jameses, and I will be looking at the log :)
[00:23:32] <Jamesofur> jayvdb: did you have a question/want an update on the math request or was that more an FYI?
[00:23:52] <Jamesofur> Thanks for coming Josh
[00:23:57] <rastus_vernon> That makes sense, considering testing is the only way to keep track of whether everything works as expected with input methods.
[00:24:04] <rastus_vernon> Since many of them are for wikis which aren't edited by a lot of people.
[00:24:10] <James_F> rastus_vernon: Whenever you get a pawn or a snowman character, that's a sign that the input code has failed. We're aiming to work out at least the /causes/, if not the /fixes/, as quickly as possible.
[00:24:35] <rastus_vernon> Oh. Is that related to the problems some people had about pawns getting inserted in articles for some reason?
[00:24:41] <James_F> Indeed. Also, we're talking about ~ 500 IMEs or equivalents used on our wikis right now. It's not sensible to try to test that manually, even if we had unlimited funds.
[00:24:48] <James_F> Exactly the reason, sadly. :-(
[00:24:51] <jayvdb> Jamesofur: it was a ftr ;-)
[00:24:58] <Jamesofur> cool :)
[00:25:05] <rastus_vernon> Well, 500 input methods is amazing.
[00:25:16] <James_F> I can talk in some (more) length about it, but I don't want to crowd out other questions.
[00:25:19] <rastus_vernon> Maybe we need to start a new Wikimedia project for collaboratively editing input method definitions!
[00:25:33] <James_F> rastus_vernon: Just iBus 1.8.3 vs. 1.8.4 has differences in our experience, for example.
[00:25:52] <James_F> rastus_vernon: That's keeping everything else (OS/browser/site/language/input stream) exactly the same.
[00:26:03] <jayvdb> Has the list of 500 IMEs been pasted to meta ? if not, could it be, please?
[00:26:05] <Jamesofur> while James and rastus continue, is there anyone else who has a question or clarification they want?
[00:26:41] <James_F> rastus_vernon: Let alone all the random IMEs that "just work" in wikitext editor (because it's a plain text box and so the OS does all the work).
[00:27:07] <rastus_vernon> You mean the input methods for text areas are handled by the OS?
[00:27:11] <rastus_vernon> I thought they had to use keyboard events too.
[00:27:11] <James_F> jayvdb: The "500" is a rough figure based on the complexities we've found; there's not an exact list of what we're supporting.
[00:27:24] <VojtechDostal> Hi! Last time I asked about the Reference Toolbar implementation. Did you have some progress in it?
[00:27:32] <rastus_vernon> I'm not really aware of all the DOM's API for input in text areas and text inputs.
[00:27:55] <James_F> rastus_vernon: Browser/IME/OS integration is pretty stable for text areas - it Just Worksâ„¢, and if it doesn't, people blame the IME.
[00:28:00] <jayvdb> Has there been any analytics on mobile devices using VE ?
[00:28:22] <James_F> rastus_vernon: Whereas for contentEditable we have to work around every issue (or actively decide not to let that IME work in VE).
[00:28:50] <James_F> VojtechDostal: Hey. Yes.
[00:29:10] <James_F> VojtechDostal: Since last month, we've made some more progress on designs for the reference inserter, which you can see at https://www.mediawiki.org/wiki/VisualEditor/Design/Reference_Dialog
[00:29:37] <James_F> VojtechDostal: The discussion continues, though we feel we're relatively comfortable with iterating in code now with the designs as shown there, based on feedback from the community.
[00:29:52] <James_F> VojtechDostal: But there's no actual code I can point at yet, sorry.
[00:30:19] <jayvdb> James_F: do you have a partial list of IME that Are Working (TM) and ones you are aware of but are not yet tested or verified to be working
[00:30:43] <VojtechDostal> James_F: yes the design looks good. I am only looking forward to the reference toolbar. Without it, I can hardly teach the VisualEditor to my students...
[00:30:46] <Jamesofur> jayvdb: for analytics on mobile you mean on if they're using VE?
[00:30:50] <James_F> jayvdb: When you say "analytics on mobile devices using VE", what do you mean? Counts of how many users/uniques/impressions are currently using mobile? Currently using VE? Who we…
[00:30:56] <James_F> Yeah, what Jamesofur said.
[00:31:24] <James_F> VojtechDostal: Understood. We hope it works for you, and if you have any issues, please do shout. :-)
[00:31:41] <Ironholds> data on mobile + VE sounds fairly simple; they're both just tags
[00:32:48] <James_F> jayvdb: Currently we don't have enough testing to declare than any IME works in all cases, so it's the null list, sadly. This is why we've built the automated testing framework for this.
[00:36:35] <Jamesofur> James_F: I know you're working on the list of wikis for the next deploy, do you know when that might be or is it a bit down the road?
[00:38:25] <James_F> Jamesofur: I don't know. The list is probably mostly ready, but we want to sanity-check it first before just posting it as "surprise!". :-) It will be a request for input, with at least a couple of weeks' notice before we change anything.
[00:38:45] <James_F> jayvdb: Ping again about your question about mobile analytics - not sure what you wanted to know?
[00:40:12] <jayvdb> e.g. how many mobile edits are done with VE?
[00:40:26] <James_F> jayvdb: Right now? Almost none.
[00:40:32] <Jamesofur> and another ping for anyone around who may have other questions :) Obviously if you don't have any don't feel bad, they ended a bit early this morning as well.
[00:41:03] <Keegan> Jamesofur: By a bit early you mean about ten minutes ;)
[00:41:22] <Jamesofur> yes, I heard I missed your dramatic reading
[00:41:33] <jayvdb> James_F: do you have any lists of IMEs ? e.g. a list of ones you are testing?
[00:42:33] <James_F> jayvdb: VE isn't integrated with the mobile interface, so you have to use the desktop side. We're working with the Mobile team on integrating VE with Mobile and focussing on what blockers there are to making it available there.
[00:42:56] <James_F> That's initially for tablets, but later we'll be working on getting it available for 'phones too.
[00:43:01] <rastus_vernon> If there are around 500 input methods, that means there are more than 2 input methods in average per language, right?
[00:43:13] <Keegan> Jamesofur: You missed nothing. pb wouldn't let me do King Solomon's Mines
[00:43:41] <James_F> jayvdb: The Language Team has a list of ~ 160 IMEs (just the IME, not counting the OS/browser/version combos that explode the number forther), which we're using as our starting point. Not sure where that is though, sorry.
[00:44:04] <James_F> jayvdb: Is there a particular IME or set of IMEs on which you think we should particularly focus?
[00:44:39] <rastus_vernon> I'm mostly active on Wikimedia projects, but I also contribute on one Wikia project, and because of this I heard they're integrating VisualEditor.
[00:45:04] <rastus_vernon> I'm curious on how exactly the two teams collaborate.
[00:45:06] <jayvdb> James_F: is there any analytics on device/OS/browser combos using VE ?
[00:45:10] <rastus_vernon> (since they have their own feedback page)
[00:45:36] <James_F> rastus_vernon: Yes, we've been working with Wikia for a couple of years on the VisualEditor.
[00:45:49] <Jamesofur> jayvdb: do you mean in a general sense including desktop etc?
[00:45:57] <jayvdb> James_F: re IME, if you want users to test for you, I think we need a wiki page listing the IMEs and a status for each
[00:46:07] <jayvdb> Jamesofur: yes
[00:46:33] <James_F> rastus_vernon: They've been a core part of the team, though recently they've focussed on the Wikia-specific work.
[00:46:47] <James_F> rastus_vernon: Not sure if that answers your question?
[00:47:21] <James_F> jayvdb: In terms of browser-/OS-level break down of VisualEditor usage, sadly no, I do not have that data.
[00:48:07] <James_F> jayvdb: We don't particularly want users to be trying to test VisualEditor's support for IMEs themselves, as it's a huge amount of work and would be a huge burden on volunteers' time. Hence why we built the automated testing rig.
[00:48:18] <jayvdb> there is about 60 pages re IME at https://en.wikipedia.org/wiki/Category:Input_methods
[00:49:20] <jayvdb> An automated testing rig is going to need a config entry for each IME to be simulated/tested
[00:49:42] <James_F> jayvdb: Yes. As I said above. :-)
[00:51:21] <Jamesofur> about a 10 minute warning for people until the end of the session
[00:52:17] <James_F> … though we seem to have run out of questions now.
[00:52:25] <James_F> Keegan: You ready with the dramatic reading? :-)
[00:52:37] <Keegan> Absolutely. Let me get my fedora
[00:52:38] * Jamesofur starts dancing the hula in the background as a screen saver�
[00:53:10] <Jamesofur> Risker: was there anything further from your question earlier you wanted?
[00:53:19] <Keegan> Thanks to the James (take your pick), this was quite informative in a different manner from earlier
[00:53:22] <Risker> I think you'
[00:53:42] <Risker> *you answered as well as possible. Is there somewhere we can go to see what characters are going to be available?
[00:54:26] <James_F> Risker: No, because we haven't decided yet. We're not going to "fire and forget" - it will be open to change as needed.
[00:54:56] <James_F> Risker: But at first probably something like é è ï – — for those as can't type them.
[00:55:30] <Jamesofur> would this be somewhere specific (wikitech/mediawiki/area of gerrit etc) ?
[00:55:34] <Risker> good to hear it James_F. Will it also include the nonstandard punctuation like the emdashes and the "usual" symbols like arrows?
[00:55:37] <Jamesofur> when it's ready
[00:56:05] <James_F> Risker: Yup.
[00:56:35] <rastus_vernon> Will it include a mechanism to insert any Unicode character by selecting the table and clicking on the appropriate character?
[00:56:44] <Risker> great, James_F. One more step in making it easy
[00:56:45] <James_F> Jamesofur: Don't know for sure.
[00:56:45] <rastus_vernon> (some systems don't have something built-in for this)
[00:56:59] <rastus_vernon> (and on those systems, it can be awfully hard to insert some particular characters)
[00:57:02] <James_F> rastus_vernon: Possibly; I don't want to make it too complicated, though.
[00:57:06] <jayvdb> The list of ~150 IMEs appears to be here: https://github.com/wikimedia/jquery.ime/blob/master/src/jquery.ime.inputmethods.js
[00:57:57] <James_F> jayvdb: I think that's the list of input methods that ULS itself supports, not the non-ULS IMEs Language Engineering test against.
[00:58:21] <jayvdb> What are the plans for Internet Explorer support?
[00:58:26] <rastus_vernon> Will all the input methods be merged into ULS eventually?
[00:59:13] <James_F> jayvdb: We're looking at it right now. Long-term we plan to support IE 9+.
[00:59:21] <James_F> rastus_vernon: No?
[00:59:27] <jayvdb> how long is long-term ?
[00:59:42] <Jamesofur> rastus_vernon: I'm not sure I understand exactly what you mean by that
[00:59:55] <James_F> jayvdb: I don't know. It will take as long as it takes. Anyone who pretends to tell you otherwise for a research project is just being dishonest. :-(
[01:00:06] <rastus_vernon> Jamesofur: Well, there are non-ULS input methods. Those can be selected in various ways, if I'm not wrong.
[01:00:27] <rastus_vernon> Some of those are perhaps pre-ULS, and unifying them into the ULS could be a good idea in the very very very very long term.
[01:00:30] <jayvdb> James_F: can we have 'beta' in the UI renamed to 'research project'? ;-)
[01:00:53] <James_F> jayvdb: IE is vastly different in some crucial ways, hence why we need a research effort to work out how we can get it to work.
[01:01:02] <rastus_vernon> jayvdb: 'experimental development research project' would be better.
[01:01:37] <James_F> rastus_vernon: VE is going to support the ULS-provided IMEs early on, but I don't think it's reasonable to require users to use ULS's IMEs rather than the ones they use elsewhere, just because VE works with it.
[01:02:34] <rastus_vernon> Oh, I thought those IMEs worked only on MediaWiki.
[01:02:46] <jayvdb> fwiw, the roadmap for IME needs to be filled in https://www.mediawiki.org/wiki/Roadmap#Visual_Editor_-_IME_integration
[01:03:08] <Jamesofur> I think we're going to wind down now given that the designated time has come. If James_F has any finishing to do I'll keep logging that but other then that will close down.
[01:03:31] <Risker> thanks folks
[01:03:48] <James_F> jayvdb: I don't think [[mw:Roadmap]] is used any more.
[01:03:53] <jayvdb> thanks James&James
[01:03:54] <Jamesofur> Thank you, everyone, for coming
[01:03:54] <James_F> jayvdb: It's certainly not used by me.
[01:03:58] <James_F> Anyway.
[01:04:05] <James_F> Next time, or on-wiki.
[01:04:08] <James_F> See you all!
[01:04:54] <Jamesofur> I will get this log up by later tonight and give Pb a big poke
[01:05:12] <rastus_vernon> Do the four extra minutes compensate for the 10 minutes loss in the previous office hour?
[01:05:20] <Jamesofur> lol
[01:05:41] <Jamesofur> I guess we can look at it that way but I was more just thinking of it as compensating for questions still on going :)
[01:06:12] <rastus_vernon> Well, I missed the previous office hour even though my computer was on.
[01:06:22] <rastus_vernon> I somehow had forgotten about it.
[01:06:55] * Jamesofur refused to get up �
[01:09:01] <Jamesofur> ok, I think I'm going to get off soon and will prep the log and go searching for the chef hiding in the kitchen.