[20:21:25] <Denny_WMDE> so why not here? [20:21:42] <Denny_WMDE> (then again, you could just add a (en) to it) [20:22:29] <JeroenDeDauw> Denny_WMDE: yeah, that is what Krinkle changed [20:22:38] <liangent> Denny_WMDE: afaik entities fetched from searchentities behave the same as wbUsedEntities [20:22:41] <liangent> is this correct? [20:23:24] <Denny_WMDE> JeroenDeDauw: he add a typo to the filename? (confused) [20:23:50] <JeroenDeDauw> "(22:14:51) Denny_WMDE: huh? i cannot merge? JeroenDeDauw ?" [20:23:53] <Denny_WMDE> liangent: i do not know [20:24:01] <Denny_WMDE> JeroenDeDauw: ah, that one [20:24:14] <Denny_WMDE> JeroenDeDauw: ok,, but it seems merged now [20:25:36] <Krinkle> Denny_WMDE: https://gerrit.wikimedia.org/r/#/c/74001 now shows that you merged it manually. You are familiar with gatte-and-submit? [20:25:53] <gerrit-wm> New patchset: Jeroen De Dauw; "Serialze comparators as strings" [mediawiki/extensions/Ask] (master) - https://gerrit.wikimedia.org/r/74046 [20:25:53] <gerrit-wm> New patchset: Jeroen De Dauw; "Fix typo in file name and updated @covers tags" [mediawiki/extensions/Ask] (master) - https://gerrit.wikimedia.org/r/74047 [20:26:11] <liangent> weird I can't really find the place where wbsearchentities is used, in javascript [20:26:21] <Denny_WMDE> Krinkle: no, sorry, what is gatte-and-submit ? [20:26:43] <aude> krinkle gets alerted with alarm bells when someones merges [20:26:46] <Denny_WMDE> Krinkle: also, I am confused. I only CR+2'ed it - why does it show I merged it manually? [20:26:59] <Krinkle> Denny_WMDE: It means the repository has a Jenkins pipeline configured that runs unit tests, and you shouldn't merge manually. [20:27:03] <aude> alarms and sirens [20:27:06] <Krinkle> Instead you just do "Code-Review: +1" [20:27:09] <Krinkle> Instead you just do "Code-Review: +2" [20:27:22] <Krinkle> and then Jenkins will do Verified: +2 and merge it [20:27:36] <Krinkle> Denny_WMDE: It shows you merged it because you clicked "Publish and Submit" [20:27:38] <gerrit-wm> New patchset: Jeroen De Dauw; "Remove obsolete TODOs" [mediawiki/extensions/Ask] (master) - https://gerrit.wikimedia.org/r/74048 [20:27:48] <Denny_WMDE> i never verified it [20:27:51] <Denny_WMDE> ah, that's the culprit [20:27:52] <Krinkle> Denny_WMDE: Jenkins already verified it [20:27:57] <Krinkle> Submit is Gerrit language for Merge [20:28:01] <Krinkle> you should press "Publish comments" [20:28:01] <Denny_WMDE> so i should be using only "publish" ? [20:28:07] <Krinkle> instead of "Publish and Submit" [20:28:07] <Denny_WMDE> ah ok, sorry [20:28:08] <Krinkle> yeah [20:28:21] <Denny_WMDE> didn't know. I have been told to use Publish and Submit [20:28:27] <Denny_WMDE> sorry, my fault, won't do it again [20:28:35] <aude> Denny_WMDE: it changed at some point [20:28:37] <Denny_WMDE> I guess I popped up on your alarm wall a few times [20:28:58] <Krinkle> This is exactly why I remove the Submit button from most repositories because people (accidentally or not) misuse it, and there's no justifiable case for using it (unless it is an utmost emergency that needs merging regardless of failing tests or Jenkins being down - at which point you can contact operations to override it - not me) [20:29:17] <Krinkle> Denny_WMDE: Yep, I have an alarm on "V: 2" and non-jenkins merges. [20:29:25] <JeroenDeDauw> Finals of the "N.O." competition: Denny vs Jenkins, tonight at 8 [20:29:27] <Denny_WMDE> nah, it's fine. I know now. Won't do it anymore :) thanks for letting me know Krinkle [20:29:33] <Krinkle> yw [20:29:50] <Denny_WMDE> JeroenDeDauw: :D [20:29:55] <Krinkle> Denny_WMDE: If you want I can easily 'upgrade' the settings of this repository to match others (e.g. don't show Verified and Submit) [20:30:04] <aude> Denny_WMDE: which by the way, we need to setup jenkins jobs for the deployment branch(es) and figure that out [20:30:26] <aude> they often will fail because they are testing older branch of wikibase together with master data values, data model, etc [20:30:30] <JeroenDeDauw> Denny_WMDE: looks like Ask is ready for initial release after https://gerrit.wikimedia.org/r/#/q/status:open+project:mediawiki/extensions/Ask,n,z [20:30:30] <Denny_WMDE> liangent: try the entity selector widget [20:30:54] <Denny_WMDE> Krinkle: no, not without asking the team first. This is not a decision I can make alone [20:31:21] <Krinkle> Sure, I'm just letting you know it is an option. We can't fix the confusing labels in Gerrit but we are able to hide them on request. [20:31:38] <Krinkle> (well, I suppose we could fix the labels if we really tried) [20:31:42] <aude> Krinkle: Denny_WMDE yes, the inability to merge would've been a probelm for our backports [20:31:58] <Krinkle> note this only applies to master [20:32:03] <aude> Krinkle: ok [20:32:09] <Krinkle> (the settings change) [20:32:20] * aude would've had to cherry pick two patches into the branch for the tests to pass [20:32:35] <Denny_WMDE> i can put it on the biweekly to discuss [20:32:42] <Denny_WMDE> thanks for the offer, krinkle [20:33:19] <Denny_WMDE> JeroenDeDauw: what is COMP_MEQ as a comparator? I understand GEQ, but not MEQ? [20:35:22] <gerrit-wm> New patchset: Jeroen De Dauw; "Update version, will be 1.0, not 0.1" [mediawiki/extensions/Ask] (master) - https://gerrit.wikimedia.org/r/74051 [20:35:22] <gerrit-wm> New patchset: Jeroen De Dauw; "Put release notes in README file" [mediawiki/extensions/Ask] (master) - https://gerrit.wikimedia.org/r/74052 [20:35:48] <JeroenDeDauw> Denny_WMDE: LEQ = less [20:35:55] <JeroenDeDauw> Denny_WMDE: having less and greater is odd [20:35:57] <JeroenDeDauw> less and more [20:36:06] <JeroenDeDauw> And more is nicer then grtr [20:36:08] <Denny_WMDE> hmm [20:36:21] <Denny_WMDE> actually that makes sense [20:36:38] <JeroenDeDauw> No! I'll have to revert in that case [20:36:48] <Denny_WMDE> ? [20:37:17] <JeroenDeDauw> Denny_WMDE: I'm just kidding, no worries [20:37:37] <Denny_WMDE> i am just used to LEQ and GEQ . if you google for it, you'll see they usually go together [20:37:38] <JeroenDeDauw> Denny_WMDE: if you could do a quick review of whole of Ask soonish, that'd be nice [20:38:00] <JeroenDeDauw> Denny_WMDE: I can change it back if you think taht is better [20:38:23] <Denny_WMDE> i think it is. same thing with the HTML entities gt lt [20:38:28] <Denny_WMDE> or in latex leq and geq [20:38:31] <Denny_WMDE> etc. [20:38:38] <Denny_WMDE> just less surprising to users [20:38:51] <JeroenDeDauw> Denny_WMDE: the only thing hindering release now is that one of Asks dependencies has not seen a single release yet (DVs), so specifying requirements is rather impossible now [20:38:51] <Denny_WMDE> shall i merge and then you change you amend and then i merge? [20:39:06] <Denny_WMDE> yah, true [20:39:36] <JeroenDeDauw> Denny_WMDE: just merge all of it, I'll make a commit on top of the chain changing this back - that is quickest [20:39:49] <Denny_WMDE> ok [20:40:47] <gerrit-wm> New review: Denny Vrandecic; "(1 comment)" [mediawiki/extensions/Ask] (master) C: 2; - https://gerrit.wikimedia.org/r/74046 [20:41:45] <liangent> Denny_WMDE: I realize extending wbsearchentities doesn't resolve everything [20:42:00] <gerrit-wm> New patchset: Jeroen De Dauw; "Use "geq" rather then "meq" for "greater then or equal" comparator" [mediawiki/extensions/Ask] (master) - https://gerrit.wikimedia.org/r/74055 [20:42:06] <Denny_WMDE> liangent: hardly anything does :) [20:42:53] <gerrit-wm> Change merged: jenkins-bot; [mediawiki/extensions/Ask] (master) - https://gerrit.wikimedia.org/r/74046 [20:43:01] <gerrit-wm> Change merged: jenkins-bot; [mediawiki/extensions/Ask] (master) - https://gerrit.wikimedia.org/r/74047 [20:43:15] <gerrit-wm> Change merged: jenkins-bot; [mediawiki/extensions/Ask] (master) - https://gerrit.wikimedia.org/r/74048 [20:43:52] <liangent> for example: my language is de and a fallback chain de|en; an entity has a de and an en label; I'm inputting an en search term and got that entity; so wbsearchentities is returning data in en (for what got matched) [20:44:23] <JeroenDeDauw> Danwe_WMDE: http://programmers.stackexchange.com/questions/204837/component-design-getting-cohesion-right [20:44:23] <liangent> however the system should show mw de label after saving [20:44:23] <Denny_WMDE> why update the since from 0.1 to 1.0 ? that sounds confusing. it has been there in 0.1…? [20:44:25] <Denny_WMDE> because 0.1 was no release? [20:44:32] <Denny_WMDE> JeroenDeDauw: ^^ [20:44:33] <liangent> that means an extra wbgetentities call is needed [20:44:51] <JeroenDeDauw> Denny_WMDE: There was never a 0.1 release yes [20:45:28] <Denny_WMDE> liangent: wait, why do you do saving? wbsearchentities is not about saving..? [20:45:54] <liangent> saving a claim [20:46:15] <liangent> it saves the claim and exit from edit mode, right? [20:46:27] <Denny_WMDE> for saving you use the ID [20:46:50] <Denny_WMDE> it does not matter what label you are showing, or if you are even showing any label [20:47:26] <liangent> after saving, the interface is updated to show the entity's label [20:48:16] <Denny_WMDE> where doesi thought we were talking about wbsearchentities? [20:48:37] <liangent> because the label is taken from wbsearchentities :/ [20:49:14] <liangent> the label to show on ui after saving a claim pointing to an entity fetched from wbsearchentities [20:49:56] <gerrit-wm> Change merged: jenkins-bot; [mediawiki/extensions/Ask] (master) - https://gerrit.wikimedia.org/r/74052 [20:49:56] <gerrit-wm> Change merged: jenkins-bot; [mediawiki/extensions/Ask] (master) - https://gerrit.wikimedia.org/r/74051 [20:50:26] <gerrit-wm> Change merged: jenkins-bot; [mediawiki/extensions/Ask] (master) - https://gerrit.wikimedia.org/r/74055 [20:52:18] <JeroenDeDauw> Denny_WMDE: I see our new url data type is currently using StringValue - why not IriValue? [20:53:28] <Denny_WMDE> liangent: well, that's not too terrible. I would start with just using the label then [20:53:34] <Denny_WMDE> we can later try to fix this [20:54:25] <Denny_WMDE> JeroenDeDauw: don't know. i thought you would have discussed this with daniel? [20:54:25] <gerrit-wm> New patchset: Hashar; "Make ReferencedEntityFinder use Entity::getAllSnaks" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/72962 [20:54:33] <gerrit-wm> New review: Hashar; "retriggering for https://bugzilla.wikimedia.org/show_bug.cgi?id=51461" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/72962 [20:55:02] <JeroenDeDauw> Denny_WMDE: he did not ask me. Nor was I even added as reviewer [20:55:33] <liangent> Denny_WMDE: ok then will I be fixing it by making an extra call to wbgetentities, or will there be a better solution? [20:55:37] <Denny_WMDE> JeroenDeDauw: don't know. ask him tomorrow [20:55:38] <gerrit-wm> New review: Jeroen De Dauw; "(1 comment)" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73993 [20:56:52] <Denny_WMDE> liangent: i don't know. maybe changing wbsearchentities to include the language label. [20:57:04] <hashar> hey [20:57:27] <Denny_WMDE> liangent: a call to wbgetentities is possible, too. whatever is simpler. [20:57:28] <hashar> so there was an evil bug in the configuration of mwext-Wikibase-testextensions-master [20:57:36] <Denny_WMDE> hashar: hey [20:57:42] <hashar> I got it fixed by refreshing the configuration https://bugzilla.wikimedia.org/show_bug.cgi?id=51461 [20:57:52] <hashar> basically, it was not properly fetching the extension dependencies :( [20:57:57] <liangent> Denny_WMDE: ok [20:58:39] <Denny_WMDE> hashar: thanks! [20:58:50] <Denny_WMDE> i remember daniel being loud about this one today [20:59:13] <hashar> yeah I guess I owe your team a bunch of rounds of beers [20:59:38] <hashar> we really need to work on composer to resolve the dependencies :-] [21:00:00] <Denny_WMDE> i think that would be appreciated :) [21:01:41] <JeroenDeDauw> hashar: yeah, we should really get https://gerrit.wikimedia.org/r/#/c/73760/ merged :p [21:02:59] <hashar> :-] [21:03:01] <JeroenDeDauw> Denny_WMDE: do you have any thoughts on this? http://programmers.stackexchange.com/questions/204837/component-design-getting-cohesion-right [21:03:16] <JeroenDeDauw> I want to get on with this but first get some feedback... [21:03:22] <JeroenDeDauw> No one replying tho :/ [21:09:40] <JeroenDeDauw> hashar: y u no merge http://inglorioushipsters.com/comic/wp-content/uploads/2012/01/sad_kitten.jpg [21:10:04] <hashar> JeroenDeDauw: cause I need to look at the code ? :-D [21:10:23] <hashar> and probably actually try it hehe [21:10:27] <JeroenDeDauw> hashar: all 6 lines that I moved! [21:10:45] <Denny_WMDE> JeroenDeDauw: wait longer for input and work on the WikibaseQuery API instead? [21:10:50] <JeroenDeDauw> hashar: I geuss this counts as me rewriting our entire composer support? :D [21:11:21] <JeroenDeDauw> Denny_WMDE: my work on WikibaseQuery is not blocked by this, or much affected at this point at all [21:11:43] <Denny_WMDE> JeroenDeDauw: exactly. [21:11:51] <JeroenDeDauw> Denny_WMDE: I do however want this split to happen soonish, its really starting to be annoying [21:12:30] <JeroenDeDauw> Denny_WMDE: I can work on WikibaseQuery and rage about other things in parallel, so no need to worry [21:19:36] <JeroenDeDauw> Danwe_WMDE: [21:19:37] <JeroenDeDauw> https://github.com/wikimedia/mediawiki-extensions-DataTypes [21:19:42] <JeroenDeDauw> ALL of the relevant history [21:21:25] <Denny_WMDE> JeroenDeDauw: re component design [21:21:40] <Denny_WMDE> what you would have in the end is a set of abstract interfaces, that basically never change again [21:21:58] <JeroenDeDauw> Denny_WMDE: yeah [21:22:02] <Denny_WMDE> and a number of implementations, that are much less stable [21:23:53] <Denny_WMDE> putting the interfaces in one component sounds right [21:24:22] <Denny_WMDE> and then suddenly it does not matter where the implementations are [21:24:36] <Denny_WMDE> they could be in the program using the component [21:24:48] <Denny_WMDE> they could be in a component of its own for a particular hard implementation [21:25:13] <Denny_WMDE> they could be some already in the component containing the interfaces [21:25:27] <Denny_WMDE> think of lucene or almost any larger software package [21:26:29] <Denny_WMDE> just because it is in the same library, doesn't mean it depends on it [21:27:08] <Denny_WMDE> only if a specific implementation is pulled in, you also add the implementation as a dependency [21:28:14] <Denny_WMDE> this would allow not to have to reimplement the hierarchy in all different components and have them more or less be aware of each other [21:28:32] <Denny_WMDE> not explicitly, but spookily somehow [21:29:08] <Denny_WMDE> i think queryresultprinters in smw is a nice example [21:29:19] <Denny_WMDE> smw itself offers the interfaces [21:29:30] <Denny_WMDE> srf offers a few more implementations [21:29:37] <Denny_WMDE> but there can be also implementations in other places [21:29:43] <gerrit-wm> New patchset: Jeroen De Dauw; "Move out DataTypes to its own repo" [mediawiki/extensions/DataValues] (master) - https://gerrit.wikimedia.org/r/74065 [21:30:26] <gerrit-wm> New review: Jeroen De Dauw; "Awaiting preparations elsewhere" [mediawiki/extensions/DataValues] (master) C: -2; - https://gerrit.wikimedia.org/r/74065 [21:31:34] <Denny_WMDE> and if i wanted to implement something to deal with chemical formulas, i can do it in one place, and offer the implementation as a realizer of the one component [21:31:37] <hashar> JeroenDeDauw: make sure the new ext ends up in wmf branches and get included in the mediawiki-config :-] [21:31:49] <hashar> JeroenDeDauw: I guess beta will complain anyway [21:34:39] <hashar> sleep time [21:36:21] <gerrit-wm> New patchset: Jeroen De Dauw; "Update magical loading code so it keeps working after DataValues split" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/74066 [21:37:58] <gerrit-wm> New review: Daniel Werner; "Wondering why this is unsing the string DV type and not the iri one. Will you change this later?" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73993 [21:38:29] <Denny_WMDE> JeroenDeDauw: in my opinion, a library should not get a 1.0 before it is used by two projects [21:38:33] <JeroenDeDauw> Denny_WMDE: Agree with the stuff you wrote, though it does not seem to reach a conclusion further then usage of interfaces and polymorphism being nice [21:38:58] <aude> JeroenDeDauw: can we please do the split after we branch stuff tomorrow? [21:39:10] <JeroenDeDauw> Denny_WMDE: except "just because it is in the same library, doesn't mean it depends on it" - though perhaps you mean something differently then I am reading [21:39:13] <aude> it will allow time for the dust to settle :) [21:39:33] <JeroenDeDauw> aude: yeah sure, this is ahy I -2 my current submitted change for DataTypes [21:39:41] <aude> k [21:40:03] <aude> we've got enough settings changes this time to add wikivoyage and other new features [21:41:54] <JeroenDeDauw> Denny_WMDE: re 1.0: mhh, that might work for some things, though does not seem very applicable here [21:42:03] <Denny_WMDE> JeroenDeDauw: what i meant with that sentence was that if an implementation in a library changes, but you are not using that implementation, then you do not care about that change [21:42:11] <Denny_WMDE> JeroenDeDauw: why? [21:42:43] <JeroenDeDauw> Denny_WMDE: I do not except huge changes to come about that warant bumping the primary version number. Bimping it just because wikibase started using it is odd, as this is not actually a change to the lib itself at all [21:43:23] <JeroenDeDauw> Denny_WMDE: do you mean the impelemntation of an interface? [21:43:52] <Denny_WMDE> so you are saying you got it right, the library is reusable as a library, and it is good and stable - but it is actually not used besides in a single project [21:43:58] <Denny_WMDE> how do you know that you got it right? [21:44:33] <Denny_WMDE> how do you know you are not merely reflecting the idiosyncrasies of the single project that uses the library? [21:44:51] <Denny_WMDE> or, for that matter, of a single dev or dev team? [21:45:06] <JeroenDeDauw> Denny_WMDE: Keep in mind this thing was not written from scratch [21:45:14] <JeroenDeDauw> The core of it has basically been around for years [21:45:41] <JeroenDeDauw> Denny_WMDE: And no, I'm not sure there will be no huge changes [21:45:57] <JeroenDeDauw> Denny_WMDE: I however think it is unlikely for those to happen [21:46:23] <Denny_WMDE> because...? [21:47:16] <JeroenDeDauw> Many factors, intuition [21:47:33] <JeroenDeDauw> Denny_WMDE: and if some huge change does come about, 2.0 works just fine [21:48:09] <JeroenDeDauw> Denny_WMDE: I anticipate changes will happen to this library [21:48:20] <JeroenDeDauw> Though it seems much mroe likely they will be iteratively [21:48:30] <JeroenDeDauw> And only warrant secondary version nr bumps [21:49:01] <JeroenDeDauw> If you where to not release for a year, and then consider all the changes, then sure, a primary bump might make more sense [21:50:22] <Denny_WMDE> a library that is only used by a single project is like code that is not tested [21:55:35] <JeroenDeDauw> Denny_WMDE: I do see what you are saying [21:55:44] <JeroenDeDauw> It is very hard to make truely reusable software [21:56:00] <Denny_WMDE> i definitively agree with that :) [21:56:03] <JeroenDeDauw> And the best way to avoid doing it wrong is designing it for multiple usages at the same time [21:56:17] <JeroenDeDauw> Denny_WMDE: this applies to the Ask code in SMW [21:56:22] <JeroenDeDauw> It is not reusable as is [21:56:32] <JeroenDeDauw> So we looked at how to change it so it would work for both SMW and Wikidata [21:56:48] <Denny_WMDE> and once it does actually work for both [21:57:10] <Denny_WMDE> that's the moment where i'd say, this seems to be reusable [21:57:20] <JeroenDeDauw> Denny_WMDE: do you anticipate we'll need serious changes to Ask to make it work for both? [21:57:37] <Denny_WMDE> no, i do not anticipate these [21:58:59] <JeroenDeDauw> "reusability" is also ofc not binary, so I find it odd to say "this needs to happen first before we can consider it as such" [21:59:17] <JeroenDeDauw> Denny_WMDE: and one also has to consider the difference in types of components [21:59:24] <JeroenDeDauw> Some are easier to make reusable then others [21:59:34] <JeroenDeDauw> For some it simply does not make sense to even attempt it [21:59:42] <Denny_WMDE> code quality is also not binary, nevertheless you would require tests in order to consider it as such — why is that odd? [22:00:25] <JeroenDeDauw> Denny_WMDE: in case of Ask, which pretty much just defines PHP respresentation of some language, it is sort of hard to mess up reusability if you pay attention to it [22:01:06] <JeroenDeDauw> Denny_WMDE: of course the language itself can turn out to not be reusable [22:01:20] <Denny_WMDE> for those where it makes no sense to attempt to make it reusable, i don't see the advantage of making it a library in the first place [22:01:22] <JeroenDeDauw> Though that'd mean not using Ask, rather then modifying it to be something else [22:02:01] <JeroenDeDauw> Denny_WMDE: yeah - I did not mean to imply one should. In fact I was saying one should not [22:04:22] <Denny_WMDE> well, coming back to the original question — i do not see the advantage of splitting the Data* into several different components unless we actually see concrete users for these components who would benefit from that split [22:05:21] <JeroenDeDauw> Denny_WMDE: right now every SMW install running master is loading ValueView and DataTypes [22:06:17] <Denny_WMDE> and are they supposed to use the validators in the mid term? [22:07:00] <JeroenDeDauw> Denny_WMDE: the ValueValidators where originally created for ParamProcessor, which is used by SMW [22:07:02] <Denny_WMDE> good point [22:07:11] <JeroenDeDauw> SMW itself could also use the, but that'd mean reworking the whole DV/DI system firsyt [22:07:29] <JeroenDeDauw> That is the big thing to do to increase interoperability [22:07:36] <JeroenDeDauw> With emphasis on "big" [22:08:04] <JeroenDeDauw> Denny_WMDE: though ATM as maintainer of SMW, I'm not willing to even start with this [22:08:11] <JeroenDeDauw> Since there is no release of DataValues [22:08:15] <JeroenDeDauw> Or the other components [22:08:25] <JeroenDeDauw> And they are all in one repo, like if it is one big component [22:08:34] <JeroenDeDauw> While it contains very different things [22:08:47] <JeroenDeDauw> Including things that are not, and are not going to be, relevant to SMW [22:09:05] <JeroenDeDauw> The same is true for anything else using ParamProcessor at this point [22:09:09] <JeroenDeDauw> ie Maps and SubPageList [22:09:14] <JeroenDeDauw> I'd like to make a new SPL release [22:09:18] <JeroenDeDauw> But there is no DV release [22:09:20] <JeroenDeDauw> ... [22:09:26] <Denny_WMDE> of most libraries that i use i do not use all available functions [22:09:42] <JeroenDeDauw> And if we regard this whole thing as one big component, there is not going to be a release any time soon, as there is much unfinished stuff in there [22:10:07] <JeroenDeDauw> Denny_WMDE: this is about releases mostly [22:10:40] <JeroenDeDauw> There also is the "needless affected by stuff I do not care about" part, though that is less of an issue [22:11:00] <Denny_WMDE> i do not understand that point [22:11:07] <JeroenDeDauw> Which point? [22:11:22] <Denny_WMDE> if the interfaces do not change, why would you be affected? [22:11:34] <JeroenDeDauw> Denny_WMDE: also, one cannot actually fully satisfy all component principles, they pull in different directions - see the triangle on the pillar in the office [22:11:37] <Denny_WMDE> "There also is the "needless affected by stuff I do not care about" part," [22:11:58] <JeroenDeDauw> Denny_WMDE: that is a reply to "of most libraries that i use i do not use all available functions" [22:12:16] <Denny_WMDE> right, i do not understand how this is a reply [22:12:19] <JeroenDeDauw> Which I presume to be a reply to "Including things that are not, and are not going to be, relevant to SMW" [22:12:30] <Denny_WMDE> which it was [22:13:06] <JeroenDeDauw> I'm saying this is not my main concern with current DataValues [22:13:31] <JeroenDeDauw> If there is some code in some component being loaded that is not needed, this is not an issue [22:13:49] <Denny_WMDE> why loaded? [22:13:55] <JeroenDeDauw> For instance, loading DataValues while only needing some of its basic values, and not stuff such as Geo or Multilang is fine [22:14:23] <JeroenDeDauw> I can even live with loading a component I don't actually need for now (such as ValueView) [22:14:51] <JeroenDeDauw> The blocking issue is lack of releases - if something does not have releases, it is not reusable [22:14:56] <Denny_WMDE> wait - [22:15:07] <JeroenDeDauw> The granule of reuse is the granule of release -- RCM [22:15:46] <Denny_WMDE> why would they be loaded? [22:16:28] <JeroenDeDauw> Denny_WMDE: remember all those things that assume you only have one extension per repo? [22:16:40] <JeroenDeDauw> root/DataValues.php [22:16:50] <JeroenDeDauw> That "entry point" will load all actual entry points [22:17:09] <Denny_WMDE> but only to register them for autoloading if required, no? [22:17:19] <Denny_WMDE> it's not like we load the whole tree? [22:17:22] <Denny_WMDE> or am i wrong? [22:18:02] <JeroenDeDauw> Denny_WMDE: by "loading component X" I mean that component X is registered. It means its autoloader is registered. It does not mean every class in it got loaded. [22:20:21] <Denny_WMDE> hmm, i fail to understand the badness of this, but let's get back to the issue [22:22:02] <JeroenDeDauw> Denny_WMDE: for instance, ValueParsers adds an API module. No reason for it to be available on SMW installs. And ValueView adds resource loader modules, which people can access as well [22:22:27] <JeroenDeDauw> Plus if any of those things breaks for some reason, my code that does not even care about it, also breaks [22:22:49] <JeroenDeDauw> Or rather, someones website, that has no use for the broken code, breaks [22:22:53] <Denny_WMDE> hmm, ok, those examples convince me [22:23:15] <lazowik> JeroenDeDauw: just sent you an email [22:23:37] <JeroenDeDauw> Denny_WMDE: also, I'd like to be able to install my SMW via composer [22:23:50] <JeroenDeDauw> By specifying stable versions of its dependencies [22:23:59] <JeroenDeDauw> That is not going to happen is stuff stays in the same repo [22:24:22] <JeroenDeDauw> Same thing for SubPageList or Maps [22:24:46] <JeroenDeDauw> Denny_WMDE: you where talking about how usage by multiple projects shows how reusable your code actually is [22:25:06] <JeroenDeDauw> Denny_WMDE: while this process of no releases and everything in one repo works for Wikidata.org, it clearly hinders reusability [22:25:21] <JeroenDeDauw> Or rather, pretty much prevents reuse [22:26:21] <JeroenDeDauw> Denny_WMDE: I anticipate pain for me as an extension maintainer reusing DataValues in its current state, even while I have a better overview then anyone else on what is going on there and how it affects things [22:26:50] <Denny_WMDE> ok, let's make it so. [22:26:56] <JeroenDeDauw> Denny_WMDE: if some dev from the php community stumbles accross it on packagist, and then looks at the source, they'll be "no, not using this stuff" [22:28:23] <JeroenDeDauw> lazowik: will look at it tomorrow, done with using my brain for today [22:28:29] <lazowik> :) [22:29:24] <Denny_WMDE> me too [22:29:27] <Denny_WMDE> i need to sleep [22:29:44] <Denny_WMDE> regarding your question on stackexchange [22:41:53] <JeroenDeDauw> Denny_WMDE: without the thing I'm speculating about in that question, these components will end up right/above the line here http://pdepend.org/screenshots/full/jdepend.html [22:46:08] <Denny_WMDE> and off. gn8 [23:47:48] <Sven_Manguard> anyone here speak Japanese? [23:53:58] <Guerillero> no [23:54:17] <Guerillero> but I did buy steam games last night

[08:30:58] New patchset: Daniel Kinzler; "Fix conflict detection in MapPatcher." [mediawiki/extensions/Diff] (master) - https://gerrit.wikimedia.org/r/73933 [08:31:37] New patchset: Daniel Kinzler; "Fix conflict detection in MapPatcher." [mediawiki/extensions/Diff] (master) - https://gerrit.wikimedia.org/r/73933 [08:53:33] New review: Liangent; "This particular line isn't problematic currently because ->language is always eagerly constructed fo..." [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73854 [08:55:41] New patchset: Tobias Gritschacher; "Make use of Serializers in API results for sitelinks" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/69689 [08:55:57] New patchset: Henning Snater; "jQuery.ui.suggester: Proper event handler management" [mediawiki/extensions/DataValues] (master) - https://gerrit.wikimedia.org/r/73937 [09:06:10] New patchset: Henning Snater; "jQuery.wikibase.entityselector: Reposition menu when refreshing" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73938 [09:16:16] DanielK_WMDE_: Tobi_WMDE Denny_WMDE https://bugzilla.wikimedia.org/51376 [09:17:56] aude: I can look into it a bit later.. [09:18:09] ok [09:19:42] New patchset: Addshore; "Ask DeserializerFactory and Tests" [mediawiki/extensions/Ask] (master) - https://gerrit.wikimedia.org/r/73786 [09:33:13] New patchset: Henning Snater; "Converted wikibase.ui.Tooltip to jQuery widget" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73939 [09:41:11] New patchset: Addshore; "move and fix wblinktitles validation" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73941 [09:43:47] Is it just me or is jenkins broken? O_o [09:44:12] works for me 8 min ago [09:44:21] Cache is not usable if WikibaseClient is configured for direct access to the repo database [09:44:41] oh our tests are broken? [09:44:52] :> [09:45:24] i think DanielK_WMDE_ is aware of the propertylist unseriailzer thing [09:45:32] :> [09:45:46] addshore: regarding https://gerrit.wikimedia.org/r/#/c/73316/ [09:45:53] all the skips ones are okay [09:46:04] somwhere along the way of patching array_values() must be used [09:46:19] because the last test checks removing against patching [09:46:27] and patching also breaks things [09:47:01] New review: Addshore; "I think we can ignore the failed tests and re run jenkins later." [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73941 [09:47:01] aude: "aware" is saying to much. it randomly happens to me. [09:47:01] i have no clue why [09:47:01] :> [09:47:01] oh, okay [09:47:08] well it's appearing on jenkins now [09:48:04] addshore: also try opening the page not being logged in [09:48:25] ahh yes, you wont be able to see the draft :P [09:48:41] I imagine you can see it now you are a reviewer though? [09:48:48] yeo [09:48:49] *p [09:48:56] that's why I added myself [09:49:00] :> [09:49:03] it's a gerrit bug [09:49:20] found out only because Status is RULE_ERROR [09:49:23] that was strange [09:49:37] *goes to find which test makes it fail again* [09:50:28] I thought https://gerrit.wikimedia.org/r/#/c/73797 would be OK [09:50:49] but patcher can get arrays with gaps [09:50:57] and then things would've broken [09:51:09] New patchset: Henning Snater; "Converted wikibase.ui.Tooltip to jQuery widget" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73939 [09:52:06] addshore: ^ [09:53:53] New patchset: Tobias Gritschacher; "Make use of Serializers in API results for sitelinks" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/69689 [09:54:50] New review: Daniel Kinzler; "@Jeroen: I think ListDiffer/ListPatcher should treat the array as a *list*, that is, completely igno..." [mediawiki/extensions/Diff] (master) - https://gerrit.wikimedia.org/r/73797 [09:54:54] New review: Tobias Gritschacher; "(1 comment)" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/69689 [09:55:12] DanielK_WMDE_: done: https://gerrit.wikimedia.org/r/#/c/69689/ [09:57:29] New review: Tobias Gritschacher; "Hey!" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/62598 [09:57:43] New review: Tobias Gritschacher; "Hey!" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/63857 [09:58:23] aude, Denny_WMDE, Tobi_WMDE: this is currently causing conflicts to be silently overwritten: https://gerrit.wikimedia.org/r/#/c/73933/ [09:58:43] well, the patch fixes it, it doesn't casue it, obviously [09:59:29] DanielK_WMDE_: that would have been my next question.. :) [10:01:51] New patchset: Daniel Kinzler; "(bug 46352) Implement EntityHandler::getUndoContent" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73942 [10:23:56] New patchset: Addshore; "Improve errors in wblinktitles" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73943 [10:33:11] heh, Denny_WMDE > https://gerrit.wikimedia.org/r/#/c/73943/ [10:33:33] addshore: ? [10:33:44] build tests also fail :P [10:33:51] no merging :) [10:34:08] yeah, but once it's fixed, it will merge, no? [10:34:25] hehe *starts adding to the queue* [10:34:25] not automatically [10:34:35] working on it [10:34:36] i don't think.... [10:35:12] ah, we can probably kick it then to try again [10:42:30] New patchset: Daniel Kinzler; "ByPropertyListUnserializer must check structure." [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73945 [10:43:07] aude, addshore, Denny_WMDE: see above [10:43:32] please verify, merge and then rebase any changes for which jenkins fails on ByPropertyListUnserializerTest [10:48:05] Change merged: jenkins-bot; [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73945 [10:48:19] addshore: can i "rebase" https://gerrit.wikimedia.org/r/#/c/73316/ to allow jenkins to run? [10:48:30] and the link titles patches? [10:48:44] errr, https://gerrit.wikimedia.org/r/#/c/73316/ is in data model [10:48:50] the other ones, can i rebase? [10:50:57] * aude is going to "rebase" everything that jenkins failed in wikibase [10:51:18] DanielK_WMDE_: addshore Henning_WMDE Tobi_WMDE speak up if you are working on something particular that you don't want me to rebase :) [10:51:23] addshore, Tobi_WMDE, Henning_WMDE: maybe worth a try? http://chocolatey.org/ [10:52:36] New patchset: Aude; "move and fix wblinktitles validation" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73941 [10:53:39] New patchset: Aude; "Make use of Serializers in API results for sitelinks" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/69689 [10:54:00] New patchset: Aude; "(bug 46352) Implement EntityHandler::getUndoContent" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73942 [10:54:25] New patchset: Addshore; "Improve errors in wblinktitles" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73943 [10:55:10] addshore: http://blog.jetbrains.com/webide/2011/02/zero-configuration-debugging-with-xdebug-and-phpstorm-2-0/ [10:55:15] Henning_WMDE: gerrit won't allow me to rebase your patches. they need manual rebase probably [10:55:40] addshore: if it doesn't work for you, i'll have a look at what i did in my apache/xdebug config [10:55:53] okay :) *looks at the many links above* [10:56:04] Tobi_WMDE: hmmm, jenkins still says "no" [10:56:35] addshore: meanwhile, i'm just going to fix the bug :P [10:56:42] ;_; go for it ;p [10:57:47] New review: Aude; "looks like the SiteLinkSerializerTest.php has a dependency on wikibase repo, which is not available ..." [mediawiki/extensions/Wikibase] (master) C: -1; - https://gerrit.wikimedia.org/r/69689 [11:02:08] aude: looking at https://bugzilla.wikimedia.org/show_bug.cgi?id=51376 now. if nobody else did yet [11:02:25] I dont think so :) [11:03:00] Tobi_WMDE: not yet [11:04:40] Tobi_WMDE: also be aware of https://bugzilla.wikimedia.org/show_bug.cgi?id=50983 [11:05:31] which addshore is working on [11:06:09] mhhm, that bug itself is fixed from the patch [11:06:18] this is a different bug now ;p [11:06:27] errr, it's the one about the formats being different [11:06:39] we haz too many bugs :o [11:06:49] thats nothing to do with the alises being remove though :P [11:06:56] yeah [11:07:00] you had another bug for it ;p [11:07:05] https://bugzilla.wikimedia.org/show_bug.cgi?id=51192 [12:33:26] New review: Jeroen De Dauw; "(1 comment)" [mediawiki/extensions/WikibaseDataModel] (master) - https://gerrit.wikimedia.org/r/70443 [12:35:38] Change merged: Jeroen De Dauw; [mediawiki/extensions/Diff] (master) - https://gerrit.wikimedia.org/r/73933 [12:39:57] New patchset: Jeroen De Dauw; "Update rel notes and bump to 0.8 alpha" [mediawiki/extensions/Diff] (master) - https://gerrit.wikimedia.org/r/73955 [12:40:08] Change merged: Jeroen De Dauw; [mediawiki/extensions/Diff] (master) - https://gerrit.wikimedia.org/r/73955 [12:42:18] New patchset: Jeroen De Dauw; "(bug 51363) Add more tests to for bad diffs." [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73772 [12:58:17] New patchset: Addshore; "Entity removeAliases now resets array keys" [mediawiki/extensions/WikibaseDataModel] (master) - https://gerrit.wikimedia.org/r/73316 [12:58:24] DanielK_WMDE_: ^^ :) [12:58:50] everything should now be passing :) [12:59:19] lazowik: also ^^ [12:59:39] addshore: got an email, looking [13:00:43] yep [13:01:14] New review: Michał Łazowik; "Looks good" [mediawiki/extensions/WikibaseDataModel] (master) C: 1; - https://gerrit.wikimedia.org/r/73316 [13:02:15] New review: Jeroen De Dauw; "The ListPatcher is currently completely ignoring keys yes. This patch changes that" [mediawiki/extensions/Diff] (master) - https://gerrit.wikimedia.org/r/73797 [13:04:08] New patchset: Daniel Kinzler; "(bug 51363) Add more tests to for bad diffs." [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73772 [13:04:23] liangent: ? [13:05:22] New review: Michał Łazowik; "(1 comment)" [mediawiki/extensions/WikibaseDataModel] (master) C: -1; - https://gerrit.wikimedia.org/r/73316 [13:05:50] Denny_WMDE: hi [13:06:00] shall we? [13:06:35] liangent: you prefer to chat in this channel, or one on one? [13:06:47] Denny_WMDE: is DanielK_WMDE_ also around? [13:07:04] Denny_WMDE: two topics: 1) getentities & serialized form used in other places 2) searchentities [13:08:14] 1) was already mentioned in a mail, and I've already talked about 2) with you a few days ago [13:08:26] liangent: i would prefer to structure it a bit differently [13:08:32] re your priorities and the way the work is done [13:08:44] by getting guided by specific stories [13:08:51] i.e. functionalities for the end user [13:09:39] one story "readers of wikidata can see a label in a different language if their language is not available" [13:09:56] another story "editing ..." [13:10:00] one story "search" [13:10:02] etc. [13:10:17] search is tied with editing [13:10:22] yes [13:10:29] because our search relies on the editing [13:10:38] but the first story, display, is not [13:10:46] right [13:11:08] for display, simply replacing wbEntity and wbUsedEntities just work [13:11:11] as mentioned in the mail [13:11:38] but we also need to think about how to display it [13:12:06] yeah [13:12:40] but first, the infrastructure to supported [13:12:50] js library and data format [13:12:55] *support it [13:13:06] yes [13:13:41] for that also we don't need yet automatic translation etc. [13:13:46] let us start as simple as possible [13:13:51] and try to get visible results [13:13:57] then we will get feedback [13:14:03] while we make the backend more powerful [13:14:08] and add translation etc. [13:14:19] or rather, transliteration [13:14:54] well in backend, translation is handled transparently when fetching labels [13:15:16] but it is not yet displayed anywhere [13:15:26] which makes it basically dead functionality as far as the user is concerned [13:15:39] that means untested, no feedback, etc. [13:15:55] no it's already displayed [13:16:02] https://gerrit.wikimedia.org/r/#/c/71845/ https://gerrit.wikimedia.org/r/#/c/71865/ [13:16:23] one minor place though [13:16:32] that is a surprising place [13:16:33] but cool [13:16:37] let's get it more visible [13:17:56] yeah those page titles don't need and can't have complex s so I went with them first [13:19:15] ready for primetime? [13:19:55] i mean, shall we display it on the page next? [13:19:55] on the item view? [13:19:55] Denny_WMDE: yeah [13:19:56] New patchset: Addshore; "Entity removeAliases now resets array keys" [mediawiki/extensions/WikibaseDataModel] (master) - https://gerrit.wikimedia.org/r/73316 [13:20:30] liangent: is this for conversion? or fallback? [13:20:31] so what needs to happen to get there? [13:21:10] e.g. en-gb -> en or ku -> ku-arab [13:21:20] aude: upper layers don't distinguish them [13:21:27] liangent: ok [13:22:52] Change merged: Jeroen De Dauw; [mediawiki/extensions/WikibaseDataModel] (master) - https://gerrit.wikimedia.org/r/73316 [13:23:13] Denny_WMDE: choose a correct format to pass data to entity view <- this is the thing I'm now considering [13:24:13] with it done I'll tweak js code to show language tags when it realized that a label is from a fallback language [13:24:15] s/realized/realizes/ [13:25:38] can you go a bit into more detail what you mean with the correct format? [13:25:47] mhhm Denny_WMDE I have just found core uses 'missingparam' as an error code which is different to what we use :/ Should I change WB to be like core for this also? [13:26:05] i like consistency [13:26:09] :P [13:26:15] addshore: any reason against? [13:26:36] not really, other than cores error codes look silly because they dont have dashes in them :> [13:26:50] New review: Daniel Werner; "(1 comment)" [mediawiki/extensions/Wikibase] (master) C: -1; - https://gerrit.wikimedia.org/r/73942 [13:27:16] New review: Jeroen De Dauw; "(1 comment)" [mediawiki/extensions/Ask] (master) C: -1; - https://gerrit.wikimedia.org/r/73786 [13:27:37] Denny_WMDE: current serialized format writes " labels:{de:{value:'something',language:de}} " when the user is viewing de and there is a de label, or " labels:{} " if there's no de label, right? [13:27:52] labels:{de:{value:'something',language:'de'}} [13:27:57] addshore: then let's switch [13:28:07] everything to have no dashes? :O [13:28:22] liangent: i think so [13:28:31] i would rather change core to have dashes ;p [13:28:44] Denny_WMDE: I want to change it to labels:{de:{value:'something',language:'en'}} or labels:{qqc:{value:'something',language:'en'}} when no de label [13:28:50] i guess one way to change it would be to "de-ch : { value: something, language: de} ? [13:29:05] i like the first one you suggest, liangent [13:29:27] addshore: I understand, but let's say no :) [13:29:33] :> [13:29:35] okay that's what DanielK_WMDE_ thinks too [13:30:37] but there're many client code assuming de-ch must be identical to de in de-ch : { value: something, language: de} [13:30:37] as they say in Germany "two dumbs, one thought" [13:31:16] let's list those places and fix them? [13:31:22] or, I should say: they assume in labels:{de:{value:'something',language:'en'}} de is always identical to en [13:31:38] (to use a incompatible language pairs) [13:31:50] ok [13:32:30] now by fixing them, they'll add a language tag to the label when they find that de != en [13:33:24] sorry i was disconnected for a moment (stupid network) [13:33:34] and before this takes place (and after labels:{de:{value:'something',language:'en'}} is done), user is seeing en text in de interface without any indication [13:33:43] if you said something, please repeat before [13:34:28] DanielK_WMDE: Looks like I need to remove the line again from the diff rel notes. Regression got introduced after the previous release: https://gerrit.wikimedia.org/r/#/c/68652/3/includes/differ/MapDiffer.php [13:35:21] New patchset: Jeroen De Dauw; "Remove incorrect line from rel notes" [mediawiki/extensions/Diff] (master) - https://gerrit.wikimedia.org/r/73960 [13:35:47] New review: Jeroen De Dauw; "Regression was only present in 0.7 alpha, not 0.6." [mediawiki/extensions/Diff] (master) C: 2; - https://gerrit.wikimedia.org/r/73960 [13:35:48] Change merged: Jeroen De Dauw; [mediawiki/extensions/Diff] (master) - https://gerrit.wikimedia.org/r/73960 [13:36:24] Denny_WMDE1: which was the last sentence you received? [13:36:53] or, I should say: they assume in labels:{de:{value:'something',language:'en'}} de is always identical to en [13:38:13] Denny_WMDE1: and before this takes place (and after labels:{de:{value:'something',language:'en'}} is done), user is seeing en text in de interface without any indication [13:38:15] and no more [13:38:28] ok, thanks [13:38:43] JeroenDeDauw1: remove which line? [13:38:59] DanielK_WMDE: the one about the regression you fixed [13:39:01] liangent: are there many places where we have that assumption? [13:39:45] JeroenDeDauw1: ah, ok. uh... but the error was present in the 0.7 release, no? [13:40:19] the code was merged june 24 [13:40:19] DanielK_WMDE: no [13:40:29] 0.7 was just releases yesterday [13:40:34] Do you mean 0.6? [13:40:56] no i mean that yesterday, the error was on master. [13:41:00] so the error is in 0,7 [13:41:16] am i missing something? [13:41:31] DanielK_WMDE: I merged it remember? Looking at the stuff is what promted me to release [13:41:48] How else would I have mistakenly have put the line in the rel notes? [13:41:52] .confusion [13:42:07] Denny_WMDE1: of course everywhere [13:42:32] basically the library just look at keys in json objects [13:42:38] and ignore the language attribute in it [13:42:46] JeroenDeDauw1: i fixed two bugs, one in MapDiffer (merged yesterday) and ine in MapPatcher (merged today) [13:43:25] liangent: oh, access to that field is not mediated by a function? [13:43:28] JeroenDeDauw1: if the first bug prompted the release of 0.7, the second one would go into 0.8, right? [13:44:42] even if there is, the function just return a single label, and can do nothing if it finds some inconsistency [13:44:57] except for throwing an error, or refusing to return a label [13:45:29] New patchset: Jeroen De Dauw; "Move stuff to src/ and remove duplicate testsuite" [mediawiki/extensions/WikibaseQueryEngine] (master) - https://gerrit.wikimedia.org/r/73961 [13:45:50] Change merged: Jeroen De Dauw; [mediawiki/extensions/WikibaseQueryEngine] (master) - https://gerrit.wikimedia.org/r/73961 [13:46:00] it cannot return HTML? that would solve everything, no? [13:46:47] Denny_WMDE1: I haven't really checked it [13:47:04] did you have an alternative plan? [13:48:12] Denny_WMDE1: so that's why I've thought about qqc [13:48:37] and add a getPreferredLabel or something in js lib [13:48:59] which returns a js dict (object) maybe [13:49:07] to include language info in it [13:49:52] so callers aware of qqc or getPreferredLabel (which must be added afterwards) know about this possible inconsistency too [13:50:08] to avoid breakage of existing stuff [13:50:54] but this doesn't look so good in terms of long time plan [13:51:10] (to have a better serialized format maybe [13:51:54] DanielK_WMDE: ^ [13:52:10] re your mail: I'm talking about that with denny here [13:52:27] i just told our Danwe Henning_WMDE and TobiasG_WMDE to join us as well, they have done most of the work on the front end [13:52:42] liangent: so, the alternatives are: a) ask for a "real" language + fallback mode, or b) as for a real language (no fallback) or alternatively qqc (all fallback). [13:52:46] did i get that right? [13:53:07] the latter would need no change to the data structure [13:53:18] both sound fine [13:53:30] the qqc approach has several drawbacks [13:53:51] e.g. you don't know what language the returned value is in. you can't apply direction tags, etc [13:54:00] the qqc proposal would also refrain from overloading the meaning of the other label-keys and not introduce unintentional side-effects [13:54:08] you can't generate links for editing it (you don't know the original language), etc [13:54:34] Denny_WMDE1: we'd lose/lack a lot of valuable information [13:54:37] DanielK_WMDE: why? the qqc key would still point to an object of the form { value : Vienna, language: de } [13:54:39] DanielK_WMDE: the real language in qqc is indicated by language attrib [13:54:53] and don't we have a client side language info database [13:54:54] if we always have that, fine [13:54:59] I mean directionality etc [13:55:08] i thought the current data structure doiesn't allow for that? did i get that wrong? [13:55:18] allow for what? [13:55:23] DanielK_WMDE: i think you did. it's always an object. [13:55:31] not a plain key -> string [13:55:34] ok then [13:55:51] right, it's luckily not using the internal serialization [13:56:11] so, if that is the case, i see even less use for qqc. [13:56:25] less use for or less issues with? [13:56:36] both [13:57:38] :-/ [13:58:00] and dropped again after my :-/ [13:58:03] Denny_WMDE1: what advantage does it have? [13:58:18] less unintentional side-effects [13:58:19] liangent: my approach to B/C would be to not apply any fallback u8nless the client asks for it. [13:58:36] the only place where that doesn't work is wbEntity, I think [13:59:13] Denny_WMDE1: you mean for peopple assuming that the value for a given language will always be in that language? [13:59:13] if you have a de : { value : Vienna, language : en } a client might currently just use de.value instead of also checking de.language [13:59:20] aye [13:59:37] i dont know which is better [13:59:41] these are the two options [13:59:46] New patchset: Tobias Gritschacher; "(bug 51376) Pass missing SerializationOptions" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73962 [13:59:46] that's why I think fallback should only be apüplied if the client asks for it [13:59:50] DanielK_WMDE: yeah. about wbEntity, for code using the builtin wikibase js lib, it might be easy: just let getLabel return nothing when it finds a mismatch (key and language attrib) [13:59:51] then the client also knows about it [13:59:58] Danwe: Henning_WMDE TobiasG_WMDE — any opinion? what would be easier to deal with in the UI? [14:00:37] liangent: yea, though it should also be possible to get the lable with fallback. getLabelObject or getLabelWithFallback or some such [14:01:19] * DanielK_WMDE thinks that wbEntity is an implementation detail and not part of our interface, and anything relying on it has to deal with the consequences [14:01:26] DanielK_WMDE: yeah but what about codes (usually user / site scripts) accessing wbEntity directly? [14:01:44] maybe "do we want to well support it"? [14:02:14] I don't, but that'S up to Danwe/Henning [14:03:18] aude: addshore: fix for the xml result issuw: https://gerrit.wikimedia.org/r/#/c/73962/ [14:04:35] w00t! :) [14:04:44] how do i reproduce the bug? [14:04:52] to test the fix? [14:05:20] api sandbox? [14:05:23] aude: just use the apisandbox [14:05:26] ok [14:05:45] and try to change a description or aliases with wbeditentity [14:05:59] and set the result format to xml [14:06:03] ok [14:06:08] *looks* :) [14:06:14] it worked for labels, because there I passed the options [14:06:17] * aude does not understand why people use xml in the api, but okay [14:06:22] just forgot them for descriptions and aliases [14:06:27] ok [14:06:30] easy enough then [14:06:35] ill let you test it aude :) [14:06:42] ok [14:06:47] no point in 2 people testing such a tiny change ;p [14:06:53] yep [14:06:59] it needs backporting [14:07:47] aude: I can backport once it's merged [14:07:52] or just du it yourself [14:08:00] *do [14:08:04] diy! [14:08:07] i can take care of it [14:08:24] i might cherry pick hennings patch for the edit links link thing [14:08:39] so the link goes to the site links section [14:09:11] #sitelinks-wikipedia ? [14:09:17] yep [14:09:47] *runs tests* [14:11:53] liangent: we had a long and loud discussion right here now [14:12:24] Denny_WMDE1: :) [14:12:30] and you want to use another channel? [14:12:34] no [14:12:43] i mean, here in the office we had a very loud discussion [14:12:57] about whether to use ggc or lang-key with inconsistent language value [14:13:16] and we all agree on the language-key [14:13:37] and update the front end code [14:13:49] to deal with appropriately [14:13:49] ：） [14:13:49] :) [14:14:14] but unfortunately there will be a bit of a mess, because the label on the top is done very differently from the labels in the statements [14:14:29] hmm [14:15:15] but Danwe and Henning_WMDE will help you there [14:15:28] they probably will also be the ones mostly reviewing the javascript part [14:15:43] and they can point you to the right place and help answering questions [14:17:23] we can try in experimental first to see what happens if you just add a fallback language about a referenced item [14:17:28] that is used int he statements [14:17:43] and then start fixing them where they appear [14:20:11] in this approach I may need to fix js earlier than changing php output [14:20:31] depends. mark the changed php output as experimental [14:20:35] then it doesnt go into production [14:20:46] how is this done? [14:21:09] you mean to use a series of dependent changesets [14:21:26] or you have some mechanism to avoid some changesets in deployment? [14:21:34] eg revert them in deployment branch? [14:22:29] we have a mechanism to have some pieces not be executed in deployment [14:22:41] what do you want experimental? [14:23:05] Denny_WMDE1: any guide for using it? [14:23:06] adding "de: { value : language, language : en }" to the entityview for the dom to be available [14:23:23] liangent: i hoped DanielK_WMDE and aude can help with that [14:24:00] liangent: instead of registering hooks, api modules, resource loader stuff, etc in the main config/setup files, they can go in the config/Wikibase.experimental files [14:24:07] there also is a constant [14:24:15] New review: Daniel Kinzler; "(1 comment)" [mediawiki/extensions/WikibaseDataModel] (master) - https://gerrit.wikimedia.org/r/73316 [14:24:24] Change merged: jenkins-bot; [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73962 [14:24:35] can be used to switch off certain bits of code, but it's better to put stuff in the experimental file if possible [14:24:40] easier to maintain and keep track of [14:25:03] in javascript, i have no idea how to do experimental except put resource loader modules in the experimental setup [14:25:48] that's for the php now. [14:26:16] the frontend doesnt seem to know about experimental [14:29:22] so it should inject the fallback labels in the php side, based on the experimental flag [14:29:33] and the front end should be tweaked to learn to deal with it until it does [14:29:40] and then we switch that feature on [14:29:47] and make the world a better place :) [14:30:18] New patchset: Daniel Kinzler; "Create Stores from settings via builder function." [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/72939 [14:30:38] liangent: ^^ [14:31:15] New review: Jeroen De Dauw; "(1 comment)" [mediawiki/extensions/WikibaseDataModel] (master) - https://gerrit.wikimedia.org/r/73316 [14:31:17] Published patchset: Addshore; "Api errors now similar to core" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/73966 [14:31:45] addshore: are you addressing DanielK_WMDE his comment here? https://gerrit.wikimedia.org/r/#/c/73316/ [14:31:47] Denny_WMDE1: got it [14:31:53] JeroenDeDauw1: just about to :0 [14:31:56] :) [14:32:13] not again ... [14:32:30] rebase, rebase, rebase, ... [14:32:47] :D [14:34:47] JeroenDeDauw1: you added comment to Patch Set 3 whereas the last one is 4: https://gerrit.wikimedia.org/r/#/c/73316/ [14:35:47] New patchset: Addshore; "Clear aliases when setting all aliases" [mediawiki/extensions/WikibaseDataModel] (master) - https://gerrit.wikimedia.org/r/73968 [14:36:29] New patchset: Addshore; "Clear aliases when setting all aliases" [mediawiki/extensions/WikibaseDataModel] (master) - https://gerrit.wikimedia.org/r/73968 [14:36:32] New patchset: Tobias Gritschacher; "Make use of Serializers in API results for sitelinks" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/69689 [14:36:38] lazowik: thanks for that :) I addressed them :) [14:36:51] New patchset: Jeroen De Dauw; "Work on store interface" [mediawiki/extensions/WikibaseQueryEngine] (master) - https://gerrit.wikimedia.org/r/73969 [14:36:59] I'm terrible with spaces :) [14:37:53] addshore: I'm terrible with times :) [14:38:52] lazowik: I know - I made the comment earlier but apparently did not submit it [14:39:02] ok [14:39:14] hoo: poke [14:39:37] addshore: it's not that bad: https://www.mediawiki.org/wiki/Manual:Coding_conventions/PHP#Spaces [14:39:50] Denny_WMDE1: should I add another attrib, like de : { value : Vienna, language : en, fallbackmode : user } for eg based on user babel info [14:40:03] * hoo waves at aude :) [14:40:07] or de : { value : Vienna, language : en, fallbackmode : system } for system fallback chain only [14:40:24] hoo: at one point recently, the item move test was failing on jenkins [14:40:30] do you remember which patch fixed it? [14:40:41] * aude not sure if it fails on the deployment branch [14:41:03] oh wait, dumb [14:41:05]