On Sun, Dec 6, 2015 at 7:30 PM, Bartosz Dziewoński matma.rex@gmail.com wrote:
On 2015-12-04 18:59, Adam Baso wrote:
I do wonder though if we've spent much time studying the ease of getting at least some part of oojs-ui split out or making it so that new stuff going forward is part of the oojs-ui family but it's not as monolithich?
Not any more than what is written at < https://phabricator.wikimedia.org/T113681#1673158%3E, as far as I know.
OOjs UI doesn't have a dedicated team (or even a dedicated person) supporting it. The changes are mostly driven by whatever the VisualEditor or Design teams need, and what they are willing to implement themselves (and to a lesser extent, Multimedia team and random people like Brad or Florian). This makes it difficult to make any big changes happen.
If OOjs UI is the thing that we're supposed to be using in the future for our UI stuff, it's very concerning that further development is blocked on T113681 but no one is planning on working on that task or feels ownership for the thing.
Is someone going to step up to actually own and maintain the thing? Or should we consider declaring it a failure and figure out a plan to move away from it, if no one wants to maintain our home-grown JS UI library anymore?
OOjs UI is used as a core component of MediaWiki's user interfaces, including heavy use in WMF-sponsored initiatives such as VE and Flow and slowly increasing use within core; as such it should probably be maintained by the MediaWiki Core team at Wikimedia Foundation.
Unfortunately since the mid-2015 Engineering department reorgs, there is no such team... If we can't lobby management to (re)create one, we'll have to set up a "cross-functional" skunkworks team if we want anything big to get done.
I would love to be involved, but I'm already stretched thin with multimedia-related commitments and probably don't have time to be the main point person. But if there's a few others who would like to chip in... maybe we can make some decisions and push some code around!
-- brion
On Wed, Dec 9, 2015 at 10:26 AM, Brad Jorsch (Anomie) <bjorsch@wikimedia.org
wrote:
On Sun, Dec 6, 2015 at 7:30 PM, Bartosz Dziewoński matma.rex@gmail.com wrote:
On 2015-12-04 18:59, Adam Baso wrote:
I do wonder though if we've spent much time studying the ease of getting at least some part of oojs-ui split out or making it so that new stuff
going
forward is part of the oojs-ui family but it's not as monolithich?
Not any more than what is written at < https://phabricator.wikimedia.org/T113681#1673158%3E, as far as I know.
OOjs UI doesn't have a dedicated team (or even a dedicated person) supporting it. The changes are mostly driven by whatever the VisualEditor or Design teams need, and what they are willing to implement themselves (and to a lesser extent, Multimedia team and random people like Brad or Florian). This makes it difficult to make any big changes happen.
If OOjs UI is the thing that we're supposed to be using in the future for our UI stuff, it's very concerning that further development is blocked on T113681 but no one is planning on working on that task or feels ownership for the thing.
Is someone going to step up to actually own and maintain the thing? Or should we consider declaring it a failure and figure out a plan to move away from it, if no one wants to maintain our home-grown JS UI library anymore?
-- Brad Jorsch (Anomie) Senior Software Engineer Wikimedia Foundation _______________________________________________ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
On Wed, Dec 9, 2015 at 2:40 PM, Brion Vibber bvibber@wikimedia.org wrote:
Unfortunately since the mid-2015 Engineering department reorgs, there is no such team...
I agree, such a team should be created. I'll be first in line to sign up, and as part of such a team I'd be happy to help "own" OOjs UI (although I'd not be comfortable being the lead on owning it).
If we can't lobby management to (re)create one, we'll have to set up a "cross-functional" skunkworks team if we want anything big to get done.
IMO a skunkworks team for critical infrastructure would be a sad outcome.
I would love to be involved, but I'm already stretched thin with multimedia-related commitments and probably don't have time to be the main point person. But if there's a few others who would like to chip in... maybe we can make some decisions and push some code around!
Me too, with the API and the AuthManager project. Plus I don't understand our JS ecosystem well enough to perform the requisite major surgery on OOjs UI without hand-holding from someone who does know how the packaging should work.
Short-cut answer to the title question: Me.
On 9 December 2015 at 10:26, Brad Jorsch (Anomie) bjorsch@wikimedia.org wrote:
On Sun, Dec 6, 2015 at 7:30 PM, Bartosz Dziewoński matma.rex@gmail.com wrote:
On 2015-12-04 18:59, Adam Baso wrote:
I do wonder though if we've spent much time studying the ease of getting at least some part of oojs-ui split out or making it so that new stuff
going
forward is part of the oojs-ui family but it's not as monolithich?
Not any more than what is written at < https://phabricator.wikimedia.org/T113681#1673158%3E, as far as I know.
OOjs UI doesn't have a dedicated team (or even a dedicated person) supporting it. The changes are mostly driven by whatever the VisualEditor or Design teams need, and what they are willing to implement themselves (and to a lesser extent, Multimedia team and random people like Brad or Florian). This makes it difficult to make any big changes happen.
If OOjs UI is the thing that we're supposed to be using in the future for our UI stuff, it's very concerning that further development is blocked on T113681
"Further development" is not blocked on this task. A few things that some people want to do are. Please do not exaggerate for effect to try to get your way. I'm sorry that we disagree as to whether your patch belongs in the library in its current form.
but no one is planning on working on that task or feels ownership for the thing.
The value-add of doing that accrues to the people currently not using OOUI. The work is, and has always been, a group effort by engineers, designers and product people making our front-end architecture more sane and consistent. Responsibility for change lies with those that want it and can justify the time, mostly. This is the same model as used for most shared things around MediaWiki. The passion, care and effort from volunteer and staff developers that goes into improving the library as people extend its use to more places is great. I'd love for there to be dedicated engineers to support the use of OOUI, but however.
Is someone going to step up to actually own and maintain the thing?
I am, and remain, the product owner responsible since 2012. The plan was to transfer the product/design side of it entirely to Volker, but we've not formally 'handed over the keys' yet. It doesn't appear that there's a particular rush, given the lack of movement to re-do how skins in MediaWiki work, as had been the ambition in 2013.
[Trim the rest.]
J.
On Wed, Dec 9, 2015 at 3:40 PM, James Forrester jforrester@wikimedia.org wrote:
Short-cut answer to the title question: Me.
I'm glad to hear that you are accepting responsibility for OOjs UI development! Do you have a timeline on a fix for T113681, or a page that indicates what higher-priority development you and your team are working on in the near future?
If OOjs UI is the thing that we're supposed to be using in the future for our UI stuff, it's very concerning that further development is blocked on T113681
"Further development" is not blocked on this task. A few things that some people want to do are.
Let's not chop logic here. If "a few things that some people want to do" cannot be done due to T113681, then T113681 is indeed blocking some further development even if other further development isn't blocked. This email thread isn't even about T113681 specifically, it's about that there are no development resources for fixing things in OOjs UI unless someone is willing to do it as a skunkworks project, and OOjs UI isn't yet a finished product where we might be able to justify that.
I'm disappointed that you don't think that the fact that "some things people want to do" are blocked and no development resources are available to remedy the situation is cause for concern. When the situation was brought up in today's Scrum of Scrums, the consensus was that it is indeed concerning.
Please do not exaggerate for effect to try to get your way. I'm sorry that we disagree as to whether your patch belongs in the library in its current form.
Since you brought it up, let's look at my patch. There are two concrete blockers that have been raised on my patch. Neither of them actually have to do with the form of the patch itself.
The long-standing blocker has been disagreement over how the widget can be internationalized in the context of OOjs UI: The Language and translatewiki faction wants OOjs UI developers to integrate cldrjs, while the OOjs UI developers are unwilling to make any decision as to whether cldrjs is the way to go or translatewiki will just have to deal with providing translations for month and weekday names as they do for everything else. The closest we have to a decision is really a cop-out: "shove it into MediaWiki even though it doesn't belong there, because MediaWiki already happens to have most of the needed i18n strings and we can't make any decision here".
In last week's Scrum of Scrums, you brought up T113681 as a new blocker: OOjs UI is already too large, so we can't add new stuff until someone reworks it to be able to load individual components. MatmaRex then stated that no one owns or maintains OOjs UI to the extent that we can expect T113681 to be solved any time soon, which brought the lack of maintainership in OOjs UI into clear view.
MatmaRex also raised some other objections (disagreement with Design's design, non-use of moment.js despite moment.js not gaining us anything, doubt that anyone actually needs <input type="datetime"> despite evidence to the contrary), but no one else has agreed with those and he hasn't deigned to respond to attempts at further discussion in Gerrit.
If you have objections to the actual form of my patch, as opposed to lack of a willingness to make any decision on the i18n issue or any progress on the form of OOjs UI as a whole, you should raise them in Gerrit instead of continuing to sit on them. Although I wonder why you haven't done so already.
This might be a good discussion for the dev summit?
I talked to Moriel about this a couple of days ago. I too am a bit concerned and feel like this needs a dedicated team, preferably without a product to manage and mediate/prioritise requests against it as otherwise the library will be biased towards a single product rather than all our products.
Ideally, I feel that we need a team determining how it evolves and its architecture. A big rewrite to split out OOjs UI into components/making it support mobile /adding a new component to OOjs UI is not something that should be done in an ad-hoc nature - it should be done by people with a vision of what this library needs to grow into, the problem it is solving and knowledge of its history and mistakes of the past - guardians as such - ensuring that the library is the best it can be.
I worry about its success if arranged in a cross-functional skunkworks team.
On Wed, Dec 9, 2015 at 1:58 PM, Brad Jorsch (Anomie) bjorsch@wikimedia.org wrote:
On Wed, Dec 9, 2015 at 3:40 PM, James Forrester jforrester@wikimedia.org wrote:
Short-cut answer to the title question: Me.
I'm glad to hear that you are accepting responsibility for OOjs UI development! Do you have a timeline on a fix for T113681, or a page that indicates what higher-priority development you and your team are working on in the near future?
If OOjs UI is the thing that we're supposed to be using in the future for our UI stuff, it's very concerning that further development is blocked on T113681
"Further development" is not blocked on this task. A few things that some people want to do are.
Let's not chop logic here. If "a few things that some people want to do" cannot be done due to T113681, then T113681 is indeed blocking some further development even if other further development isn't blocked. This email thread isn't even about T113681 specifically, it's about that there are no development resources for fixing things in OOjs UI unless someone is willing to do it as a skunkworks project, and OOjs UI isn't yet a finished product where we might be able to justify that.
I'm disappointed that you don't think that the fact that "some things people want to do" are blocked and no development resources are available to remedy the situation is cause for concern. When the situation was brought up in today's Scrum of Scrums, the consensus was that it is indeed concerning.
Please do not exaggerate for effect to try to get your way. I'm sorry that we disagree as to whether your patch belongs in the library in its current form.
Since you brought it up, let's look at my patch. There are two concrete blockers that have been raised on my patch. Neither of them actually have to do with the form of the patch itself.
The long-standing blocker has been disagreement over how the widget can be internationalized in the context of OOjs UI: The Language and translatewiki faction wants OOjs UI developers to integrate cldrjs, while the OOjs UI developers are unwilling to make any decision as to whether cldrjs is the way to go or translatewiki will just have to deal with providing translations for month and weekday names as they do for everything else. The closest we have to a decision is really a cop-out: "shove it into MediaWiki even though it doesn't belong there, because MediaWiki already happens to have most of the needed i18n strings and we can't make any decision here".
In last week's Scrum of Scrums, you brought up T113681 as a new blocker: OOjs UI is already too large, so we can't add new stuff until someone reworks it to be able to load individual components. MatmaRex then stated that no one owns or maintains OOjs UI to the extent that we can expect T113681 to be solved any time soon, which brought the lack of maintainership in OOjs UI into clear view.
MatmaRex also raised some other objections (disagreement with Design's design, non-use of moment.js despite moment.js not gaining us anything, doubt that anyone actually needs <input type="datetime"> despite evidence to the contrary), but no one else has agreed with those and he hasn't deigned to respond to attempts at further discussion in Gerrit.
If you have objections to the actual form of my patch, as opposed to lack of a willingness to make any decision on the i18n issue or any progress on the form of OOjs UI as a whole, you should raise them in Gerrit instead of continuing to sit on them. Although I wonder why you haven't done so already.
-- Brad Jorsch (Anomie) Senior Software Engineer Wikimedia Foundation _______________________________________________ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
On Thu, Dec 10, 2015 at 5:40 PM, Jon Robson jrobson@wikimedia.org wrote:
[OOjs UI ownership] might be a good discussion for the dev summit?
Perhaps. I'd say the idea in Phab. It's obviously way past the scheduled date for submissions[1], but the schedule isn't etched in stone, and it might make a good unconference session anyway.
Rob [1] And, admittedly, we're behind the schedule we planned on for putting the schedule together.
wikitech-l@lists.wikimedia.org