Incubator talk:Test-administrators

Project page?

Maybe it's better to move this to something like "Incubator:Test-sysops"? SPQRobin 16:33, 10 January 2008 (UTC)

Why not. --MF-W {a, b} 17:30, 15 January 2008 (UTC)

Done :-) SPQRobin 19:17, 15 January 2008 (UTC)

Do we stick to our own policy or not?

"You must have a user page on the Incubator and be a contributor to this wiki."

Either I'm colour blind, or three test-administrators do not have a user page. Are we going to pay a bit more attention to this or shall we simply omit that rule? --OWTB (talk) 22:31, 2 February 2016 (UTC)

Oh, a good point. I asked the three T-As now whether they want to create a user page. --MF-W {a, b} 00:44, 6 February 2016 (UTC)

Test wiki administrators who became Incubator administrators

IMO test wiki administrators who became Incubator administrators should (must) not be removed from the list (nor from the relevant user group, although the rights are redundant), because then it changes the reality.

For instance currently the info template on test wikis main pages lists the admins for given project. After removal it seems the wiki has no admin then. There are also further planned places which would use this information.

Presence on the list of test wiki admins also signalizes to other Incubator admins that admin requests for given wiki should be primarily and preferably be handled by test admins of that test wiki.

Also consider (low, but still) possible removal of Incubator administrator rights, where the test-wiki adminship should preserve, but in this situation it would have to go through the nominating process again, although originally the user might be still in the granted period.

Personally I don't see any harm in concurrence of the user groups (OTOH I see benefits described above) so unless someone will provide some significant reason against, I'm going to put it in the desired status quo.

— Danny B. 08:05, 11 January 2017 (UTC)

I was just thinking about this recently, which is why I ran a couple of tests at MediaWiki:Administrated test wikis.json recently. I placed my own name there, with expires = "Never", to see how it would look (at Wb/lad, for example).

I'm not sure we need to use concurrent user groups, and I'm very sure we don't need to add full admins to the .js script that adds those "please only deal with your own test" warning messages. But I think it's quite reasonable for administrators who are also involved in specific test projects here to add their names to the .json, so that they appear on the test project root pages as local administrators.

As far as whether test-admins should preferentially deal with their tests, that's theoretically true, of course. But part of how to do that is for tests to set up some infrastructure to facilitate that—infrastructure that would, beneficially, move to the subdomain when the test does. For example, if a prefixed {{Wx/xyz/Delete}} template placed items into Category:Wx/xyz/Maintenance:Delete, then test-admins could easily patrol those categories for deletion requests, and our central delete category wouldn't be cluttered with them. As a practical rule, I'm not prepared to allow central backlogs to grow just because test-admins don't patrol the central maintenance infrastructure regularly. StevenJ81 (talk) 14:54, 11 January 2017 (UTC)

I was not talking about anything like "please only deal with your own test". I was saying that it is preferable that test wikis are at first handled by their own test admins if they have any. Simply because they understand the language and the given test-wiki and therefore can more knowledgeably judge and deal with the situation. (I've experienced many situations when eager steward or global-admin deleted something because it has been marked to deletion although there was consideration of local admins in process or even on wiki idiscussion running somewhere. Similarly here, I remember cases from further past when testwiki pages have been manipulated using advanced rights though the case was not finished by the testwiki users yet.) Sure this does not apply to obvious vandalisms and so on.
Having separate Maintenance:Delete categories for wikis with their own test-admins was one of the ideas in planned steps when I started to work on the shared and reusable test-admins list. Although it does not need the local delete templates, the default global one can handle it itself. (Though having the local one concurrently does not harm anything, and as you mentioned, then it can be easily moved to the wiki when created.) The idea was, that it would stay in that local delete category for some time (week?) and if not handled by relevant test-admin then it would automatically be recategorized to the global Maintenance:Delete category and therefore might be handled by Incubator admins then (which clears your worries about growing log). (That's one of possible ways how to handle, there are also others of course.)
In fact on one hand we allow users to administer their projects via becoming test-admins, but on the other hand, in reality Incubator admins are mostly much faster and deal with things prior any test-admin even notices so test-admins do not have anything to do. My intention is to encourage test-admins to actively participate on maintaining of by-them-administrated projects, so they can gain experiences they can utilize when the project has its own wiki (or elsewhere of course).
Once again, the entire thing is about preferred process, not about setting up any formal bureaucratical process. So going back to the beginning for illustration, it would rather be message in form of "Hey, this wiki already has its own admins (<list>), you may like to leave it on them..." rather than "please only deal with your own test" you mentioned.
— Danny B. 12:45, 11 April 2017 (UTC)

I disagree. "It changes the reality"? What nonsense. The reality is that test-admins just have restricted access, while admins have access to the tools across all tests. What does it matter if someone is listed on the info page or not? We should better remove this information again as it pretends a wrong sense of ownership of tests by test-admins. Also, if an admin stops being an admin, he can easily request test-adminship again, if he needs it. It's not like requesting it is a very tedious process. --MF-W {a, b} 06:18, 12 January 2017 (UTC)

I appreciate the ownership issue being discussed here. And I would agree that we need to be very careful not to do anything to make it appear to test-admins as if they "own" their tests. It's a bad enough problem at small subdomain wikis; many tests here are even smaller, and the temptation to ownership would have to be even greater.

At the same time, having someone's name on the infopage also provides a contact point for any new contributor if s/he is looking for advice/​mentorship/​etc. for contributing to the test. It's reasonable, though not inevitable, for a sysop who was once a test-admin to remain involved in his/her tests. If so, that person remains an appropriate contact point for that test.

I would never insist that a sysop continue to list himself/herself on this page. In fact, best practice, if there are other test-admins around, would probably be not to do so. But in cases where the test has no other active test-admins, I don't see where there is any harm in listing an involved sysop here. StevenJ81 (talk) 19:36, 23 February 2017 (UTC)

The reality is that Incubator admins to test wikis are the same relation like stewards to single existing wikis. When you become steward, you are not automatically removed all admin/bureaucrat rights on all wikis where you have them, however, here you are. Furthermore the given test-wiki in fact still has its own admin therefore such user should be listed.
I think that running into the question of ownership is creating of artificial problem here as I do not remember any such issue in the entire history of the Incubator. Also, as pointed out above by StevenJ81, the list is more like primary contact point (assuming that test-admins are typically the most active people (and typically most tech experienced) in given test-wiki). Nevertheless, it is easy to do the wording the way it will explicitely ensure that the listed users are not owners and basic WMF principles apply here, however, per what I mentioned earlier, I am not convinced it is necessary, though I do not object it.
Again, the major issue I was pointing out is, that removing from the test-adminship group while the given user actually still can (and does) admin the given testwiki (because unlike for stewards we don't have the rule for Incubator admins to not intervene in their wikis) does not reflect the reality, because the user did not loose the "adminship-of-the-given-wiki" status (modulo expiration). Incubator admin and test-admin rights are independent on each other and therefore should be treated that way. To illustrate: If you have driving license for small motorcycles and then after some time you make it for cars as well, the marking of "you can drive small motorcycles" is not removed from the license, but stays there forever although "you can drive car" automatically implies "you can drive small motorcycle". Complementarily to that, when you first make the license for cars, you are concurrently given the license for small motorcycles as well. Also concurrent rights via different usergroups do not harm anything, on the other hand differrent usergroup membership indisputably helps to better orientation in the status quo.
— Danny B. 13:18, 11 April 2017 (UTC)

For the record, clearly having both the "sysop" and "test-administrator" bits flipped is redundant, and for the sake of transparency I wouldn't go there. However, I am fine if a sysop wants to continue to list himself/herself here on I:TA in his/her chosen tests, either with automatically updating one-year expiration dates or with an expiration date of "none". StevenJ81 (talk) 16:04, 13 April 2017 (UTC)

Looks like we obviously have different views of what transparency means. The current system is not transparent because first of all it does not mark wikis which have their own test-admins which is important in many aspects and has an influence on other existing or developped/planned tools. As I mentioned, it also does not match the reality - please read carefully the parable of relationships of stewards to projects and incubator admins to test wikis. Also, last but not least, redundancy doesn't imply lack of transparency, besides the rights are obviously redundant only partially.
Going further, are you going to remove the "Autoconfirmed user" user group from admins and test-admins, because it is "redundant"? Please mind, that the original user group serves as a fallback when you loose the "higher" group. I remember cases, when eager stewards removed some lower usergroup from the user who had the higher one. When the user lost the higher group, he didn't even have the lower one so couldn't perform the tasks he was doing before, although he had the mandate for that.
Redundancy in this particular case is not only not harmful at all, but guarantees the transparency and fallback.
— Danny B. 11:35, 14 April 2017 (UTC)

I don't see any rights that test-sysops have while sysops don't have it. --MF-W {a, b} 22:47, 14 April 2017 (UTC)

Actually the situation is very different from "stewards vs. local admins". Stewards are a "fallback" for when no local admins exist, so their policy doesn't allow them to do stuff when there are local users who can do it. But the policy on Incubator is different: Incubator admins can do everything everywhere, regardless of whether they are active editors in a particular test-wiki or not. Adminship replaces any previous test-adminship (which is simply a restricted and time-limited adminship). --MF-W {a, b} 22:47, 14 April 2017 (UTC)

My position falls between the position of User:MF-Warburg and that of User:Danny B.. I favor being able to list an administrator at I:Test-administrators (meaning on the .json), as Danny does, because it gives anyone contributing to those specific tests a point of contact that is close to the test and its specific rules and policies. I oppose assigning both rights flags (sysop and test-administrator) to a given individual simultaneously, for the reasons MF-W has stated above. StevenJ81 (talk) 18:39, 16 April 2017 (UTC)