Requests for comment/Core user preferences

An evaluation of user preferences (Special:Preferences) in MediaWiki core.

Request for comment (RFC)
Core user preferences
Component General
Creation date
Author(s) MZMcBride
Document status in discussion
See Phabricator.

New dataset

edit

See announcement and raw data. en.wp for now, others to follow. Please read the caveats before jumping to conclusions. Some initial thoughts:

  • I'd suggest using 1% as an aggressive cut-off for initial culling and then deciding whether we want to go beyond that. If a pref/feature is used by fewer than 1% of active users, and doesn't have accessibility implications, it's really hard to justify its continued existence.
  • Here are examples of features that could be ditched immediately:
    • The External Editor/Diff pref/feature. While the External Editor capability is also surfaced on file description pages, I think based on the tiny usage it's safe to remove it entirely. It was always a hackish approach (by yours truly no less), doesn't rely on the API, and should just die. Perhaps a better approach can be found in the long term, perhaps not.
    • Some of the watchlist filters. "Hide logged in users" (used by 33 users), "Hide anonymous editors" (used by 44 users), perhaps "Add pages I delete" (used by 151 users).
    • The ability to disable the Table of Contents feature (used by 86 users).
    • The ability to disable section editing (used by 86 users).
    • The ability to disable the collapsing of sidebar navigation menus in the Vector skin (used by 261 users).
    • The ability to disable AJAX search suggestions (used by 112 users).
    • The ability to render preview below the edit box (used by 372 users).
    • The ability to show diffs without rendering the corresponding page content (used by 402 users). (This one I'd like to see more cross-language stats for.)

--Eloquence (talk) 21:57, 12 October 2012 (UTC)[reply]

  • I'm not sure if this is the right place to post replies, but with regard to "Add pages I delete": 151 users out of 1,458 admin on en.wiki is over 10%. Also, please wait for some results from non-English Wikipedias and non-Wikipedias before jumping to conclusions about usage; not every one has the same workflows and user cases. Matma Rex (talk) 22:31, 12 October 2012 (UTC)[reply]
  • What is 1 % in absolute terms anyway? And shouldn't it be computed only on the total of users who ever set a preference, to avoid counting the "silent majority"?
  • Also, why are we considering only active users? It's useless data for several of those preferences, which can be used by read-only users who registered precisely to be set preferences (as many help pages suggest). --Nemo 06:26, 13 October 2012 (UTC)[reply]

Thank you very much for getting some stats. I think as long as we make a commitment to provide a snippet to replace functionality for any removed user preference (most of these should be trivial to re-implement in JS or CSS snippets), it's fine to kill most of the preferences you've listed with fire. --MZMcBride (talk) 03:28, 13 October 2012 (UTC)[reply]

How many users out of the 40 thousands who used "mark as minor by default" took advantage of the provided snippet? Workarounds can silence the most vocal opposers but don't cancel the feature loss (just a reminder). Indeed before jumping to conclusions we need to clarify how we can measure the net positive/loss of a removal...
As for the mentioned options:
  • cherry-picking watchlist options to remove would increase confusion about what is where (bugzilla:31882), better to fix bugzilla:31881;
  • TOC, collapsing, search and diffonly might affect readers, see above;
  • AJAX search: does it degrade gracefully on browser where it doesn't work?;
  • diffonly does have accessibility implications, because even when you're interested in a couple bytes only (the diff) you can be forced to render kB of HTML (the page), it can be a show-stopper for people low on CPU (or bandwidth?); rollbackers (including sysops) and RC patrollers in general are probably a more correct population to consider (alongside the others?); finally, does it make sense to remove this and not its twin "Omit diff after performing a rollback"? --Nemo 06:26, 13 October 2012 (UTC)[reply]
I'm using the diffonly preference since I'm a sysop on an active wiki and I have to review *lots* of diffs everyday, using a relatively modest computer with no much RAM and CPU clock frequency. I developed a script which adds the diffonly=0/diffonly=1 to diffs links when I need them, but I almost never enable the display content after diff. Also, I think that this particular preference is targetted to specific roles of users: sysops and patrollers, which are a very minimal part of the average users, since normal editors or readers doesn't review diffs and recent changes at all. This should also be considered when removing such preferences. --Ciencia Al Poder (talk) 11:16, 14 October 2012 (UTC)[reply]
It's also useful for people who often view diffs from their watchlist like myself. I've used it since it was introduced in January 2007 and I'm rather surprised it isn't used more widely. I use a screen reader and therefore can't really use navigation popups. Graham87 (talk) 11:38, 17 October 2012 (UTC)[reply]
Low usage isn't an argument for getting rid of a preference if there is little cost to maintaining it. It is an argument for getting rid of a preference which is expensive to maintain, makes implementing other more desired features very difficult etc. Kerry Raymond (talk) 01:40, 15 October 2012 (UTC)[reply]
Low usage of a preference is exactly a reason to get rid of it. Almost all preferences could be considered low-cost (eg: it's just one if statement), but every preference increases the complexity of the software and number of possible outcomes. It's why we're very wary to add new preferences these days (and many existing preferences wouldn't be added now, if they were being newly proposed). ^demon (talk) 16:52, 15 October 2012 (UTC)[reply]
FYI, I am one of the people who set 'The ability to render preview below the edit box (used by 372 users).' and am surprised by the stat. If the edit box is at the top of the page, after checking your preview you can just hit HOME to get to the top of the page, where the whole edit box is visible. If it is at the end of the page, hitting END takes you to the bottom of the page which leaves half the edit box cut off, due to the amount of cruft that sits underneath it. Therefore having it at the end is a lot more fiddly than having it at the start if you're not on a large desktop monitor. I wonder what the stat for this setting would be if the default value had been the opposite - I wouldn't be surprised if it was pretty similar (i.e. the inverse) and that most people haven't really considered changing it from the default, and wouldn't have done so no matter what the default was. --HappyDog (talk) 15:25, 15 October 2012 (UTC)[reply]
^demon, I agree that low usage is a reason to remove, but there are also other reasons to keep some of them. For example, there are probably A LOT of people that never changed their preferences, or did it only once to set the language or timestamp, and simply didn't care about the other preferences, because I agree there are lots of them and people probably don't know what effect have some of them. Wikipedia has millions of readers, but only a very small percentage edited. Should it be a reason to remove the editing interface because it's "low usage"? It may seem a bit extreme, isn't it? The key is "low usage" in comparison to what? I'd be glad if you could weight (and discuss here) the real usefulness of every option before trying to remove them with the generic "low usage" reason. Thanks. --Ciencia Al Poder (talk) 19:37, 17 October 2012 (UTC)[reply]


I personally have used two of the above, and would like to see them kept.
When manually doing repetitive tasks, especially on a slower connection, I find "The ability to show diffs without rendering the corresponding page content" very useful. It's not something that I need all the time, though, so I can understand why it would have a lower apparent usage.
"Enable section editing via [edit] links" - I always have this box unchecked. I never use the section editing feature. I understand how it can be useful to some, but, especially on talk pages (and let's not talk about edit conflicts) I've found it confusingly problematic. And this preference removes extra links from needing to be loaded on the page.
I think out of all the things eloquence notes above, the filter by IP definitely should go. It's contrary to "anyone can edit" and "We're all Wikipedians". I can almost see a potential usage for when someone is vandal patrolling, but I think the ability to filter out/in both IPs AND non-autoconfirmed together would be more helpful, I would think? - Jc37 (talk) 00:46, 18 October 2012 (UTC)[reply]
I've sometimes used the watchlist filter. "Hide logged in users" but I'm not counted as one of those who does so because it isn't how my filters are currently set. It is a useful way to target possible vandalism, and it doesn't breach the "anyone an edit" ethos because it doesn't stop IPs editing, just makes it easier to identify and revert vandalism. WereSpielChequers (talk) 11:45, 17 February 2014 (UTC)[reply]
WereSpielChequers, you sit there and assume that unregistered users are (mostly) vandals. This is often true; however, such attitude is harmful to such users themselves — specifically those who tried to do something, messed up the formatting or simply didn't convey their idea, and got it reverted instead of your attention to detail. —Gryllida 20:34, 17 February 2014 (UTC)[reply]
Jc37, “vandal patrolling” is a barbarous concept. When not focused on a topic, you end up engaging in routine work with little interaction with the affected contributors and article content, with issues like this and that. (I'm ok with disabling 'hide logged in users' in the watchlist.) —Gryllida 20:40, 17 February 2014 (UTC)[reply]
Those filters can be switched on/off in the watchlist itself, no need to have a preference for them. That selection could be remembered, though --Ciencia Al Poder (talk) 10:20, 18 February 2014 (UTC)[reply]

I'm only ok with disabling “render preview below the edit box”. That's something I tend to use, personally, although AJAX preview makes it useable regardless. —Gryllida 20:40, 17 February 2014 (UTC)[reply]

I would perhaps agree with the suggestions to keep the ability to show diffs without rendering the corresponding page content. Some pages are quite large with me only needing a specific diff. —Gryllida 20:41, 17 February 2014 (UTC)[reply]

Generally

edit

Tabs on top still a good idea? Kind of stack weirdly. Maybe need vertical tabs in a sidebar thing?

Putting the tabs in a sidebar would be problematic on narrower screens (which are the ones having the stacking problem in the first place), since that way the preferences themselves could get pretty squished. Dealing with squished stuff is no fun. None at all. And either way there should really just be LESS TABS. We could totally kill about seven of them with no great loss... merge relevant stuff, get rid of pointless stuff, and profit. -— Isarra 19:49, 23 September 2012 (UTC)[reply]
I think we should switch from tabs to collapsible sections making use of <details> native collapsing code. As a bonus the <summary> could include some text describing what kind of preferences are inside the section. Daniel Friesen (Dantman) (talk) 20:21, 23 September 2012 (UTC)[reply]
Tabs, whether at the top or side, are a much more intuitive structure for navigation; collapsing things is more like to just keep people from seeing them entirely. Might be worth looking into for extended descriptions or the like, however, for anything that may need one. -— Isarra 20:29, 23 September 2012 (UTC)[reply]
It actually looks pretty good using <details> even without any nice vector styling added to it: http://i.imgur.com/Mb4L9.png
And if you wanted it wouldn't be hard to use JS to implement a sort of accordion like collapse functionality that would have most of the pros that tabs have. Daniel Friesen (Dantman) (talk) 21:58, 23 September 2012 (UTC)[reply]

Stats about usage of these preferences is desperately needed.

See w:Wikipedia:Database_reports/User_preferences, but it would be better to have an aggregate of the usage over all wikis. Helder 21:09, 23 September 2012 (UTC)
This doesn't include metrics for most of the prefs discussed below, is anyone working on a complete report? Rather than just going by gut feel, we should rank-order prefs and start with removing the least frequently used one that have minimal justification to exist.--Eloquence (talk) 23:44, 8 October 2012 (UTC)[reply]
Replied to you here. Say, you have shell access, don't you... >:-) --MZMcBride (talk) 00:52, 9 October 2012 (UTC)[reply]

User profile

edit
  • Username — no (change) link
  • Number of edits (link to Special:Contributions?)
  • Language option — move up to top?
  • E-mail preferences are weirdly situated... combine into a generic "Communications" section?

Appearance

edit

Skin

edit
  • Skins should surely be using a dropdown menu?
    • No, no, no, no, no, no, no, hell no... that would eliminate the preview link, the other links there to help the user, and it would preclude us from trying to make the interface more usable in the future by using things like screenshots.
      • What? You could load additional content (such as a screenshot or links to power-user features [custom JS or CSS pages]) when the drop-down selection changes, as relevant and appropriate. --MZMcBride (talk) 21:41, 23 September 2012 (UTC)[reply]
      • Doing a dropdown like that would still kill the usability. Screenshots should be scannable visually. If a user has to individually each skin from a dropdown just to see if it's the skin they want then it's a usability failure. Additionally a user should not be required to take an action that implies changing of preferences (changing a dropdown) just to preview a theme. WordPress, Drupal, and Joomla ALL show a screen listing themes with a name, screenshot, and a short description. That is the standard most user-friendly way to do skin preferences. A dropdown is a complete step backwards away from user-friendly preferences (which is supposed to be what we're trying to aim for here in the first place). Daniel Friesen (Dantman) (talk) 21:54, 23 September 2012 (UTC)[reply]
  • Eliminate some skins?
  • Fix skinning system generally...

YES. -— Isarra 19:50, 23 September 2012 (UTC)[reply]

Maybe look for a way to combine some skins? - Jc37 (talk) 01:17, 18 October 2012 (UTC)[reply]

Files

edit

Weird section.

Again, useful, depending my connection and/or computer. - Jc37 (talk) 01:17, 18 October 2012 (UTC)[reply]

Advanced options

edit
  • Threshold for stub formatting — gadgetize?
    • Does anyone even use this? Since most folks aren't even going to have the slightest idea what that even means (I sure didn't), it probably would be better in a gadget for the subset of folks who do. -— Isarra 19:57, 23 September 2012 (UTC)[reply]
    • It's a preference that adds a stub class to internal links pointing to pages below a certain size. Basically it makes internal links pointing to pages that exist look almost like redlinks if the page doesn't have much content (ie: it's a stub an editor would want to edit). This works at the parser level (it has to fragment the cache to work) and it depends on information locked away inside the database. You can't turn it into a client-side gadget. Trying to find a way to make it work without fragmenting the cache is more important than deciding what we're going to do with the user preference. Daniel Friesen (Dantman) (talk) 21:06, 23 September 2012 (UTC)[reply]
      Appropriate api calls should be more than sufficient for a js gadget to determine that information, and without fragmenting the cache. -— Isarra 21:28, 23 September 2012 (UTC)[reply]
      Note: stub formatting (last I checked) does not fragment the cache. It disables it totally! As far as js - That's quite a lot of api calls to do on the client side... 129.173.208.184 17:20, 24 September 2012 (UTC)[reply]
      That is (number_of_unique_pages_linked)/500 calls. Which is 1 call for most pages. There's a gadget on pl.wikipedia that does this for disambig links: [1]. Matma Rex (talk) 16:37, 29 September 2012 (UTC)[reply]
      Another option is to output on each link an HTML5 -data- attribute with the page size (would it require $wgHtml5 = true?). That would make it compatible with gadgets and don't interfere with cache. Note: I'm using this preference option on some wikis, with a value of "2000" and it's very useful for me. --Ciencia Al Poder (talk) 16:39, 29 September 2012 (UTC)[reply]
  • gerrit:27202 Disable browser page caching — why does this option exist?
    • Some browsers are dumb. Might work well as a gadget for those, though, but there really won't be very many people who have this problem and know what it is. -— Isarra 19:57, 23 September 2012 (UTC)[reply]
      This preference controls HTTP level caching headers. The preference exists because of users who need their cache cleared regularly (editing site/user scripts/styles?). You can debate the idealistic topic of this pref vs. everyone using proper dev tools. But you can't gadgetize something like this. Daniel Friesen (Dantman) (talk) 21:06, 23 September 2012 (UTC)[reply]
      Sure you can; just change the links the people click on. It could also be done by instead toggling an invisible user preference to do the same thing, of course, if you don't like the whole silly workaround thing. -— Isarra 21:28, 23 September 2012 (UTC)[reply]
      I think this preference is misleading. Most people assume it refers to parser cache. People who actually understand what this does should know enough about their computers to make their browsers not send caching headers. 129.173.208.184 17:20, 24 September 2012 (UTC)[reply]
  • Auto-number headings — gadgetize?
    • Be better if it had a css way to do it, but it's not something terribly likely to screw anything up if it is done in js. -— Isarra 19:57, 23 September 2012 (UTC)[reply]
      Actually, I forgot we can now use preference flags without it actually showing up in the user preferences. Having this as an invisible option in core but using a gadget to toggle it might be a good way to make it work, especially if we're shipping gadgets with core... -— Isarra 20:07, 23 September 2012 (UTC)[reply]
      In that case I don't understand the difference between it being a preference and it being a gadget. 129.173.208.184 17:20, 24 September 2012 (UTC)[reply]
      • There is actually a potential way do do this via CSS, since gerrit:9980. We would need to change mediawiki defaults by setting the preference as enabled wiki-wide through $wgDefaultUserOptions['numberheadings'] = 1 in LocalSettings.php (or alternatively, getting rid of that option altogether, and making the code always produce the auto-numbering), and adding .mw-headline-number { display: none; } to the default skin css files. Then, the numbering will be present by default, but hidden; users could enable it by setting .mw-headline-number { display: inline; } to their user style. This is sort of documented at Manual:Table of contents. --Waldir (talk) 17:11, 9 January 2013 (UTC)[reply]
  • gerrit:30173 External editor and External Diff
    • Very very confusing if accidentally clicked, most people who use external editor (all 5 of them I'm sure) don't use it via the preference. user:Bawolff 17:20, 24 September 2012 (UTC)[reply]
    • This feature is actually pre-api. It's entire existence relies on an application interpreting a special mime-type, screen scraping the login page, screen scraping the edit page, and making an edit without using the api. The application still needs to do all the work of downloading and saving the file. And it typically needs to ask the user for their password. So since we want applications to stop screen scraping and asking passwords. And when we have something like OAuth you'll be able to build an application that does external editing without using this feature. I'm in favour of removing this feature from MediaWiki entirely. Daniel Friesen (Dantman) (talk) 17:42, 24 September 2012 (UTC)[reply]
    • [above was me] There's still people who use it (especially for editing images, but also for editing pages). There's no reason the feature cannot be used with the api, or be integrated with oAuth. Its really just a mechanism to trigger an external program. What the external program does is up to that program. However its almost useless to be enabled by a preference because most people don't want to always edit with an external editor [and if they did, in can probably be gadgetized by messing with the links in the edit tab via js]. Bawolff (talk) 18:47, 26 September 2012 (UTC)[reply]

Math

edit

See also Requests for comment/Reduce math rendering preferences and bugzilla:36496.

  • Eliminate math preferences? gadgetize?
  • Now in a MediaWiki extension... hmm
  • The preferences for this are provided due to inconsistent browser support, from the looks of it. Ideally the extension should be smart enough to figure that out itself without ever bothering users with the details, but meantime turning the preference into a purely js-based thing as such a gadget would need to be could prove problematic for the same reason this is there in the first place. -— Isarra 19:58, 23 September 2012 (UTC)[reply]

Date and time

edit
  • Can't all of the date and time formatting be done in JS?
    Is there any particular reason why the servers can't handle this? Losing this to a sea of gadgets would also be unwise given the international nature of many mediawiki-based projects. -— Isarra 19:59, 23 September 2012 (UTC)[reply]
    If implemented right there is no cache breakage to worry about from this preference so it's not a burden. Browsers also do not always have the timezome the user using them wants. And none of them expose the user's preferred date format. So this cannot be done without a server side preference. This one absolutely should stay. Daniel Friesen (Dantman) (talk) 20:51, 23 September 2012 (UTC)[reply]

Editing

edit
  • gerrit:27257 Size of editing window has no preview option
    • Yeah, what the crap does this row/coll nonsense even mean? -— Isarra 20:16, 23 September 2012 (UTC)[reply]
      They're the value of rows="" and cols="" in the <textarea> on the edit page. They end up controlling the size. cols="" doesn't have much effect in our modern skins because we use width: 100%; though it does affect some legacy skins, in fact there was a bug in that area recently. rows="" does have the effect of making the textarea bigger/smaller. However after we exterminate legacy skins I'm open to eliminating the preferences, hardcoding the rows/cols used by the textarea, and telling anyone obsessed with their size to just use their user-css to resize it. Daniel Friesen (Dantman) (talk) 20:45, 23 September 2012 (UTC)[reply]
      That was a rhetorical question. It is completely unreasonable to expect our users to know this. I'm not sure what 'exterminating legacy skins' has to do with it, but the size of the text area either needs to be adjustable somehow, or would need intelligent determination according to screen size simply because there just plain is no one-size-fits-all solution with this. -— Isarra 21:15, 23 September 2012 (UTC)[reply]
      This is one of the few features I frequently toggle, depending on what I am doing and where I am working. Having it further hidden would make it considerably more difficult, because ppeople wouldn't know about it. More generally, when people talk about doing things is css, they should realize that not everyone who needs a modification is comfortable with that--perhaps we should have a css preferences page for commonly used changes. 65.88.88.208 18:22, 27 September 2012 (UTC)[reply]
      Can you explain further what you do with it? Why is the default size not good enough for you? Doesn't your browser support a native resize handle for textareas? Or is there another reason that native resize isn't good enough for you? Would that issue be fixed if we added some JS that will make it so that when you resize the textarea the height will be stored in localStorage and re-applied when you next visit the edit page? Daniel Friesen (Dantman) (talk) 19:09, 27 September 2012 (UTC)[reply]
      I use the feature, changing my default size rather frequently, presetting the size according to what I will be doing and what computer I am working from. I know there are other ways or doing this, but I find this simplest. DGG (talk) 03:56, 26 October 2012 (UTC)[reply]
  • gerrit:27197 Mark all edits minor by default — gadgetize?
    This has already been disabled on some projects. Pulling the preference out of core entirely would probably be wise for the same reason they did that; anyone who does need it could probably use a gadget. -— Isarra 20:16, 23 September 2012 (UTC)[reply]
    right! better to remove entirely--has very unfortunate consequences for anyone trying to maintain quality. Some wikignomes may need it, so it should remain as a gadget, with a warning.65.88.88.208 18:22, 27 September 2012 (UTC)[reply]
    Why not make this a user-right instead? Other user-rights related to minor edits: minoredit and nominornewtalk. It would be useful for the bot group, for example, and anyone using automated tools specifically for minor edits. - Jc37 (talk) 01:52, 18 October 2012 (UTC)[reply]

Beta features

edit

The WikiEditor extension adds this section and its two prefs. English Wikipedia names this section "Usability features" (with message prefs-beta).

Why does this section still exist? (Related bug: 28555)

Recent changes

edit

Hmmm.

  • bugzilla:35768 Current positioning of the "Enhanced watchlist" preference is confusing. Helder 22:12, 23 September 2012 (UTC)

Watchlist

edit

Hmmm.

Not possible, per bugzilla:35768#c1. Helder 23:36, 25 September 2012 (UTC)

Search options

edit

Hmmm. Another per page results option? Hmmm.

  • This and the two above it could probably be merged, as they're all pretty similar stuff to the point of overlapping in places. -— Isarra 20:17, 23 September 2012 (UTC)[reply]
    I agree on the first two. I think the 2 entries in the "display options" section of recent changes and watchlist, could probably be combined. Both of these are directly reliant on how active a particular wiki is. Search is a little different in usage though. - Jc37 (talk) 02:05, 18 October 2012 (UTC)[reply]

Misc

edit

Why does this tab exist?

WikiLove should be a gadget, surely.

Anyone know why it was implemented as an extension in the first place? Is that reason still relevant? -— Isarra 20:18, 23 September 2012 (UTC)[reply]
Maybe one reason is so other wikis do not need to create lots of forks of the main code? For this, the Gadgets 2.0 will be of some help. Helder 22:12, 23 September 2012 (UTC)

Extensions

edit

Gadgets

edit

Fine.

Threaded discussion

edit

LQT wtf bbq.

Contests

edit
  • Show a link to My Contests in the user menu.

A whole tab for this? Seems disproportionate and excessive. Really outside the scope of this RFC, though.

Edit review

edit

FlaggedRevs wtf bbq.

  • At least the flagged revisions tab has a name that makes sense. Pending changes has a tab literally called 'Pending changes', which is not going to mean anything to most people. And these both could probably be merged into the editing section, since they're directly related to editing. -— Isarra 20:19, 23 September 2012 (UTC)[reply]

General comments

edit

Krinkle should make that magical searchable gadget repository thing he was talking about. Unless that was someone else. But whoever it was... well, that would be useful. A lot of the more random preferences would go well dumped in a searchable repository, along with a whole slew of rather bizarre user scripts. And some not even remotely bizarre ones, for that matter.

Also, we should be careful when considering gadgetising commonly used things and accessibility features, if there are any. I'd also mention things added for the sake of compatibility, but chances are most people will have no idea what they mean and thus having an extra tick box with the main stuf won't actually help anything for them.

Also also, while gadgets normally function as snippets of js and css, would it not also be possible to have some that exist simply as a piece of js that toggles a non-visible preference flag? Or... something? -— Isarra 20:24, 23 September 2012 (UTC)[reply]

I believe that the "searchable gadget repository thing" will only be available on Gadgets 3.0. Helder 22:12, 23 September 2012 (UTC)

Mark some user preferences as "Advanced" instead of removing them

edit
See also bugzilla:40346#c12

We're talking about removing user preferences, but what about "hiding" them by default? Some programs already follow the convention of having an option to "turn on advanced options", so advanced or obscure options are hidden by default to newbies who still may not understand what they actually do, and advanced users can enable them.

So, if I didn't miss anything, the rationale for the removal of user preference is to not overwhelm the user with so many options by one side, and probably delete some of them so they don't need to be taken into account when making changes to the code. The first part can be done by just hiding them, and yet allow advanced users to use them. Then we could argue which of them are useful at all and may be removed.

And I'm talking about some of them that probably only makes sense to a specific role of users, like sysops or patrollers, one of them being the "diffonly" preference (see my personal experience on #New dataset). --Ciencia Al Poder (talk) 11:28, 14 October 2012 (UTC)[reply]

Just for clarification, my intention is not to move them to another tab (like bugzilla:40346#c12 suggests). They may still remain on the actual tab as they are now, but hidden. That way they're still organized for people that want to see them. The idea is to add an additional boolean option "show advanced options", unmarked by default, that when marked would render them as they are now. Each option could be marked as "advanced" using a variable like $wgHiddenPrefs --Ciencia Al Poder (talk) 12:30, 14 October 2012 (UTC)[reply]
Strong support. "Show Advanced Options" is a widespread interface option. w:Progressive disclosure.
Because: We need Photoshop for powerusers, as well as MS Paint for the newcomers and casual editors.
  • MS Paint is great! It's Welcoming, and easy to learn via experimenting, and easy to create simple (sometimes even complex) projects in!
  • Photoshop is great! A dense abundance of menus, with a profusion of tiny and detailed-metadata, for those who need it! For those who spend hours, every day, for many years, working *hard* within it.
We want and need both. Progressive disclosure is the sanest way to do that. –Quiddity (talk) 01:34, 11 March 2014 (UTC)[reply]
edit
  • phab:T42346 – Way to bring back user preferences that were removed from core (e.g. convert some preferences into JavaScript gadgets)