Subject says it all. In truth, lots of edits I ignore.
Talk:Edit Review Improvements/New filters for edit review
About this board
Leave your thoughts and ideas here about the New Filters for Edit Review filtering tools and interface.
These are now standard on Recent Changes and the Watchlist. What works well? What could work better?
Leave feedback in any language.
How to provide feedback
- do you have that bug when you are not logged-in?
- explain how to reproduce the bug (step by step)
- tell us what is your configuration (browser version, scripts you use...)
- say on what page it is happening (Recent Changes, Watchlist...)
Also see the FAQ.
|Edit Review Improvements (ERI)|
Nice idea. Takes a little time, may take a few tweaks.
Thanks! If you need any information about the filters, please tell us.
Whoa whoa whoa
First off, now I can't even start a new talk page section without massive confusion!
Anyway. Where's the frequently asked question "HOW DO I MAKE IT STOP"
Even if the answer is only as below, you should include it. In fact, you should never have gone live without including it in the first place!
Q. How do I return to the old watchpage? A. You can't. We've decided what's best for you. Have a nice day.
When you had first access the watchlist, you had that pop-up offering you to opt-out. If not, I'm sorry about the inconvenience, but you may have a script or a gadget conflicting with that notice (and probably other stable tools).
In any case, there is an option in your preferences, Matěj gave you the link.
I went here immediately, since the optout popup is hidden behind the first.
In the future, please consider adding a "Get me outta here/Bring everything back as it was" button already to the first popup.
Btw, how do I get back to the regular talk pages? (Suspecting each wiki has its own setting; please add "luddite" option to globally make any and all new developments opt-in) CapnZapp (talk) 10:42, 17 July 2018 (UTC)
Have the opt-ing out button on the first pop-up is definitely a lesson learnt.
Concerning the talk pages, all talk pages are using Structured discussions on mediaiwiki.org. You can't opt them out. (And you don't need to sign anymore.)
The amount of white space is blinding. I realize this is not the place, but please tell the "structured" devs to add a "condense" option so less screen estate is wasted. Please don't "beginnerify" Wikipedia (or rather, do, but for new users)
Having to sign is obviously something I can live without, though.
Have a more dense version of structured discussions has already been asked. But for now the development is stalled.
Sorry that is unacceptable - I call that hit and run development. Revert back the introduction of new features if the programmers are unavailable right after release to polish their new stuff properly, that should lit a fire under their collective arses.
Testing - does this create an indent?
What the frak - if discussions are now linear only with no indentation levels, all hope is lost.
Please direct me to the proper feedback/discuss page of "structured" talk pages, so I can give my mind to those responsible: leaving development in this shape is unacceptable.
You can go to Help talk:Structured Discussions to share your feedback. On your way, I advise you to make a stop to the introduction page for Structured discussions, where you will learn more about indentations among other things.
"learn more about indentations" - why not simply say "indentation is not supported"? That page says "you don't need to indent anylonger" as if that's a good thing?! No wonder English Wikipedia rejected this mess. Luckily I'm not a frequent contributor to mediawiki, so I leave it to others to battle down this misguided and overwrought mess. The simple truth is that the Flow Project should have been shut down entirely; its team members dispersed over many other projects so their collective will is broken. It's not unheard of for smaller communities like this one to be usurped by a fanatical core of devs who ignore common sense arguments, and I will certainly not waste anymore effort here. For your sakes, I hope the flow programmers are ousted sooner than later. Signing off (because I want to), yours truly CapnZapp (talk) 09:44, 18 July 2018 (UTC)
I hope that next time the discussion would be more open and pleasant.
And I hope the next time wiki isn't afflicted by hit and run programming. Implementing something and then immediately "stalling" development (without a clear opt-out path) is like a punch to the face, and deserves all the unpleasant feedback it can get. Have a nice day. CapnZapp (talk) 15:44, 4 December 2018 (UTC)
We still look for people who could help to announce new features and improvements.
Is there statistics of users opting-out the newer watchlist UI?
As of August 6th, out of 120,258 active users on English Wikipedia at that time, 1,101 users have opted out of the watchlist (0.91%) and 630 have opted out of RC (0,05%).
Now that I see growing number of less satisfied users and no (if not very little) positive feedback, I would like to see the statistics of users opting the newer feature out. Just curious.
I can ask for it.
Also, I see some positive feedback, and don't forget that happy users tend not to say that they are happy. :)
May you please provide links to the positive feedback? Thanks.
Having been initially sceptical, I really like the new watchlist. Just discovered the coloured filters and that's really useful. I also like that I can filter out all the bot edits that I rarely care about. The interface seems to have improved too, or maybe I'm just used to it now.
You have access to the same feedback page I have, where you can easily find some positive comments. However, we are not counting points here: only look for people who are complaining about the filters versus the one who aren't is not really meaningful. Let's wait for the statistics and compare them to the number of active people on the wikis.
It may depend on whether it is used on RecentChanges or Watchlist.
Statistics are not available yet on Gafana or query-able. I've asked a developer to find those statistics. We will see if it is possible to have them visible publicly.
As of yesterday, on English Wikipedia, 1101 users have opted out of the watchlist (0.91%); 630 for RC (0,05%). To be compared to the 120,258 active users at that time.
Can we see how many active users use the Watchlist regularly, and how many of those use filters? Total number of active users is not relevant, as a significant percentage might not use the Watchlist or filters.
Hmm.... Can I ask for next month's stats? I can see numbers growing (steadily?) but not as big compared to 32,000+ enwiki users (part of 70,000+ total) enabling the feature before graduation out of beta.
If you remind me about it next month, yes.
When the release on watchlist happen, only 32,000+ were using that feature (because it was already default on RCs). Now, that's more 199,000+ users using it on watchlist. That's a lot.
One month later. Umm... Must I ask every month? I would like to have a website dedicated to statistics tracking this.
As of yesterday, for English Wikipedia, Recent changes: 691; Watchlist: 1368, Active users: 123,774.
What do you expect from those stats?
I don't know why the task to remove the opt-out option of RC is created and still intact rather than declined.
The numbers to opt-out the new RC UI are steadily increasing, indicating that those registered users would prefer the older RC interface. I bet the stats increased in three months.
EDIT: I wonder whether proposing to keep the old RC would be within the scope of the next year's community wishlist.
I don't see a reply somewhere. Is there a link, or did you just email? Thanks.
Sugestion: ignore precise bot/people
It might be useful to ignore a single bot (or more) and not all.
For example, MsnBot makes a lot of minor changes and my list is full of these changes, while they are not risky, so it's harder to track changes from other bots.
Ignore people would be done through a feature that would allow people to watch what people do. this feature has been abandoned to avoid harassement.
Maybe only apply it to bot-flagged accounts? What do you think?
Yes, for me it's only useful for bots. I want to follow most of them, but some make a lot of changes and I can't check them anyway.
So, applying this type of feature only to bot-flagged account will be perfect, and this will avoid the harassment of people accounts.
Another useful way to reduce the visual number of results with attributes (authors, bots, categories, dates, tags, etc) not of concern would be to let users group results by such attribute(s) and then click to expand/collapse desired groups, and also specify such as part of saved filters and/or queries, as needed.
@Wikicat, I'm not sure to understand. You mean creating batches of edits from a given group, and then expand them?
The current system could do that, role by role (or by combining roles) if we create filters for those groups. There is a ticket about that, but declined with no explanations. I've reopened it.
@Trizek: Just a hierarchical sorting tree with collapsible branches, like a file-directory; and depending on which and what order attributes are selected each time, for example each "author" folder might contain all "months" for that author, and each "month" all "categories" that author edited that month, and each of those the pages; or each "month" might contain all "authors" for that month, and each "author" all their editing "days". Sorting first by "author", or "bot", or whatever, and not expanding selected top-level folders would hide all their activity in the results. This could also be a multi-sortable table with collapsible row-groups. Further, to save strain on the server, could dump top-level-sorted blocks of records to browser and do the remaining sorting in a wikitable. Also, this sorting-tool could apply to any generated list of pages.
It looks like a new tool, not an improvement of the existing filters. I'm afraid that's out of the scope of the small improvements that can be made through the maintenance process.
"available on all wikis" or "on certain Wikimedia wikis only"?
There's a hatnote saying that "All features below are now standard on the Recent Changes page and on Watchlist and are available on all wikis", but many features start with a limiting comment such as "Available on certain Wikimedia wikis only". Which is true?
All features under "New filters for edit review" banner are now default if you install the lastest version of Mediawiki and default on all Wikimedia wikis. Both are available immediately if you create an account, it is possible to opt-out.
This is true on the Recent Changes and Recent Changes Linked pages and on Watchlist.
Can you change the inaccurate sentences you've found?
How to filter (out) AWB edits
Work on filtering our/excluding tags is scheduled for April 2019.
Hello, On frwikisource, some contributors use AWB for maintenance tasks, without bot flag. These edits, done by hundreds in a row, generally flood the "recent changes" : they indeed include a specific summary, mentioning AWB -> see https://fr.wikisource.org/wiki/Sp%C3%A9cial:Modifications_r%C3%A9centes?hidebots=1&hidecategorization=1&limit=500&days=30&enhanced=1&urlversion=2
How could they be filtered out ?
Bonjour! Do you have a local filter for AWB on your wiki? A previous discussion has happen about this topic, you may find there some information.
ok. I just added a filter (hope I did it right here ^^)
Now, how can I filter the tag out? ...
I only have the option to display only the tag, but not to display anything but the tag...
Filtering tags out is not possible (et c'est bien dommage). You have to select all tags but the one you cant to exclude. :/
There is a ticket about that and some plans to work on it starting April 2019.
ok, so, having a filter for AWB is no use at all for this :(
ok, so I'll have to find something else...
If you've selected all filters but AWB, you can save that filter set. Sorry, that's the best I can offer. :/
yes, indeed : but all our filters are designed to filter potentially problematic edits... many edits I want to see are not tagged because they are not problematic, so...
-> just did what you propose... and got absolutely no answer - that's a good sign : it means we do not have a lot of really problematic contributors ^^
but that is not what I was hoping for... just wanted all edits except those AWB that are made by one of our admins...
well, we'll have to wait for a solution from our dear devs ;p
Thanks for your efforts...
And thank you for your patience!
Accessibility concerns on mobile devices
Virtually unusable on a mobile device. Screen jumping all over the place.
Switching to the mobile interface will display the old display, without the filters. If you force purposefully the desktop view on mobile, you will get the filters, which are, indeed, not mobile compatible.
I wouldn't suppose the developers wanted to punish the people with old and slow computers. I would suppose they would have missed this use case (it's really hard to develop software that works out-of-the-box for everyone!) . It might be of help to them if you could provide more information about your configuration in which you face this issue: OS, browser (with versions). They might use it to see if they could do something about it.
A temporary solution would be to switch to the old interface if that worked well for you. You could do that via
What do you mean by "jump all over the place"?
I mean that the page loads, and then further additional things load (like extra tabs), and as those things appear, the page "reflows" (like adding words in the middle of a paragraph) so that things I am trying to click on might continue to move to new positions. It would be better if the page didn't appear at all until everything was in place -- since having things jump away from under the mouse pointer is very frustrating.
I think I can visualize in my head what you mean, but a recording of your screen would help a lot.
The filtering tick boxes are confusing
In most software, a tick box either selects or deselects a particular option: simple. So you'd expect a group with no boxes ticked to be equivalent to selecting none of the options, and you tick one or more of them to add the options you want. In fact, no ticks seems to be the same all all-ticks: ticking a single box in an unticked group surprisingly and without notice deselects all the options other than the one selected.
Now it's true that this is documented half way down the long documentation page, but it should be clear on the face of the controls themselves.
I've also felt this behaviour a little odd.
What would you suggest? :)
Checking the appropriate check boxes by default accompanied by 'Select all' and 'Deselect all' options.
For example, by default results from all Namespaces are shown but none of the boxes are checked. Checking the box for a namespace results in results being only shown for that namespace. In this case, the suggested change would be as follows. Check boxes for all the namespaces should be checked by default and the user could use a combination of 'Select all', 'Deselect all', checking some namespaces to achieve their desired output.
Unfortunately, this seems to result in some added UI elements which might add up to page clutter (the namespace drop down might become over crowded, for example). I don't know how to avoid this though :-)
To wszystko działa za wolno. Zarówno wrzucanie zdjęć jak i przeglądanie zmian.
Sorry to use English.
What is the problem with the filters, precisely?
Does not make sense
In the old UI was a simple tick box for edit types, like say bot edits. Could temporary switch off and rerun list. How does that work now?
Well starting to get it but works opposite to what was thinking. Need to filter on what you want to see not filter out what you do not want to see.
That's exactly it: you ask the filters to display what you want to get.