The Visual Editor is a huge failure

[Kelly Martin]

Is there an antipattern name for "find a bug, write a test for it, fix the bug" as a crap form of testing?

There's nothing wrong with writing a test for a specific bug. In fact, you want to do that, so that after you've fixed the bug, your regression testing will catch it if you should accidentally revert the fix for the bug in a later revision. So that's not really an antipattern in itself.

The antipattern here is in not writing enough tests, at any level, to ensure that the product actually performs as specified. Their testing methodologies fail to come even close to providing a complete coverage of the code or of the product space, and that's the real problem here. Which is, of course, a failure of the product managers. Forrester, specifically, is obviously a moron. Which is strange, because when I knew him he seemed to be fairly reasonable. Was he hit by a bus sometime between 2006 and 2008?

[Vigilant]

Is there an antipattern name for "find a bug, write a test for it, fix the bug" as a crap form of testing?

There's nothing wrong with writing a test for a specific bug. In fact, you want to do that, so that after you've fixed the bug, your regression testing will catch it if you should accidentally revert the fix for the bug in a later revision. So that's not really an antipattern in itself.

The antipattern here is in not writing enough tests, at any level, to ensure that the product actually performs as specified. Their testing methodologies fail to come even close to providing a complete coverage of the code or of the product space, and that's the real problem here. Which is, of course, a failure of the product managers. Forrester, specifically, is obviously a moron. Which is strange, because when I knew him he seemed to be fairly reasonable. Was he hit by a bus sometime between 2006 and 2008?

They, like other engineering teams before them, are enamored of 'embracing new technologies'; doing the hard grunt work of engineering, not so much.

They're TechHipsterDouchebags, "This bugtracker is pretty new, you probably don't know how it works. :scoff: :throw head to mess up hair: :adjust stupid scarf: :sip Pabst blue ribbon: :adopt jaunty pose on back leg:"

Knowing how to test is hard and it's tedious, painstaking, mind breakingly specific work.
I've done FPGA and ASIC verification in verilog, system verilog and C/C++ on a very large and complicated design.
In silicon, you don't get 'do overs' and 'software patches', you get fucked in the ass for a new mask set. Thanks TSMC!

I have yet to see anything inside the WMF that makes me think they know the first thing about testing.

[Kelly Martin]

My fiancee was, years ago, on a team at a large consultancy that was charged with maintaining an internal product that was used for federated single signon throughout the organization. I remember the testing matrix they had developed for this product: there were thirteen rows and thirteen columns (a coincidence, as far as I could tell) and each combination had to be tested, exhaustively, for each new release. Some of the combinations had subcases, and a few were impossible, so it worked out that they had to test about 150 combinations. Signoff, saying that each and everyone one of these combinations had been tested, was required before a new release of the product could move from "testing" to "production".

If the WMF were doing it, they'd test row 1, column 1, and call it tested. If we're lucky, maybe they do row 2 column 2, too.

Testing is boring, tedious work, and while unit testing can be automated, most other forms of testing cannot be.

This is where the WMF is failing.

[Hex]

Is there an antipattern name for "find a bug, write a test for it, fix the bug" as a crap form of testing?

There's nothing wrong with writing a test for a specific bug. In fact, you want to do that, so that after you've fixed the bug, your regression testing will catch it if you should accidentally revert the fix for the bug in a later revision. So that's not really an antipattern in itself.

The antipattern here is in not writing enough tests, at any level, to ensure that the product actually performs as specified. Their testing methodologies fail to come even close to providing a complete coverage of the code or of the product space, and that's the real problem here.

Sorry, yes. I could have been more explicit. Find a bug, write a test for it, fix the bug and nothing else, not systematically testing features that you expect to work, let alone producing a coherent and easy-to-extend testing strategy for your whole codebase. And certainly not writing the tests first.

[Zoloft]

I have recently backed into being an application development lead.
First thing I did was redesign the process for gathering product owner requirements.
Second thing I did was introduce a strict testing regimen, based on a 'user case universe' where the application is tested extensively on all discovered user cases at each iteration. It's difficult. It's frustrating. In other words, it's real work.

[Malleus]

I have recently backed into being an application development lead.
First thing I did was redesign the process for gathering product owner requirements.
Second thing I did was introduce a strict testing regimen, based on a 'user case universe' where the application is tested extensively on all discovered user cases at each iteration. It's difficult. It's frustrating. In other words, it's real work.

When did Jacobsen's "use cases" transmogrify into "user cases"?

[tarantino]

It's fixed, we just know it!
It'll deploy on this Thursday's patch and the problem will go away forever...

Jdforrester-WMF added a subscriber: Jdforrester-WMF.Via Web · Wed, Jul 1, 7:37 PM
Confirmed fixed on http://en.wikipedia.beta.wmflabs.org/wi ... ction=edit

Huh, proxies and Tor aren't blocked from editing or creating accounts on en.wikipedia.beta.wmflabs.org. It doesn't really show up Google search results though, unless you search for site:wikipedia.beta.wmflabs.org.

[Zoloft]

I have recently backed into being an application development lead.
First thing I did was redesign the process for gathering product owner requirements.
Second thing I did was introduce a strict testing regimen, based on a 'user case universe' where the application is tested extensively on all discovered user cases at each iteration. It's difficult. It's frustrating. In other words, it's real work.

When did Jacobsen's "use cases" transmogrify into "user cases"?

It's a fine blend of 'user story' and 'use cases' - the development environment I inherited is more customer-requirement-oriented than a traditional 'waterfall' methodology. The established methodology was a heavily modified version of Scrum (software development) (T-H-L).

The product owner provides feedback at every stage of development.

User cases are based on what user requirements are developed, expressed as possible interactions by the user, based on the 'user story' which is another 'Scrum' ideology.

Yes, I know, clear as mud.

[The Joy]

http://www.qwantz.com/index.php?comic=1178

[Zoloft]

Here, let me throw this short article at Malleus to illustrate why I came up with 'User case' and User case Universe':

Use cases vs user stories in Agile development

It is my own method as a 'Scrum Master' to wrap a structure around 'user stories' so the team can quickly move to building code.

We had to develop a financial tracking application in six weeks with a team of five people. Done and on time. Minor bugs in beta, one refinement (interface tweak) before production.

It has performed well through four monthly cycles, and we'll see how it handles end-of-fiscal-year processing.

I am not a software engineer. But in my job, I do what needs to be done. We have six more projects coming up.

[Malleus]

Here, let me throw this short article at Malleus to illustrate why I came up with 'User case' and User case Universe':

Use cases vs user stories in Agile development

It is my own method as a 'Scrum Master' to wrap a structure around 'user stories' so the team can quickly move to building code.

We had to develop a financial tracking application in six weeks with a team of five people. Done and on time. Minor bugs in beta, one refinement (interface tweak) before production.

It has performed well through four monthly cycles, and we'll see how it handles end-of-fiscal-year processing.

I am not a software engineer. But in my job, I do what needs to be done. We have six more projects coming up.

I see.

[Notvelty]

Here, let me throw this short article at Malleus to illustrate why I came up with 'User case' and User case Universe':

Use cases vs user stories in Agile development

It is my own method as a 'Scrum Master' to wrap a structure around 'user stories' so the team can quickly move to building code.

We had to develop a financial tracking application in six weeks with a team of five people. Done and on time. Minor bugs in beta, one refinement (interface tweak) before production.

It has performed well through four monthly cycles, and we'll see how it handles end-of-fiscal-year processing.

I am not a software engineer. But in my job, I do what needs to be done. We have six more projects coming up.

There is nothing wrong or objectionable about your use of "user cases".

"Use case" is a made-up technical term that means "how people are going to use what we're creating". In the grand scheme of things, the term has been around for a very short people of time (barely decades rather than centuries). That some people use the term "user cases" is neither here nor there and only a pedantic twat whose only claim to fame is delivering competent work among deficient... oh.. right.. sry.. do carry on.

[Malleus]

Here, let me throw this short article at Malleus to illustrate why I came up with 'User case' and User case Universe':

Use cases vs user stories in Agile development

It is my own method as a 'Scrum Master' to wrap a structure around 'user stories' so the team can quickly move to building code.

We had to develop a financial tracking application in six weeks with a team of five people. Done and on time. Minor bugs in beta, one refinement (interface tweak) before production.

It has performed well through four monthly cycles, and we'll see how it handles end-of-fiscal-year processing.

I am not a software engineer. But in my job, I do what needs to be done. We have six more projects coming up.

There is nothing wrong or objectionable about your use of "user cases".

"Use case" is a made-up technical term that means "how people are going to use what we're creating". In the grand scheme of things, the term has been around for a very short people of time (barely decades rather than centuries). That some people use the term "user cases" is neither here nor there and only a pedantic twat whose only claim to fame is delivering competent work among deficient... oh.. right.. sry.. do carry on.

An interesting and very revealing response.

[Vigilant]

Thank god that Thursday patch fixed things... or did it?

Article not loading

Harold Innis won't finish VE-loading for me in Firefox 39.0 or Chrome 43.0.2357.81 Riggr Mortis (talk) 16:58, 6 July 2015 (UTC)

Whatthefuckisgoingon?! aka Sherry Snyder offers up the old Microsoft response to the problem!

Riggr Mortis, would you please try it again? It opens for me in Firefox, and Ops has been having some problems with the servers. Whatamidoing (WMF) (talk) 18:40, 6 July 2015 (UTC)

Nice work, turd blossom. I'm sure your job couldn't be replaced with a sound board.

Well, at least they finally delivered a solution to the problem. One quick little bug fix (bad title, no parameter validation) to what should have been an obvious test case candidate.
But wait...

can't save

Bug report VisualEditor
Description Won't save
Intention: editing Jimna Fire Tower, make a lot of changes, wanted to save
Steps to Reproduce: made a lot of changes, clicked save, typed in an edit summary and then Save
Results: Threw up a box with "parsoidserver-http: HTTP 500" and a dismiss button. Clicking the dismiss button returned me to the Edit Summary box. I tried to Save again, same error. I resumed editing, made a minor change, tried saving again. Same error.
Expectations: I expected it to Save
Page where the issue occurs
Web browser Chrome the latest
Operating system Vista Home Service Pack 2
Skin
Notes:
Workaround or suggested solution As I didn't want to lose a lot of changes, I tried switching to the source editor from within the VE. That worked and I was able to save my changes.
Kerry (talk) 21:42, 6 July 2015 (UTC)

Pretty much the solution to every VisualCowpat problem is to just not use it.

I do wonder how that regression suite is progressing...

Citoid disabled?

Tracked in Phabricator
Task T104944
Clicking on "Cite" in VE (the text, not the arrow) just opens the normal list of citations, not the interface to generate a citation via URL. I am almost 100% sure, I used that tool just 12 hours ago with no problems (Windows XP, vector, FF 39.0). Just in case: If Visual Editor is checking Citoid's availability and disables the function accordingly, it should display a short message like "Citoid unavailable. Please use manual citation selection." for clarity. GermanJoe (talk) 12:21, 7 July 2015 (UTC)

Hmmmm... Strange... No reasonable error codes, no external message, silent failure...seems about par for the course with WMF code...

And the dumbest remark of the year goes to ...

Thanks for the report. This was not deliberate, and we are investigating. ESanders (WMF) (talk) 14:12, 7 July 2015 (UTC)

[Hex]

Pretty much the solution to every VisualCowpat problem is to just not use it.

A strange game, the only way to win is not to play.

[Kelly Martin]

Well, then we really want them to activate the VisualCowpat for all users, especially anonymous ones. Hasten the day™, and all that.

[Vigilant]

Well, then we really want them to activate the VisualCowpat for all users, especially anonymous ones. Hasten the day™, and all that.

I'm so torn.

[Moral Hazard]

Well, then we really want them to activate the VisualCowpat for all users, especially anonymous ones. Hasten the day™, and all that.

I'm so torn.

The ethical code of the ACM does make it difficult to wish that the WMF force its users to use shitty software.

I wonder whether one could file a complaint with the ACM about the WMF?

Adherence of professionals to a code of ethics is largely a voluntary matter. However, if a member does not follow this code by engaging in gross misconduct, membership in ACM may be terminated.

C.f., the ACM--IEEE-CS software engineering code of ethics and professional practice.

[Peryglus]

Pretty much the solution to every VisualCowpat problem is to just not use it.

A strange game, the only way to win is not to play.

Ah, a bit like The Game (mind game) (T-H-L)

[mac]

Pretty much the solution to every VisualCowpat problem is to just not use it.

A strange game, the only way to win is not to play.

Ah, a bit like The Game (mind game) (T-H-L)

Sorta.

[Vigilant]

@ESanders (WMF): Looks like the purge re-activated the function, thanks (still a weird glitch). 2 minor points remain though:
The arrow next to "Cite" to directly select citations from a dropdown list has vanished (but maybe that was an intentional design decision?).
Closing the "Cite"-window stores the window setup with the last active sub-window, either "Automatic", "Manual" or "Re-use" (nice feature). However, closing "Cite"-window in "Manual" or "Re-use" mode and re-opening the "Cite"-window, the "Cancel" button is missing (it's only shown after closing and re-opening in "Automatic" mode). It should be visible opening the Cite-window in all 3 modes. GermanJoe (talk) 14:45, 7 July 2015 (UTC)

How many years has this been in development?
How many man years down the drain?
How much donor money wasted?

The real question, how long will problems like this continue to plague VisualEdsel?
The real answer, it's so poorly designed and built that it will always have these types of problems cropping up...forever.


The WMF software engineer full employment act.

[Notvelty]

Pretty much the solution to every VisualCowpat problem is to just not use it.

A strange game, the only way to win is not to play.

Ah, a bit like The Game (mind game) (T-H-L)

No. You can also win The Game by being older than 13.

[Vigilant]

And on we go...

Recent changes shows a use rate of the VisualEdsel to be around 26 out of every 5000 edits or around 0.5% of all edits made on en.wp.
Edits with VE tend to be much smaller than those made with the text editor, probably due to the many bugs and poor interface choices.

WP:VEF shows:
* bugs continue to be reintroduced due to lack of regression
* customer service arguing with editors
* tables are still broken
* templates still cause havoc
* feature usage is incredibly clunky, probably because the software folks don't actually edit

But don't worry en.wp victims, I'm sure the mandatory rollout of the latest iteration of choleric diarrhea masquerading as an editing tool will work just fine.

Anyone want to hazard a guess on the overall cost of VE?
We'll need to add in Parsoid, the fabulous test group, customer abuse support, management, overhead, brand destruction, loss of goodwill, etc

I'll say $40MUSD


P.S. Here's a page of nearly dead links for those trying to figure out VE
https://en.wikipedia.org/wiki/Wikipedia ... sualEditor
P.P.S The FAQ is particularly funny
https://en.wikipedia.org/wiki/Wikipedia ... Editor/FAQ

I've found an issue with VisualEditor or a feature that is missing. How can I tell you to fix it?
If you're willing and able, then please report the issue in bugzilla in the "VisualEditor" product.

Nice to see you guys keeping an eye on these pages. FAQ == Frequently Asked Questions, you morons.

[Vigilant]

This page is a nightmare of unreasonable expectations, contradiction, and sad.
https://www.mediawiki.org/wiki/VisualEditor/Roadmap

[Vigilant]

On testing...
https://www.mediawiki.org/wiki/Wikimedi ... ditor_Team
Improve stability with increased test coverage Code coverage reaches 80% overall, and over 90% in dm and over 70% in ce – end of quarter In progress In progress

They aren't writing tests as they code...
No comprehensive test strategy...

Bugs forever in this code.
WMF engineering full employement act.

This page
https://phabricator.wikimedia.org/project/view/483/
should tell you that VE is still no ready for primetime.

[Vigilant]

Ah, creeping deployment under the covers!!

VisualEditor is being gradually deployed to all new users, as a result of a community discussion last month. I believe that 25% of new accounts have access to it automatically right now. That will probably go up to 50%, and then 100%. (The system operates by fractions, so their plan is 1/20th of new accounts, 1/10th, 1/4, 1/2, and then 1/1.) Assuming that there are no further delays, then the deployment process will end around September 1.
Also, they're looking into moving the location of the prefs setting. VisualEditor is getting kind of big for Beta Features, so the Beta Features "opt-in" button may become a normal "opt-out" button under Special:Preferences#mw-prefsection-editing. They need to preserve the settings for active/experienced editors if they make that switch, so figuring out whether they can do that reliably is the first step there. Whatamidoing (WMF) (talk) 18:28, 17 August 2015 (UTC)

Take it all!

[Vigilant]

What counts for 'consensus' when WMF engineering wants to put out a shit product on the heads of newbie editors.
linkhttps://en.wikipedia.org/wiki/Wikipedia ... w_accounts[/link]

That doesn't look like any consensus I've ever seen.

[Vigilant]

This seems unnecessarily cruel.
https://en.wikipedia.org/w/index.php?ti ... tocomplete

Word completion or autocomplete

A really good and powerful feature which most text editors (like Vim, Kate...etc) have is word completion or autocomplete. The editor tries the guess the word being typed by offering the set of possible choices from words already mentioned in the file (in this case, the edited article). This would reduce typos by a great margin and even reduce editing time. Currently there exists no such feature in regular source editing via scripts or gadgets that I know of. -Ugog Nizdast (talk) 18:47, 27 August 2015 (UTC)

Why not just ask them to eat four raw pigs for breakfast or some other impossible task...

[MMAR]

Or, y'know, people could just rely on the fact all browsers have spell checker plug ins?

No need to reinvent the wheel just to get the clown car on the road.....

[Vigilant]

Or, y'know, people could just rely on the fact all browsers have spell checker plug ins?

No need to reinvent the wheel just to get the clown car on the road.....

Shhhhh.

You're going to hurt their feelings. Don't you know that they've worked HARD on this project? For four fucking years...

[auriental]

Or, y'know, people could just rely on the fact all browsers have spell checker plug ins?

No need to reinvent the wheel just to get the clown car on the road.....

Shhhhh.

You're going to hurt their feelings. Don't you know that they've worked HARD on this project? For four fucking years...

Just associating ideas:

[Peryglus]

I don't know whether this is the right thread for this, but it is kind of related to the WMF developers.

A fuss has been kicking off at Wikipedia:Village_pump_(technical)#Notifications_indicators_have_changed (T-H-L) because of some new icons that have appeared at the top of each WP page now, which splits notifications and messages. Is this a sign that the developers don't care what the community think or consult with them, as they're just playing around adding features that aren't going to be popular?

[Jim]

I don't know whether this is the right thread for this

It's probably the right thread.
Seriously, though, can you lend me one of your dozens of admin accounts? I have a really cunning plan.

[Hex]

A fuss has been kicking off at Wikipedia:Village_pump_(technical)#Notifications_indicators_have_changed (T-H-L) because of some new icons that have appeared at the top of each WP page now, which splits notifications and messages. Is this a sign that the developers don't care what the community think or consult with them, as they're just playing around adding features that aren't going to be popular?

Actually, they're just partially implementing one of my ideas. Again. My original proposal in October 2013 was more useful, and got a pretty good reception. Except from this guy:

This is probably a bad idea and not something we will want to implement. ... Recommending CLOSE: WONTFIX, as it defeats the function of the product.

Too bad, pal.

[Vigilant]

Add "useful error messages" to the list of things that VE can't do.
https://en.wikipedia.org/w/index.php?ti ... went_wrong

Something went wrong[edit]
User agent: Mozilla/5.0 (Windows NT 10.0; WOW64; Trident/7.0; Touch; .NET4.0C; .NET4.0E; .NET CLR 2.0.50727; .NET CLR 3.0.30729; .NET CLR 3.5.30729; Tablet PC 2.0; rv:11.0) like Gecko
URL: https://en.wikipedia.org/wiki/User:Rais ... ction=edit

After I edited, I could not save it: "Something went wrong. parsoidserver-http: HTTP 400."

Raiseyourvibration (talk) 16:32, 11 September 2015 (UTC)

[Hex]

I just tried it for the first time in maybe a year. It looks way better than it used to, I must admit, and is definitely faster. But my immediate experience is this: I tried the tool to add a link. It gave me a dialog box. I clicked "external link", pasted the URL, and pressed Insert. It gave me a numeric link, like so: "[1]". Erm... okay. Clicked it; right, there's a button that says "edit". Nope, can't type a label there... so that would be this button "Add label" then. Nope! That just displays the URL as the link. So how do you title a link? I found the answer in the user guide:

If you select (highlight) text or place your cursor within a word, and then press the "Link" button, a link will be inserted using that text or word. In most cases, when creating internal links (links to other pages in the same wiki) you do not select text. But when creating external links, in most cases you do select text before starting the linking process, and text is probably something you just typed.

So instead of the very basic functionality (present in every WYSIWYG editor since the dawn of time) of having a field to type a link title in, the new user is expected to independently discover an invisible pre-step to using the dialog. And if they can't, they have to read this abysmally poor documentation. What a crock.

[The Joy]

I tried editing Wikipedia recently and got the "edit source" and "edit this page" tabs mixed up. I kept hitting "edit source" and bringing VE up instead of directly editing the page code. I let out a Homer Simpson shriek each time VE came up.

[Vigilant]

Just look at what's on the current WP:VEF aka WP:WTF
https://en.wikipedia.org/w/index.php?ti ... =681215916

Thing won't load, won't save, nowiki tags all over the place...

They are working on a broken foundation.
This turd will never be stable.

[Moral Hazard]

Just look at what's on the current WP:VEF aka WP:WTF
https://en.wikipedia.org/w/index.php?ti ... =681215916

Thing won't load, won't save, nowiki tags all over the place...

They are working on a broken foundation.
This turd will never be stable.

You are judging the WMF like it was a cow, with a sequence of stomachs that allow it to digest cellulose.
The WMF is more like a rabbit, with one stomach, which uses coprophagy.

[Vigilant]

Just look at what's on the current WP:VEF aka WP:WTF
https://en.wikipedia.org/w/index.php?ti ... =681215916

Thing won't load, won't save, nowiki tags all over the place...

They are working on a broken foundation.
This turd will never be stable.

You are judging the WMF like it was a cow, with a sequence of stomachs that allow it to digest cellulose.
The WMF is more like a rabbit, with one stomach, which uses coprophagy.

THe problem with this analogy is that everyone else is forced to help with the coprophagy.
Sort of like a shit eating brood parasite.
http://www.fernbank.edu/Birding/parasitism.htm

[Black Kite]

Just look at what's on the current WP:VEF aka WP:WTF
https://en.wikipedia.org/w/index.php?ti ... =681215916

Thing won't load, won't save, nowiki tags all over the place...

They are working on a broken foundation.
This turd will never be stable.

The "nowiki tag" bug has been known about since ... ooh, at least April 2013.

linkhttps://en.wikipedia.org/wiki/Wikipedia ... ki_tags.3F[/link]

Edit Filter 550, which scans for it, has been active since then. Currently, it has 67,822 hits. Yesterday alone there were 117. And they still can't fix the bloody thing. Talk about "unfit for purpose".

[Vigilant]

Still screwing things up to this day.
https://en.wikipedia.org/w/index.php?ti ... =681989284

Title damaged

In this edit, VE damaged a title, making a correct title into a mess with displayed "==" and text merged in the same line as the title. --NicoV (Talk on frwiki) 14:02, 19 September 2015 (UTC)

Surely something you couldn't write a regression test for...

can't find media when trying to add media

I am using the visual editor to edit a wikipedia page. To insert a picture I uploaded in the wikimedia commons, I clicked insert, then media. I then typed into the search box many possible searches for my picture, including file name, description, and words in the file name and description. I can't find my picture, thus I can't upload it. But it definitely exists in the wikimedia commons. It would be a ridiculously simple upload with the old system, I just copy the URL into the wikipedia page. As far as I know, uploaded photos can only be added with the visual editor through the insert media thing I just tried. Good lord, this should be so easy, I hope I'm just ignorant and there is an easier way I haven't found with my stupid eyes. Please help.

Why would you want to upload a picture and then USE IT?!
Surely, there's no way to test this on god's green earth...

Still ISBN inside nowiki + other unwanted formatting

SSastry, User:cscott ISBN are still put between nowiki tags. When will this stop? An example with other damages done by VE:

Useless abbr tags without parameters
Useless time tags
Errorneous links like [[reliquaire]]<nowiki/>s
Incorrect referencing, maybe a user mistake: ¹
--NicoV (Talk on frwiki) 15:37, 20 September 2015 (UTC)

Nowiki: Ebola for an encyclopedia. No way to eradicate it.

[Hex]

A fuss has been kicking off at Wikipedia:Village_pump_(technical)#Notifications_indicators_have_changed (T-H-L) because of some new icons that have appeared at the top of each WP page now, which splits notifications and messages. Is this a sign that the developers don't care what the community think or consult with them, as they're just playing around adding features that aren't going to be popular?

The eagle-eyed among you will no doubt have noticed that this is no longer the case. They had to roll back the feature after only a couple of days because it turned out to be causing huge rendering issues on some browsers.

[Vigilant]

A fuss has been kicking off at Wikipedia:Village_pump_(technical)#Notifications_indicators_have_changed (T-H-L) because of some new icons that have appeared at the top of each WP page now, which splits notifications and messages. Is this a sign that the developers don't care what the community think or consult with them, as they're just playing around adding features that aren't going to be popular?

The eagle-eyed among you will no doubt have noticed that this is no longer the case. They had to roll back the feature after only a couple of days because it turned out to be causing huge rendering issues on some browsers.

LILA! HEY LILA! OVER HERE!!!

Hire a qualified verification person.
Let them tell you about the joys of competently designed automated regression testing.

[Ihatemyusername]

A fuss has been kicking off at Wikipedia:Village_pump_(technical)#Notifications_indicators_have_changed (T-H-L) because of some new icons that have appeared at the top of each WP page now, which splits notifications and messages. Is this a sign that the developers don't care what the community think or consult with them, as they're just playing around adding features that aren't going to be popular?

The eagle-eyed among you will no doubt have noticed that this is no longer the case. They had to roll back the feature after only a couple of days because it turned out to be causing huge rendering issues on some browsers.

As someone who doesn't know much about maintaining/designing websites: How is it possible that going from one grey box that turns red when something happens to two grey boxes that turn red when something happens, causes massive issues? Surely it wouldn't be that big a change?

[Vigilant]

A fuss has been kicking off at Wikipedia:Village_pump_(technical)#Notifications_indicators_have_changed (T-H-L) because of some new icons that have appeared at the top of each WP page now, which splits notifications and messages. Is this a sign that the developers don't care what the community think or consult with them, as they're just playing around adding features that aren't going to be popular?

The eagle-eyed among you will no doubt have noticed that this is no longer the case. They had to roll back the feature after only a couple of days because it turned out to be causing huge rendering issues on some browsers.

As someone who doesn't know much about maintaining/designing websites: How is it possible that going from one grey box that turns red when something happens to two grey boxes that turn red when something happens, causes massive issues? Surely it wouldn't be that big a change?

Given what I've seen over the past four years with the VE team, I'd guess that they have architected themselves into the corner and have no ability to make moderate changes without drastic breakage across the code base, including over at parsoid land.

Just look at how many times bug that were once fixed have been reintroduced into the current build and then released, untested, into the wild.

LILA!

Halt development.
Leave it where it sits.
Start over.
Hire a VP Engineering who knows something.
Let them build a team of professional software engineers, not refugees from en.wp that need a job and use IRC.
Insist on industry standard best practices form start to finish.
Monitor their progress religiously.
Close the feedback loop.

[GethN7]

Just dropping by to say I have this position on why VE sucks:

If a MediaWiki extension requires me to jump through a ton of hoops and have a lot of dependencies just so it will do something and it STILL has a ton of bugs once you finally get it working, it's still in alpha state at best.

Once it becomes much less bug riddled and the average MW user can deploy it themselves without too much work, then it's something you can call "usable".

[MMAR]

It's had bugs for so long, they are now features.

What is needed now is VF.

[auriental]

It's had bugs for so long, they are now features.

What is needed now is VF.

Is that either "Visual Flow" or perhaps the precursor to each of "VG", "VH"..."VZ", "WA" etc.?

[MMAR]

Visual Flow