Topic on Extension talk:RevisionSlider

Problem

The revision pointers appear out of place (outside the revision bar area) when quickly clicking the left arrow (to load more revisions) after dropping the other pointer.

Steps to reproduce

1. Go to a page showing revisions (e.g. https://en.wikipedia.beta.wmflabs.org/w/index.php?diff=354925&oldid=217960&title=Main_Page&type=revision)
2. Click the icon to load the revision slider
3. Move the mouse over the yellow revision ball
4. Press the left mouse button
5. Move the mouse horizontally right until the yellow pointer is close to the blue one
6. Let go of the mouse button
7. Quickly, before the new changes load, click the "left arrow" to show older revisions

Expected

The click doesn't do anything until the changes are loaded.

Actual output

The revision bar area moves to show the new revisions, and this behaviour causes the yellow and blue pointer go "off screen" and appear outside to the left of "revision bar" area.

Note: Steps 6 and 7 need to be done very quickly to reproduce this behaviour.

Bless you. I saw this dozens of times, but never succeeded to reproduce so I could create a phab ticket.

Hey you two,

phew, that really is a tough one. I tried to follow the steps and be really really fast, but I failed to reproduce the error. :-/ Also I find it a bit hard to visualize what you mean with 'the yellow and blue pointer go "off screen"'.

It would be really awesome if you could send me a screenshot ( christoph.jauera@wikimedia.de ) or upload it somewhere and post a link here.

Thanks again for your very helpful feedback.

Best, Christoph

Next time I'll see this, I'll make a screenshot.

Indeed, a picture or two (animated) is worth a thousand words:

http://imgur.com/a/0kcg8

Maybe it happens more often when there are connectivity problems, or the bandwidth happens to be limited?

I guess limiting the bandwidth in developer tools would make it easier to replicate.

It's not the case I meant, but pretty close.

You mean not even close?

My case is due to conflicting conditions, it attempts to load the new revision and show older revisions. Since it can't do both simultaneously that happens.

Fixed the typo, thank you.

Hey,

thanks for the pictures! I think now I get what you mean.

I hope it is clear, that the pointers moving "of screen" is an intended feature to show, that the currently selected revisions are out of the scope of the current view ( in that case they are both somewhere "newer" then what you see ). So when you use the arrows on the left / on the right, and the slider loads a new screen with revisions he still gives you a hint where the currently selected revisions are situated.

But I assume what troubles you in that specific case is, that you can load a new pack of revisions into the slider view ( be clicking on the arrow ) even though he did not even finish to load the diff of the ( newly ) selected revisions.

Please correct me if I am wrong and totally misunderstood you there :-).

If you agree I would file a ticket for a feature request, that the arrows should block when the slider is still loading a diff.

Best,

Christoph

I hope it is clear, that the pointers moving "of screen" is an intended feature to show, that the currently selected revisions are out of the scope of the current view 

Well, it might be an intended feature but the visual cue doesn't highlight that fact. It is a bit like showing a train out of its railroad to indicate that the train is far away. One would expect some representation of a railroad in between the nearby railroad and the far away train. So maybe there could be a line underneath the blue and yellow pointer to show that is is expected. Consider this image (http://farm3.static.flickr.com/2773/4476401153_29fa621c5e.jpg) for example, the train is far away yet it is still clear that it is part of the track, instead of a train parked on a piece of land.

However, that is up to the designers.

If you agree I would file a ticket for a feature request, that the arrows should block when the slider is still loading a diff.

Agreed.

Hey,

I created two tickets on Phabricator, one for blocking the arrow when a diff is still loading:

https://phabricator.wikimedia.org/T163799

and one for a better "integration" of the pointers out of scope:

https://phabricator.wikimedia.org/T163800

So now PM and design can take over ;-)

In my case, I open the RS, and just after that see the problem, without arrows moving.

Indeed, it seems there are ways to reproduce stranger issues.

Here's a related issue in this revision:

https://en.wikipedia.beta.wmflabs.org/w/index.php?diff=2&oldid=2&title=Main_Page&type=revision

The only reproduction needed is to open that anonymously, and click the button to load the revision slider. Once it is done loading in a strange manner (it shows few bars and starts adding more slowly) you'll see something like this:

http://imgur.com/a/cDbyo

Reproduced in chrome and firefox.

Hey and thanks for this.

I am not sure if we want to address this issue because you try to get a diff for the same revision and that's usually nothing you would do as a normal user. Also you still can use the interface even if it might look a bit strange with the overlapping pointers.

Now I'm certain that there is a bug. From the history page, it is the first and second oldest revision of the main page, e.g. the diff of these:

• https://en.wikipedia.beta.wmflabs.org/w/index.php?title=Main_Page&oldid=54726
• https://en.wikipedia.beta.wmflabs.org/w/index.php?title=Main_Page&oldid=2

It shows up as these:

https://en.wikipedia.beta.wmflabs.org/w/index.php?title=Main_Page&type=revision&diff=2&oldid=54726

Once I click the revision slider button, it loads the revisions with the same off the screen problem for the yellow pointer. It also changes the ids to NAN in the URL for each of them:

https://en.wikipedia.beta.wmflabs.org/w/index.php?diff=NaN&oldid=NaN&title=Main_Page&type=revision

NAN (not a number probably) likely indicates some error in the javascript that loads these.

I know this NaN. Actually, because of me it was fixed once. The problem you show isn't RS's problem at all, but the Beta-cluster's one. Open the main page history. You'll see that the version 54726 is older that version 2. The RS should not support such a nonsense.

It is probably due to a creation or import issue. It doesn't need to support it, but it needs to fallback gracefully. It also clearly shows the drawback of having things off screen.

Unexpected behaviour is the reality of life, even in live wikis there are many weird conditions in which the slider might need to give up loading or fallback to some working state.

Hey hey!

Also locking in this thread again, it seems that this issue above is fixed with the RevisionSlider not assuming increasing revids. At least clicking on https://en.wikipedia.beta.wmflabs.org/w/index.php?title=Main_Page&type=revision&diff=2&oldid=54726 works for me now.

Best,

Christoph

Oddly enough they don't seem to be the same revision. I'm guessing that one was a hidden (oversighted) revision that regular users can't see. This is because the revision slider shows different text differences for each one of those. So it is either an issue on the backend, giving metadata about stuff that doesn't exist to users who shouldn't see them or the frontend for pulling ghost information from nowhere.

Maybe the issue is that they should be hidden from revision slider?

Try hovering at the second and third bar (revisions) on revision slider and you'll see what I mean. Anyway, it is a minor issue, and doesn't cause serious rendering issues, so it might be reasonable to just ignore it.

Mmmmmmmmmmmm. It's the same version.

UP the Reply@IKhitron Why not, if there's nobody there, who knows that anyone, could have logged in right-on, at this time.