search

zooniverse / planet-four

Identify and measure features on the surface of Mars
https://www.planetfour.org/
Apache License 2.0
2 stars 0 forks source link

Moving marks create artifacts in Firefox #88

Closed chrissnyder closed 5 years ago

chrissnyder commented 10 years ago

Can anyone off chance think of what might be causing the behavior shown below?

https://www.dropbox.com/s/dhsltumbs82rgab/Screenshot%202014-03-30%2017.30.31.png

https://www.dropbox.com/s/fu5t46mvokdpkzt/Screenshot%202014-03-27%2016.39.42.png

User on Win7/latest Firefox. Repeatable behavior. Could not duplicate on my Win7 machine.

michaelaye commented 10 years ago

I just duplicated that with Chrome/OSX(Mavericks) https://www.dropbox.com/s/yq6a8xcn5116evs/Screenshot%202014-05-15%2014.39.32.pdf

It's very solid repeatibility with that image http://planetfour.org/#/subjects/APF00002ec I closed the tab and launched it again with this direct URL (within our efforts to create gold standard data Meg made this list with direct access URLs) and the problem appears exactly at the same place again.

michaelaye commented 10 years ago

I cannot reproduce the issue with Chrome on OSX with above image anymore. I should mention that I updated to 10.9.3 this week, just in case that is important.

michaelaye commented 10 years ago

I have again this issue right now (14:37 PT May 27th) with this image id: http://planetfour.org/#/subjects/APF0000j3d and on Chrome with OSX. FF and Safari don't show the issue and a restart of Chrome did NOT help.

https://www.dropbox.com/s/q73cx20lvea53ak/Screenshot%202014-05-27%2014.34.09.pdf

chrissnyder commented 10 years ago

I've yet to be able to reproduce this under any circumstance. Something's causing it, but I can't pinpoint it. One of the mods also has this happen a lot.

If anyone can chime in with a repeatable set of circumstances it happens it, I'd be grateful.

michaelaye commented 10 years ago

Are you saying that with current Chrome on OSX (35.0.1916.114) and above image id you can not reproduce it? I apologize for not providing the Chrome version earlier.

chrissnyder commented 10 years ago

Any browser, any version. Doesn't matter, I haven't been able to recreate it.

michaelaye commented 10 years ago

Could it be related to a certain graphic card driver? I'm using a MacBookPro Retina from Mid 2012 with a NVIDIA GeForce GT 650M 1024 MB.

chrissnyder commented 10 years ago

Possibly, but admittedly I have no idea how to test that besides to try a bunch of different computers.

mschwamb commented 10 years ago

A volunteer is reporting the issue reappeared in this Talk thread. Could someone please take a look?

Thanks, ~Meg

mschwamb commented 10 years ago

A few more people are reporting ( this if it helps diagnose the bug

http://talk.planetfour.org/#/boards/BPF0000004/discussions/DPF0000h9d

mschwamb commented 10 years ago

A few more people are reporting ( this if it helps diagnose the bug)

http://talk.planetfour.org/#/boards/BPF0000004/discussions/DPF0000h9d?page=6&comment_id=53f1ce1f1e48620c5f000a3c

mschwamb commented 9 years ago

One of the moderators is still experiencing this interface weirdness. IE seems to be okay for him.

http://talk.planetfour.org/#/boards/BPF0000004/discussions/DPF0000h9d?page=6&comment_id=545a714c415ac16b000001d8

brian-c commented 9 years ago

Just added a hack that's fixed similar rendering issues in the past.

mschwamb commented 9 years ago

Thanks Brian. I'll let the moderator know, and I'll close the ticket if he sees improvement.

mschwamb commented 9 years ago

@brian-c the hack didn't appear to do anything according to our Talk moderator. He's still seeing the same thing. Does the moderator have to clear cookies or something for it to take effect?

brian-c commented 9 years ago

Nah, he should just have to reload the page. If it's not fixed, then the hack didn't work. If it happens across browsers, this is probably a hardware issue; I'm not sure what else we can do without being able to reproduce it.

michaelaye commented 9 years ago

Hm, could your hack possibly have improved the memory management for markings? I just went up to 136 blotches in Chrome without much noticeable slowdown. Usually, things start to crawl around 80-100.

brian-c commented 9 years ago

Probably has more to do with improving rendering performance by moving it to the GPU. I'm surprised it improved so much, since this implementation of the drawing tools is done in a canvas element and not the nested SVG structure we use now.

mschwamb commented 9 years ago

This may be helpful in tracking the bug down. Most detailed description yet from a volunteer:

Hi - same problem pops up every couple of days for me too - I've tried on Chrome, Firefox, IE and even Safari - sometimes it works fine of one, but not the others. Sometimes it has the same problem over all 4 browsers. Other days 3 out of four don't work - each time it's a different browser that is ok (yesterday IE was fine, day before it was Chrome). Very strange. Today, it is not working in Chrome, Firefox or IE - but it is fine in Safari! Always seems to be the same section of the images - right hand side and bottom - about a 2cm strip across the bottom and slightly thinner section up the right. Markers leave multiple trails and you can't adjust them if their expander points are inside the "problem area". I have still been placing the markers (adjusting their direction and shape first, then carefully dragging them into place - hoping that it is in the correct location!!)

mschwamb commented 9 years ago

Someone got what looks like a console grab during the time their browser was drawing the artifacts and there java errors listed but not sure if it's related. Looks like it's not, but I thought I'd let y'all decide that. You can see the screen grabs here on Talk.

chrissnyder commented 9 years ago

I think those are different errors. I cleaned up the language selection problems a bit ago. Shouldn't have any others problems on that front.

Still not sure what causes the weird artifacts while classifying though.