Looks like the Great Firewall or something like it is preventing you from completely loading www.skritter.com because it is hosted on Google App Engine, which is periodically blocked. Try instead our mirror:

legacy.skritter.cn

This might also be caused by an internet filter, such as SafeEyes. If you have such a filter installed, try adding appspot.com to the list of allowed domains.

Glitch in Linux (Ubuntu) again

Schnabelhund   November 24th, 2012 9:23p.m.

Hi!
Remember this problem?

http://www.skritter.com/forum/topic?id=130925491

It’s happening again in Firefox, but Skritter works fine in Chrome.

Jose   November 25th, 2012 5:59a.m.

Same here (since two/three days ago). Redraw problems are even worse than the last time. Usually the left side of the kanji input pane does not redraw at all.

Tested with Ubuntu 12.04 (64 bits) and 12.10.

Chrome not working for me (kanji input works but the left pane does not show anything).
Chromium works (well, the cursor usually does not redraw properly but I can study here).

wispfrog   November 25th, 2012 10:16a.m.

Yes, seeing horrible redraw problems here too.

Also, it wants me to manually click to save reviews, and seems to heavily undercount the items due.

I think there was just a firefox update?

wispfrog   November 25th, 2012 10:25a.m.

Still doesn't work at all in chrome, just hangs on the 3 dots.

beta version works (or doesn't) just as badly.

Also, on the manual saving of reviews, when you click away it brings up a prompt for saving, and after that prompt the message changes to "saved" so one can't click it anyway, and next click away just leaves, but actually, it hasn't really saved the reviews at all, and same ones come up next time.

Please can we have the old version back?

wispfrog   November 25th, 2012 11:40a.m.

Hmm, when I run it from chrome as another user on the same box, then it doesn't hang at the dots, but its all the stuff /outside/ the flash that doesn't draw, rather than the flash applet itself. Weird.

On android Galaxy Note it still works somewhat, except the manual save review button appears on top of the applet, the warning pop up message is better, but it still doesn't actually save the reviews. Since one can't even manually save them, this is no good at all. Please reverse this change at least for android.

nick   November 25th, 2012 1:47p.m.

We didn't make any change there that would affect the redraw problem. Did you all get the same Firefox update?

Does it behave any differently on this URL? http://www.skritter.com/study/all?ten_point_one=1

wispfrog, can you check for me to see if there are any JavaScript or network errors when that manual saving of reviews part tries to fire? I can go through how to do this with Firebug if you'd like to help me debug this.

Schnabelhund   November 25th, 2012 2:00p.m.

The problem is still there. Can’t remember if I got a ff update, my version is 17.0.

nick   November 25th, 2012 2:24p.m.

Looking at the Firefox release history, it seems that 17.0 came out 5 days ago on November 20, so you would have gotten that update in the last few days. I think the interaction between this Firefox update and the Linux Flash player is what's causing the problem.

Could you try either going to an older or newer (beta) Firefox version to see if the problem still exists?

Schnabelhund   November 26th, 2012 1:25p.m.

I’ll try that as soon as I get a chance.

wispfrog   November 26th, 2012 3:29p.m.

So using the ?ten_point_one=1 option seems to make no difference, same sorts of redraw problems and saving issue. (There are also quite a few css warnings in the normal firefox console, btw)

I've just been getting into it with firebug, and seen on the net tab a progress save request going out when the don't change page dialog comes up and getting an ok 200 response back, and then on the next refresh getting the same character as was done a few ago (tun2) again immediately. I'm going to email you a screen shot with the post data and the next few requests in the hope that it might be helpful - am I missing some obvious way to get firebug to actually /export/ its trace info?

Edit: http://www.softwareishard.com/blog/netexport/

Also, switching to the response tab of each request in there always seems to be blank. But actual response headers are visible in the headers tab, that's where I saw 200 for the save.

nick   November 26th, 2012 5:14p.m.

Thanks for the great debugging help, wispfrog.

How long do you wait before refreshing the study page and getting the same words again? It does take several seconds now for the reviews to be saved after the request goes out, so if you refresh immediately, it's possible that you load the not-due words seconds before they become due again. Could that be what's going on? Looking at the log you sent, it was only about a five-second delay between response and page refresh, which might not be long enough.

Do you see the same problems if you give it 20 seconds?

The response tab is supposed to be blank here; the client doesn't need any information from the server on a saveReviews call other than the 200 status code.

wispfrog   November 27th, 2012 1:18p.m.

Yeah, so I was of course refreshing directly back into the applet when testing in order to provoke it, so that's very plausible. I've just been experimenting, and I think its a little worse than 20s needed sometimes, but its confusing as I think there's also an issue that the currently visible review doesn't get saved if one hasn't moved onto the next? But it certainly works if given reasonably long, so no worry there. I wonder if a save could be automatically triggered preemptively by the pointer leaving the window?

On the response tabs I was just wanting to see some data coming into the client and wasn't able to, and couldn't.

So the redraw front is really just going to be a matter of trying different flash versions then?

I've still got the even odder behaviour under chrome in two different ways.

nick   November 27th, 2012 4:12p.m.

I'll keep looking into delayed review save times. Yes, the word that you're currently on is not submitted for review when you leave the page, but at least that part is the same as the old system.

I'm not sure that the pointer leaving the window would be the best trigger for a save--many people would go to click on mnemonics and such, and the more often we save, the more expensive it is--but perhaps having it trigger whenever the mouse cursor goes above the prompt status bar (below the study | progress | etc pills), which might catch some of the common ways that people leave the page. Good thinking.

Not necessarily different Flash versions, although that might work, but different Firefox versions might be a better place to start...

Schnabelhund   December 9th, 2012 12:34a.m.

So it’s Firefox 17.0.1 now, but the problem is still there. Could you find a solution?

nick   December 9th, 2012 2:17p.m.

I probably can't solve this problem. I've never been able to do anything about buggy Flash hardware acceleration in Linux browsers except trying to turn it off (only helped slightly) and disabling all new Flash features from the last two years, and they're already off now.

This forum is now read only. Please go to Skritter Discourse Forum instead to start a new conversation!