Froze on Loading

[ghost]

Hey all,

some days ago i installed etherpad-lite. on installation all went fine but i think there is a problem with the database and i don´t know how to fix it.
I created an mysql user and a database and went on like the how to it say´s.

When i start etherpad-lite via script i can read this at log:

[32m[2012-04-10 13:56:43.875] [INFO] console - [39mYour Etherpad Lite git version is 309e3b0
[32m[2012-04-10 13:56:43.877] [INFO] console - [39mReport bugs at https://github.com/Pita/etherpad-lite/issues
[32m[2012-04-10 13:56:44.031] [INFO] console - [39mServer is listening at 88.198.116.77:9001
[32m[2012-04-10 13:56:44.032] [INFO] console - [39m �[36minfo -�[39m 'socket.io started'

and when i go on the site this apperas at log:

[32m[2012-04-10 13:57:06.482] [INFO] http -[39m200, GET /p/5CrUAwYnoW
[32m[2012-04-10 13:57:06.962] [INFO] http -[39m200, GET /minified/pad.js?callback=require.define
[32m[2012-04-10 13:57:07.012] [INFO] http -[39m200, GET /minified/pad/index.js?callback=require.define

on browser etherpad-lite freeze on "Loading..."
also the log stop until a new website request.

Anyone have an idee what i do wrong /where the problem is?

[jhollinger]

The develop branch is freezing quite a bit for me on "Loading..." as well, thought it looks like a javascript problem.

[ghost]

Okay but how can i fix it?

[fourplusone]

Which Browser/OS are you using

[ghost]

I use Win 7 Mozilla 11.0/ Chrome 18.0.xxx

Server is Ubutnu 10.04 LTS with Apache

[jhollinger]

Does your browser report any js errors?

[ghost]

Yep

[17:23:21.392] The module at "/pad" does not exist. @ http://88.198.116.77:9001/static/js/require-kernel.js:1

[edy]

same issue here with Firefox 11 (Chrome and Safari are OK)

$ git clone https://github.com/Pita/etherpad-lite.git && cd etherpad-lite
$ git checkout develop
$ ./bin/run.sh

Mac OS X 10.7.3
Firefox 11.0 (new installation, new profile but with Firebug installed)

[ghost]

not at all Chrome isn´t ok @ me

[fourplusone]

@edy-b can you confirm that 1260044 works and that c65216a does not?

[edy]

yes! 1260044 works and c65216a doesn't

[jhollinger]

Firebug hangs and gets no response from http://localhost:9001/javascripts/lib/ep_etherpad-lite/static/js/ace2_inner.js?callback=require.define. If I visit that url manually however, it servers just fine.

Chromium says Uncaught ReferenceError: translate is not defined.

[0ip]

Deleting "document.domain = document.domain;" in pad.html helps.
(develop branch, tested on Ubuntu Firefox 11)

[edy]

@0ip thank you!

[jhollinger]

@0ip it doesn't seem to help in my case. Same errors as before.

[jhollinger]

Chromium is dying at pad_userlist.js:713 translate("Enter your name"). translate is used a lot, but neither Chrome or I can find it defined it anywhere.

[jhollinger]

Yes, that's definitely the problem. I defined a rather dull translate function (which simply echos what you pass it). Now both Chrome and FF work. So...where is the real translate function?

[fourplusone]

@jhollinger can you specify the commit where translate is beeing used? cant find it on develop or master

[jhollinger]

I'm at the most recent checkout of develop. It's used about a dozen times.

[fourplusone]

Can't find one here: https://github.com/Pita/etherpad-lite/blob/develop/src/static/js/pad_userlist.js

[jhollinger]

What the crap, git? Maybe my branch is screwed up.

[jhollinger]

I like git but every now and then I have to delete everything and refork for no reason...grrr. That fixed translate in Chrome. But FF still hangs on ace2_inner.js.

[0ip]

I'm running 67d9f73.
Do you get that ace2_inner.js error here: http://pad.factor.cc:9002/p/tadzlvfjAx too?

[jhollinger]

@0ip same problem at your link. (FF isn't actually throwing an error. It just hangs on "Loading" and Firebug shows that the request for ace2_inner.js never completes.)

[jhollinger]

beta.etherpad.org doesn't work for me in FF either, looks like same issue.

[jhollinger]

I now believe this to be a Minify issue. I disabled minify and now it works. Investigating.

[jhollinger]

The problem seems to be triggered in pad.html. The fact that no errors are thrown makes it difficult to track down. But so far I've found two workarounds:

1. Remove ?callback=require.define from <script type="text/javascript" src="../javascripts/lib/ep_etherpad-lite/static/js/pad.js?callback=require.define"></script>

Or

2. Put pad = require('ep_etherpad-lite/static/js/pad').pad; inside of a 1 second setTimeout.

Does this make sense to anyone more familiar with kernel.require or yajsml? It only seems to be a problem in FF, so I guess the problem could be there, too.

[cweider]

@jhollinger, I’m looking into it. First thing I ought to do is adjust 22eaa6f, That conditional means that the packaging middleware is never exercised in development… so nobody will ever notice when integration with that breaks down.

[cweider]

I don’t have a full understanding on exactly what is happening, but the end effect was that the following lines:

--- a/src/templates/pad.html
+++ b/src/templates/pad.html
@@ -287,11 +287,14 @@

                   var plugins = require('ep_etherpad-lite/static/js/pluginfw/plugins');
                   plugins.update(function () {
+                    console.log('interrupted');
                     require('ep_etherpad-lite/static/js/pad').init();
                   });

                   /* TODO: These globals shouldn't exist. */
+                  console.log('- require');
                   pad = require('ep_etherpad-lite/static/js/pad').pad;
+                  console.log('+ require');
                   chat = require('ep_etherpad-lite/static/js/chat').chat;
                   padeditbar = require('ep_etherpad-lite/static/js/pad_editbar').padeditbar;
                   padimpexp = require('ep_etherpad-lite/static/js/pad_impexp').padimpexp;

Executed in this order:

- require
interrupted
+ require

I’m not a fan of what this appears to imply (that update’s XHR is preempting other JS code) and there are plenty of other things to cause this odd behavior (communication between our ~3 frames is more feasible), but anyways, scheduling the callback fixes the issue, which is good enough for me.