Last modified: 2014-11-05 23:47:19 UTC
In Firefox, I periodically get the JS error "Error: module not found: toast" when loading a page. It seems to be a race condition with the module loading, however, as it only happens about half the time. Also, can't reproduce with debug=true.
Prioritization and scheduling of this bug is tracked on Trello card https://trello.com/c/ziP2marX
I'm unable to replicate this locally or on beta labs. Can anyone else?
Seems to be thrown by mobile.watchstar.
I also get the error on Beta Labs.
I'm using Firefox 33, logged in, in stable. Can't reproduce on mediawiki.org, so must be a recent change.
Hmm, can't reproduce on betalabs, too :( (with FF 32 and 33, logged in and in stable (and alpha)).
Just got all the following errors locally with debug=true (logged in, in stable): Module not found: toast Module not found: modules/watchstar/Watchstar Module not found: modules/uploads/LeadPhotoUploaderButton Module not found: modules/uploads/PhotoUploaderButton Module not found: modules/PageList
All came from modules.js in Mantle.
I get mobile.watchstar and mobile.editor throwing this about 50% of the time locally.
The dependencies are likely to be out of whack. Check all those modules and that they are dependent on the appropriate things. Likely to be my fault hygiening up our messy Resources.php file. Wish I could replicate this locally. RL can be a weird thing.
(In reply to Jon from comment #10) > The dependencies are likely to be out of whack. Check all those modules and > that they are dependent on the appropriate things. Likely to be my fault > hygiening up our messy Resources.php file. > > Wish I could replicate this locally. RL can be a weird thing. So I /may/ have tracked down the error to 6e32b778501c0f9bf75e0321c107b2a0505ea271 with git bisect. I'm stressing may here because I haven't seen "Module not found: toast" since I started bisecting, I now see "Module not found: Overlay".
... and now I can't reproduce the error at HEAD.
Could be caching related. ./dev-scripts/cachedpage.sh 8327a9ceac372766955ac42042e89ba5bd8e3943 Headings If you can run the output on a localhost it should be possible to see if it's something weird in the HTML causing this.
Before Beta Labs stopped working, I was able to reproduce this there even on hard refreshes. I'm very worried about this getting to production as it causes random JS failure. What do people think about reverting Jon's patch until we figure out what's going on?
Completely wiping my browser cache seems to have fixed the problem, or at least I can't reproduce it anymore.
Hmm.
