Open Bug 1384748 Opened 4 years ago Updated 4 years ago
Development version of PDF
.js extension stopped loading
STR: 1. Open Nightly in non-E10s mode 2. Install "Development version" of PDF.js extension linked at https://github.com/mozilla/pdf.js/#firefox-and-seamonkey (and maybe restart Nightly?) 3. Open https://www.adobe.com/content/dam/Adobe/en/devnet/pdf/pdfs/PDF32000_2008.pdf Expected: document to be loaded Actual: viewer hangs (but can resume after the stop page loading button is clicked) It's a regression after bug 1383215: 14:46.64 INFO: Last good revision: 853cfa601a8f4d94f94af50505e2960624e1ca12 14:46.64 INFO: First bad revision: df302a197bca4990e47bce4b31dd00d051e7f079 14:46.64 INFO: Pushlog: https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=853cfa601a8f4d94f94af50505e2960624e1ca12&tochange=df302a197bca4990e47bce4b31dd00d051e7f079
I guess what's going on is that the extension is registering a new pdf.js resource package after the built-in one has already been registered, and the built-in module has already been loaded. But it doesn't unload the original module, and just expects future calls to resolve to the new URL. Which is a bug in the extension. If it wants to replace the built-in PDF.js resources, it needs to either register its versions early enough that the built-in ones don't load first, or it needs to properly unregister and destroy the built-in ones.
> Which is a bug in the extension. It worked before the change.
(In reply to Yury Delendik (:yury) from comment #3) > > Which is a bug in the extension. > > It worked before the change. It mostly worked, I'll grant you, aside from leaving an extra copy of the platform code loaded and active. But there were almost certainly other subtle bugs prior to this change. And even if it did work before, it worked by exploiting a platform bug. Which is still a bug, as far as I'm concerned.
You need to log in before you can comment on or make changes to this bug.